Training in progress, step 4500, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1673342072
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:212c1f4aa56505d9d886b2ce32af1f73897b72544c03971b84af1ea1e779ede1
|
| 3 |
size 1673342072
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 194745274
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f4ffa38fb5be6fffd0ce3ed64a7481e8be864f293de431d33c5627f72e76c98a
|
| 3 |
size 194745274
|
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:a95d4423cecde4a78762bf58efa44aece85b1128c704d4a826647715bc2c3a6c
|
| 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:a8670c5c048d65fccc45f56703c267d59b5ec466e8f5777cf6ff01b0a32855ba
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0
|
| 5 |
"eval_steps": 900,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -28039,6 +28039,3514 @@
|
|
| 28039 |
"learning_rate": 2.2271714922049e-05,
|
| 28040 |
"loss": 0.4815,
|
| 28041 |
"step": 4000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28042 |
}
|
| 28043 |
],
|
| 28044 |
"logging_steps": 1,
|
|
@@ -28053,12 +31561,12 @@
|
|
| 28053 |
"should_evaluate": false,
|
| 28054 |
"should_log": false,
|
| 28055 |
"should_save": true,
|
| 28056 |
-
"should_training_stop":
|
| 28057 |
},
|
| 28058 |
"attributes": {}
|
| 28059 |
}
|
| 28060 |
},
|
| 28061 |
-
"total_flos": 4.
|
| 28062 |
"train_batch_size": 1,
|
| 28063 |
"trial_name": null,
|
| 28064 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
"eval_steps": 900,
|
| 6 |
+
"global_step": 4500,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 28039 |
"learning_rate": 2.2271714922049e-05,
|
| 28040 |
"loss": 0.4815,
|
| 28041 |
"step": 4000
|
| 28042 |
+
},
|
| 28043 |
+
{
|
| 28044 |
+
"epoch": 0.8891111111111111,
|
| 28045 |
+
"grad_norm": 0.5801368355751038,
|
| 28046 |
+
"learning_rate": 2.22271714922049e-05,
|
| 28047 |
+
"loss": 0.8694,
|
| 28048 |
+
"step": 4001
|
| 28049 |
+
},
|
| 28050 |
+
{
|
| 28051 |
+
"epoch": 0.8893333333333333,
|
| 28052 |
+
"grad_norm": 0.6287752985954285,
|
| 28053 |
+
"learning_rate": 2.21826280623608e-05,
|
| 28054 |
+
"loss": 1.1768,
|
| 28055 |
+
"step": 4002
|
| 28056 |
+
},
|
| 28057 |
+
{
|
| 28058 |
+
"epoch": 0.8895555555555555,
|
| 28059 |
+
"grad_norm": 0.8464820981025696,
|
| 28060 |
+
"learning_rate": 2.2138084632516704e-05,
|
| 28061 |
+
"loss": 2.4031,
|
| 28062 |
+
"step": 4003
|
| 28063 |
+
},
|
| 28064 |
+
{
|
| 28065 |
+
"epoch": 0.8897777777777778,
|
| 28066 |
+
"grad_norm": 0.878257691860199,
|
| 28067 |
+
"learning_rate": 2.209354120267261e-05,
|
| 28068 |
+
"loss": 2.1387,
|
| 28069 |
+
"step": 4004
|
| 28070 |
+
},
|
| 28071 |
+
{
|
| 28072 |
+
"epoch": 0.89,
|
| 28073 |
+
"grad_norm": 0.6297408938407898,
|
| 28074 |
+
"learning_rate": 2.2048997772828508e-05,
|
| 28075 |
+
"loss": 1.0515,
|
| 28076 |
+
"step": 4005
|
| 28077 |
+
},
|
| 28078 |
+
{
|
| 28079 |
+
"epoch": 0.8902222222222222,
|
| 28080 |
+
"grad_norm": 0.8279980421066284,
|
| 28081 |
+
"learning_rate": 2.200445434298441e-05,
|
| 28082 |
+
"loss": 1.992,
|
| 28083 |
+
"step": 4006
|
| 28084 |
+
},
|
| 28085 |
+
{
|
| 28086 |
+
"epoch": 0.8904444444444445,
|
| 28087 |
+
"grad_norm": 0.07900725305080414,
|
| 28088 |
+
"learning_rate": 2.1959910913140315e-05,
|
| 28089 |
+
"loss": 0.0116,
|
| 28090 |
+
"step": 4007
|
| 28091 |
+
},
|
| 28092 |
+
{
|
| 28093 |
+
"epoch": 0.8906666666666667,
|
| 28094 |
+
"grad_norm": 0.07973389327526093,
|
| 28095 |
+
"learning_rate": 2.1915367483296214e-05,
|
| 28096 |
+
"loss": 0.0113,
|
| 28097 |
+
"step": 4008
|
| 28098 |
+
},
|
| 28099 |
+
{
|
| 28100 |
+
"epoch": 0.8908888888888888,
|
| 28101 |
+
"grad_norm": 0.5878556370735168,
|
| 28102 |
+
"learning_rate": 2.1870824053452115e-05,
|
| 28103 |
+
"loss": 1.1704,
|
| 28104 |
+
"step": 4009
|
| 28105 |
+
},
|
| 28106 |
+
{
|
| 28107 |
+
"epoch": 0.8911111111111111,
|
| 28108 |
+
"grad_norm": 0.9955252408981323,
|
| 28109 |
+
"learning_rate": 2.182628062360802e-05,
|
| 28110 |
+
"loss": 1.8866,
|
| 28111 |
+
"step": 4010
|
| 28112 |
+
},
|
| 28113 |
+
{
|
| 28114 |
+
"epoch": 0.8913333333333333,
|
| 28115 |
+
"grad_norm": 0.876213788986206,
|
| 28116 |
+
"learning_rate": 2.178173719376392e-05,
|
| 28117 |
+
"loss": 1.9363,
|
| 28118 |
+
"step": 4011
|
| 28119 |
+
},
|
| 28120 |
+
{
|
| 28121 |
+
"epoch": 0.8915555555555555,
|
| 28122 |
+
"grad_norm": 0.8237855434417725,
|
| 28123 |
+
"learning_rate": 2.173719376391982e-05,
|
| 28124 |
+
"loss": 2.1091,
|
| 28125 |
+
"step": 4012
|
| 28126 |
+
},
|
| 28127 |
+
{
|
| 28128 |
+
"epoch": 0.8917777777777778,
|
| 28129 |
+
"grad_norm": 0.6529291868209839,
|
| 28130 |
+
"learning_rate": 2.1692650334075727e-05,
|
| 28131 |
+
"loss": 0.9921,
|
| 28132 |
+
"step": 4013
|
| 28133 |
+
},
|
| 28134 |
+
{
|
| 28135 |
+
"epoch": 0.892,
|
| 28136 |
+
"grad_norm": 0.9489926099777222,
|
| 28137 |
+
"learning_rate": 2.1648106904231625e-05,
|
| 28138 |
+
"loss": 1.8424,
|
| 28139 |
+
"step": 4014
|
| 28140 |
+
},
|
| 28141 |
+
{
|
| 28142 |
+
"epoch": 0.8922222222222222,
|
| 28143 |
+
"grad_norm": 0.973099946975708,
|
| 28144 |
+
"learning_rate": 2.1603563474387527e-05,
|
| 28145 |
+
"loss": 1.9839,
|
| 28146 |
+
"step": 4015
|
| 28147 |
+
},
|
| 28148 |
+
{
|
| 28149 |
+
"epoch": 0.8924444444444445,
|
| 28150 |
+
"grad_norm": 0.8978729248046875,
|
| 28151 |
+
"learning_rate": 2.1559020044543433e-05,
|
| 28152 |
+
"loss": 1.7646,
|
| 28153 |
+
"step": 4016
|
| 28154 |
+
},
|
| 28155 |
+
{
|
| 28156 |
+
"epoch": 0.8926666666666667,
|
| 28157 |
+
"grad_norm": 0.9603530764579773,
|
| 28158 |
+
"learning_rate": 2.1514476614699335e-05,
|
| 28159 |
+
"loss": 1.8374,
|
| 28160 |
+
"step": 4017
|
| 28161 |
+
},
|
| 28162 |
+
{
|
| 28163 |
+
"epoch": 0.8928888888888888,
|
| 28164 |
+
"grad_norm": 0.07031574100255966,
|
| 28165 |
+
"learning_rate": 2.1469933184855233e-05,
|
| 28166 |
+
"loss": 0.0148,
|
| 28167 |
+
"step": 4018
|
| 28168 |
+
},
|
| 28169 |
+
{
|
| 28170 |
+
"epoch": 0.8931111111111111,
|
| 28171 |
+
"grad_norm": 0.06550273299217224,
|
| 28172 |
+
"learning_rate": 2.142538975501114e-05,
|
| 28173 |
+
"loss": 0.0147,
|
| 28174 |
+
"step": 4019
|
| 28175 |
+
},
|
| 28176 |
+
{
|
| 28177 |
+
"epoch": 0.8933333333333333,
|
| 28178 |
+
"grad_norm": 0.06782650202512741,
|
| 28179 |
+
"learning_rate": 2.138084632516704e-05,
|
| 28180 |
+
"loss": 0.0149,
|
| 28181 |
+
"step": 4020
|
| 28182 |
+
},
|
| 28183 |
+
{
|
| 28184 |
+
"epoch": 0.8935555555555555,
|
| 28185 |
+
"grad_norm": 0.7483673095703125,
|
| 28186 |
+
"learning_rate": 2.133630289532294e-05,
|
| 28187 |
+
"loss": 0.9084,
|
| 28188 |
+
"step": 4021
|
| 28189 |
+
},
|
| 28190 |
+
{
|
| 28191 |
+
"epoch": 0.8937777777777778,
|
| 28192 |
+
"grad_norm": 0.9090237617492676,
|
| 28193 |
+
"learning_rate": 2.129175946547884e-05,
|
| 28194 |
+
"loss": 1.8329,
|
| 28195 |
+
"step": 4022
|
| 28196 |
+
},
|
| 28197 |
+
{
|
| 28198 |
+
"epoch": 0.894,
|
| 28199 |
+
"grad_norm": 0.7626523971557617,
|
| 28200 |
+
"learning_rate": 2.1247216035634746e-05,
|
| 28201 |
+
"loss": 0.9174,
|
| 28202 |
+
"step": 4023
|
| 28203 |
+
},
|
| 28204 |
+
{
|
| 28205 |
+
"epoch": 0.8942222222222223,
|
| 28206 |
+
"grad_norm": 0.6706441640853882,
|
| 28207 |
+
"learning_rate": 2.1202672605790645e-05,
|
| 28208 |
+
"loss": 0.8133,
|
| 28209 |
+
"step": 4024
|
| 28210 |
+
},
|
| 28211 |
+
{
|
| 28212 |
+
"epoch": 0.8944444444444445,
|
| 28213 |
+
"grad_norm": 0.9489988684654236,
|
| 28214 |
+
"learning_rate": 2.1158129175946547e-05,
|
| 28215 |
+
"loss": 2.003,
|
| 28216 |
+
"step": 4025
|
| 28217 |
+
},
|
| 28218 |
+
{
|
| 28219 |
+
"epoch": 0.8946666666666667,
|
| 28220 |
+
"grad_norm": 0.9574695825576782,
|
| 28221 |
+
"learning_rate": 2.1113585746102452e-05,
|
| 28222 |
+
"loss": 1.5621,
|
| 28223 |
+
"step": 4026
|
| 28224 |
+
},
|
| 28225 |
+
{
|
| 28226 |
+
"epoch": 0.8948888888888888,
|
| 28227 |
+
"grad_norm": 1.1038743257522583,
|
| 28228 |
+
"learning_rate": 2.1069042316258354e-05,
|
| 28229 |
+
"loss": 1.8653,
|
| 28230 |
+
"step": 4027
|
| 28231 |
+
},
|
| 28232 |
+
{
|
| 28233 |
+
"epoch": 0.8951111111111111,
|
| 28234 |
+
"grad_norm": 1.0262362957000732,
|
| 28235 |
+
"learning_rate": 2.1024498886414253e-05,
|
| 28236 |
+
"loss": 1.6652,
|
| 28237 |
+
"step": 4028
|
| 28238 |
+
},
|
| 28239 |
+
{
|
| 28240 |
+
"epoch": 0.8953333333333333,
|
| 28241 |
+
"grad_norm": 0.8741075396537781,
|
| 28242 |
+
"learning_rate": 2.0979955456570158e-05,
|
| 28243 |
+
"loss": 1.6058,
|
| 28244 |
+
"step": 4029
|
| 28245 |
+
},
|
| 28246 |
+
{
|
| 28247 |
+
"epoch": 0.8955555555555555,
|
| 28248 |
+
"grad_norm": 0.7687373161315918,
|
| 28249 |
+
"learning_rate": 2.093541202672606e-05,
|
| 28250 |
+
"loss": 0.8818,
|
| 28251 |
+
"step": 4030
|
| 28252 |
+
},
|
| 28253 |
+
{
|
| 28254 |
+
"epoch": 0.8957777777777778,
|
| 28255 |
+
"grad_norm": 0.06525861471891403,
|
| 28256 |
+
"learning_rate": 2.089086859688196e-05,
|
| 28257 |
+
"loss": 0.0177,
|
| 28258 |
+
"step": 4031
|
| 28259 |
+
},
|
| 28260 |
+
{
|
| 28261 |
+
"epoch": 0.896,
|
| 28262 |
+
"grad_norm": 0.6307370066642761,
|
| 28263 |
+
"learning_rate": 2.0846325167037864e-05,
|
| 28264 |
+
"loss": 0.7724,
|
| 28265 |
+
"step": 4032
|
| 28266 |
+
},
|
| 28267 |
+
{
|
| 28268 |
+
"epoch": 0.8962222222222223,
|
| 28269 |
+
"grad_norm": 1.1199438571929932,
|
| 28270 |
+
"learning_rate": 2.0801781737193766e-05,
|
| 28271 |
+
"loss": 1.7534,
|
| 28272 |
+
"step": 4033
|
| 28273 |
+
},
|
| 28274 |
+
{
|
| 28275 |
+
"epoch": 0.8964444444444445,
|
| 28276 |
+
"grad_norm": 0.9748408794403076,
|
| 28277 |
+
"learning_rate": 2.0757238307349665e-05,
|
| 28278 |
+
"loss": 1.6166,
|
| 28279 |
+
"step": 4034
|
| 28280 |
+
},
|
| 28281 |
+
{
|
| 28282 |
+
"epoch": 0.8966666666666666,
|
| 28283 |
+
"grad_norm": 0.0824805200099945,
|
| 28284 |
+
"learning_rate": 2.071269487750557e-05,
|
| 28285 |
+
"loss": 0.0188,
|
| 28286 |
+
"step": 4035
|
| 28287 |
+
},
|
| 28288 |
+
{
|
| 28289 |
+
"epoch": 0.8968888888888888,
|
| 28290 |
+
"grad_norm": 0.09000510722398758,
|
| 28291 |
+
"learning_rate": 2.0668151447661472e-05,
|
| 28292 |
+
"loss": 0.0188,
|
| 28293 |
+
"step": 4036
|
| 28294 |
+
},
|
| 28295 |
+
{
|
| 28296 |
+
"epoch": 0.8971111111111111,
|
| 28297 |
+
"grad_norm": 0.08561154454946518,
|
| 28298 |
+
"learning_rate": 2.0623608017817374e-05,
|
| 28299 |
+
"loss": 0.0185,
|
| 28300 |
+
"step": 4037
|
| 28301 |
+
},
|
| 28302 |
+
{
|
| 28303 |
+
"epoch": 0.8973333333333333,
|
| 28304 |
+
"grad_norm": 0.7661683559417725,
|
| 28305 |
+
"learning_rate": 2.0579064587973276e-05,
|
| 28306 |
+
"loss": 0.907,
|
| 28307 |
+
"step": 4038
|
| 28308 |
+
},
|
| 28309 |
+
{
|
| 28310 |
+
"epoch": 0.8975555555555556,
|
| 28311 |
+
"grad_norm": 0.9890311360359192,
|
| 28312 |
+
"learning_rate": 2.0534521158129178e-05,
|
| 28313 |
+
"loss": 1.3849,
|
| 28314 |
+
"step": 4039
|
| 28315 |
+
},
|
| 28316 |
+
{
|
| 28317 |
+
"epoch": 0.8977777777777778,
|
| 28318 |
+
"grad_norm": 0.7973209619522095,
|
| 28319 |
+
"learning_rate": 2.048997772828508e-05,
|
| 28320 |
+
"loss": 0.749,
|
| 28321 |
+
"step": 4040
|
| 28322 |
+
},
|
| 28323 |
+
{
|
| 28324 |
+
"epoch": 0.898,
|
| 28325 |
+
"grad_norm": 1.1026244163513184,
|
| 28326 |
+
"learning_rate": 2.044543429844098e-05,
|
| 28327 |
+
"loss": 1.7591,
|
| 28328 |
+
"step": 4041
|
| 28329 |
+
},
|
| 28330 |
+
{
|
| 28331 |
+
"epoch": 0.8982222222222223,
|
| 28332 |
+
"grad_norm": 1.1480908393859863,
|
| 28333 |
+
"learning_rate": 2.0400890868596884e-05,
|
| 28334 |
+
"loss": 1.9024,
|
| 28335 |
+
"step": 4042
|
| 28336 |
+
},
|
| 28337 |
+
{
|
| 28338 |
+
"epoch": 0.8984444444444445,
|
| 28339 |
+
"grad_norm": 1.0242488384246826,
|
| 28340 |
+
"learning_rate": 2.0356347438752786e-05,
|
| 28341 |
+
"loss": 1.4006,
|
| 28342 |
+
"step": 4043
|
| 28343 |
+
},
|
| 28344 |
+
{
|
| 28345 |
+
"epoch": 0.8986666666666666,
|
| 28346 |
+
"grad_norm": 1.117613434791565,
|
| 28347 |
+
"learning_rate": 2.0311804008908684e-05,
|
| 28348 |
+
"loss": 1.3118,
|
| 28349 |
+
"step": 4044
|
| 28350 |
+
},
|
| 28351 |
+
{
|
| 28352 |
+
"epoch": 0.8988888888888888,
|
| 28353 |
+
"grad_norm": 1.1382890939712524,
|
| 28354 |
+
"learning_rate": 2.026726057906459e-05,
|
| 28355 |
+
"loss": 1.3193,
|
| 28356 |
+
"step": 4045
|
| 28357 |
+
},
|
| 28358 |
+
{
|
| 28359 |
+
"epoch": 0.8991111111111111,
|
| 28360 |
+
"grad_norm": 1.2156895399093628,
|
| 28361 |
+
"learning_rate": 2.0222717149220492e-05,
|
| 28362 |
+
"loss": 1.1794,
|
| 28363 |
+
"step": 4046
|
| 28364 |
+
},
|
| 28365 |
+
{
|
| 28366 |
+
"epoch": 0.8993333333333333,
|
| 28367 |
+
"grad_norm": 0.6855819225311279,
|
| 28368 |
+
"learning_rate": 2.017817371937639e-05,
|
| 28369 |
+
"loss": 0.4808,
|
| 28370 |
+
"step": 4047
|
| 28371 |
+
},
|
| 28372 |
+
{
|
| 28373 |
+
"epoch": 0.8995555555555556,
|
| 28374 |
+
"grad_norm": 0.1426740288734436,
|
| 28375 |
+
"learning_rate": 2.0133630289532296e-05,
|
| 28376 |
+
"loss": 0.033,
|
| 28377 |
+
"step": 4048
|
| 28378 |
+
},
|
| 28379 |
+
{
|
| 28380 |
+
"epoch": 0.8997777777777778,
|
| 28381 |
+
"grad_norm": 1.0770491361618042,
|
| 28382 |
+
"learning_rate": 2.0089086859688198e-05,
|
| 28383 |
+
"loss": 0.8844,
|
| 28384 |
+
"step": 4049
|
| 28385 |
+
},
|
| 28386 |
+
{
|
| 28387 |
+
"epoch": 0.9,
|
| 28388 |
+
"grad_norm": 0.9471620917320251,
|
| 28389 |
+
"learning_rate": 2.00445434298441e-05,
|
| 28390 |
+
"loss": 0.9019,
|
| 28391 |
+
"step": 4050
|
| 28392 |
+
},
|
| 28393 |
+
{
|
| 28394 |
+
"epoch": 0.9002222222222223,
|
| 28395 |
+
"grad_norm": 0.8870931267738342,
|
| 28396 |
+
"learning_rate": 2e-05,
|
| 28397 |
+
"loss": 1.9518,
|
| 28398 |
+
"step": 4051
|
| 28399 |
+
},
|
| 28400 |
+
{
|
| 28401 |
+
"epoch": 0.9004444444444445,
|
| 28402 |
+
"grad_norm": 0.681877851486206,
|
| 28403 |
+
"learning_rate": 1.9955456570155904e-05,
|
| 28404 |
+
"loss": 1.0902,
|
| 28405 |
+
"step": 4052
|
| 28406 |
+
},
|
| 28407 |
+
{
|
| 28408 |
+
"epoch": 0.9006666666666666,
|
| 28409 |
+
"grad_norm": 0.8550397157669067,
|
| 28410 |
+
"learning_rate": 1.9910913140311806e-05,
|
| 28411 |
+
"loss": 2.1819,
|
| 28412 |
+
"step": 4053
|
| 28413 |
+
},
|
| 28414 |
+
{
|
| 28415 |
+
"epoch": 0.9008888888888889,
|
| 28416 |
+
"grad_norm": 0.5659412741661072,
|
| 28417 |
+
"learning_rate": 1.9866369710467708e-05,
|
| 28418 |
+
"loss": 1.309,
|
| 28419 |
+
"step": 4054
|
| 28420 |
+
},
|
| 28421 |
+
{
|
| 28422 |
+
"epoch": 0.9011111111111111,
|
| 28423 |
+
"grad_norm": 0.618087887763977,
|
| 28424 |
+
"learning_rate": 1.982182628062361e-05,
|
| 28425 |
+
"loss": 1.0288,
|
| 28426 |
+
"step": 4055
|
| 28427 |
+
},
|
| 28428 |
+
{
|
| 28429 |
+
"epoch": 0.9013333333333333,
|
| 28430 |
+
"grad_norm": 0.5965234637260437,
|
| 28431 |
+
"learning_rate": 1.977728285077951e-05,
|
| 28432 |
+
"loss": 1.0864,
|
| 28433 |
+
"step": 4056
|
| 28434 |
+
},
|
| 28435 |
+
{
|
| 28436 |
+
"epoch": 0.9015555555555556,
|
| 28437 |
+
"grad_norm": 0.8751803636550903,
|
| 28438 |
+
"learning_rate": 1.973273942093541e-05,
|
| 28439 |
+
"loss": 2.2079,
|
| 28440 |
+
"step": 4057
|
| 28441 |
+
},
|
| 28442 |
+
{
|
| 28443 |
+
"epoch": 0.9017777777777778,
|
| 28444 |
+
"grad_norm": 0.907996416091919,
|
| 28445 |
+
"learning_rate": 1.9688195991091315e-05,
|
| 28446 |
+
"loss": 2.0763,
|
| 28447 |
+
"step": 4058
|
| 28448 |
+
},
|
| 28449 |
+
{
|
| 28450 |
+
"epoch": 0.902,
|
| 28451 |
+
"grad_norm": 0.08640366792678833,
|
| 28452 |
+
"learning_rate": 1.9643652561247217e-05,
|
| 28453 |
+
"loss": 0.0116,
|
| 28454 |
+
"step": 4059
|
| 28455 |
+
},
|
| 28456 |
+
{
|
| 28457 |
+
"epoch": 0.9022222222222223,
|
| 28458 |
+
"grad_norm": 0.08815193176269531,
|
| 28459 |
+
"learning_rate": 1.959910913140312e-05,
|
| 28460 |
+
"loss": 0.0114,
|
| 28461 |
+
"step": 4060
|
| 28462 |
+
},
|
| 28463 |
+
{
|
| 28464 |
+
"epoch": 0.9024444444444445,
|
| 28465 |
+
"grad_norm": 0.08544806391000748,
|
| 28466 |
+
"learning_rate": 1.955456570155902e-05,
|
| 28467 |
+
"loss": 0.0115,
|
| 28468 |
+
"step": 4061
|
| 28469 |
+
},
|
| 28470 |
+
{
|
| 28471 |
+
"epoch": 0.9026666666666666,
|
| 28472 |
+
"grad_norm": 0.6461583375930786,
|
| 28473 |
+
"learning_rate": 1.9510022271714923e-05,
|
| 28474 |
+
"loss": 1.1449,
|
| 28475 |
+
"step": 4062
|
| 28476 |
+
},
|
| 28477 |
+
{
|
| 28478 |
+
"epoch": 0.9028888888888889,
|
| 28479 |
+
"grad_norm": 0.8666505813598633,
|
| 28480 |
+
"learning_rate": 1.9465478841870825e-05,
|
| 28481 |
+
"loss": 2.0281,
|
| 28482 |
+
"step": 4063
|
| 28483 |
+
},
|
| 28484 |
+
{
|
| 28485 |
+
"epoch": 0.9031111111111111,
|
| 28486 |
+
"grad_norm": 1.0003634691238403,
|
| 28487 |
+
"learning_rate": 1.9420935412026727e-05,
|
| 28488 |
+
"loss": 1.929,
|
| 28489 |
+
"step": 4064
|
| 28490 |
+
},
|
| 28491 |
+
{
|
| 28492 |
+
"epoch": 0.9033333333333333,
|
| 28493 |
+
"grad_norm": 1.1285505294799805,
|
| 28494 |
+
"learning_rate": 1.937639198218263e-05,
|
| 28495 |
+
"loss": 2.3214,
|
| 28496 |
+
"step": 4065
|
| 28497 |
+
},
|
| 28498 |
+
{
|
| 28499 |
+
"epoch": 0.9035555555555556,
|
| 28500 |
+
"grad_norm": 0.6442127823829651,
|
| 28501 |
+
"learning_rate": 1.933184855233853e-05,
|
| 28502 |
+
"loss": 0.79,
|
| 28503 |
+
"step": 4066
|
| 28504 |
+
},
|
| 28505 |
+
{
|
| 28506 |
+
"epoch": 0.9037777777777778,
|
| 28507 |
+
"grad_norm": 0.6966649889945984,
|
| 28508 |
+
"learning_rate": 1.9287305122494433e-05,
|
| 28509 |
+
"loss": 0.9614,
|
| 28510 |
+
"step": 4067
|
| 28511 |
+
},
|
| 28512 |
+
{
|
| 28513 |
+
"epoch": 0.904,
|
| 28514 |
+
"grad_norm": 1.0646114349365234,
|
| 28515 |
+
"learning_rate": 1.9242761692650335e-05,
|
| 28516 |
+
"loss": 2.0066,
|
| 28517 |
+
"step": 4068
|
| 28518 |
+
},
|
| 28519 |
+
{
|
| 28520 |
+
"epoch": 0.9042222222222223,
|
| 28521 |
+
"grad_norm": 1.0722988843917847,
|
| 28522 |
+
"learning_rate": 1.9198218262806237e-05,
|
| 28523 |
+
"loss": 1.8977,
|
| 28524 |
+
"step": 4069
|
| 28525 |
+
},
|
| 28526 |
+
{
|
| 28527 |
+
"epoch": 0.9044444444444445,
|
| 28528 |
+
"grad_norm": 0.9870444536209106,
|
| 28529 |
+
"learning_rate": 1.915367483296214e-05,
|
| 28530 |
+
"loss": 1.6182,
|
| 28531 |
+
"step": 4070
|
| 28532 |
+
},
|
| 28533 |
+
{
|
| 28534 |
+
"epoch": 0.9046666666666666,
|
| 28535 |
+
"grad_norm": 0.6224427819252014,
|
| 28536 |
+
"learning_rate": 1.910913140311804e-05,
|
| 28537 |
+
"loss": 0.7754,
|
| 28538 |
+
"step": 4071
|
| 28539 |
+
},
|
| 28540 |
+
{
|
| 28541 |
+
"epoch": 0.9048888888888889,
|
| 28542 |
+
"grad_norm": 0.06635406613349915,
|
| 28543 |
+
"learning_rate": 1.9064587973273943e-05,
|
| 28544 |
+
"loss": 0.0147,
|
| 28545 |
+
"step": 4072
|
| 28546 |
+
},
|
| 28547 |
+
{
|
| 28548 |
+
"epoch": 0.9051111111111111,
|
| 28549 |
+
"grad_norm": 0.07058946043252945,
|
| 28550 |
+
"learning_rate": 1.9020044543429845e-05,
|
| 28551 |
+
"loss": 0.0173,
|
| 28552 |
+
"step": 4073
|
| 28553 |
+
},
|
| 28554 |
+
{
|
| 28555 |
+
"epoch": 0.9053333333333333,
|
| 28556 |
+
"grad_norm": 0.6457788348197937,
|
| 28557 |
+
"learning_rate": 1.8975501113585747e-05,
|
| 28558 |
+
"loss": 0.8879,
|
| 28559 |
+
"step": 4074
|
| 28560 |
+
},
|
| 28561 |
+
{
|
| 28562 |
+
"epoch": 0.9055555555555556,
|
| 28563 |
+
"grad_norm": 1.0111377239227295,
|
| 28564 |
+
"learning_rate": 1.893095768374165e-05,
|
| 28565 |
+
"loss": 1.617,
|
| 28566 |
+
"step": 4075
|
| 28567 |
+
},
|
| 28568 |
+
{
|
| 28569 |
+
"epoch": 0.9057777777777778,
|
| 28570 |
+
"grad_norm": 0.9277496933937073,
|
| 28571 |
+
"learning_rate": 1.888641425389755e-05,
|
| 28572 |
+
"loss": 1.8035,
|
| 28573 |
+
"step": 4076
|
| 28574 |
+
},
|
| 28575 |
+
{
|
| 28576 |
+
"epoch": 0.906,
|
| 28577 |
+
"grad_norm": 1.0849852561950684,
|
| 28578 |
+
"learning_rate": 1.8841870824053453e-05,
|
| 28579 |
+
"loss": 1.6758,
|
| 28580 |
+
"step": 4077
|
| 28581 |
+
},
|
| 28582 |
+
{
|
| 28583 |
+
"epoch": 0.9062222222222223,
|
| 28584 |
+
"grad_norm": 1.02144455909729,
|
| 28585 |
+
"learning_rate": 1.8797327394209355e-05,
|
| 28586 |
+
"loss": 1.812,
|
| 28587 |
+
"step": 4078
|
| 28588 |
+
},
|
| 28589 |
+
{
|
| 28590 |
+
"epoch": 0.9064444444444445,
|
| 28591 |
+
"grad_norm": 1.1183116436004639,
|
| 28592 |
+
"learning_rate": 1.8752783964365257e-05,
|
| 28593 |
+
"loss": 1.8675,
|
| 28594 |
+
"step": 4079
|
| 28595 |
+
},
|
| 28596 |
+
{
|
| 28597 |
+
"epoch": 0.9066666666666666,
|
| 28598 |
+
"grad_norm": 1.3064316511154175,
|
| 28599 |
+
"learning_rate": 1.870824053452116e-05,
|
| 28600 |
+
"loss": 1.7242,
|
| 28601 |
+
"step": 4080
|
| 28602 |
+
},
|
| 28603 |
+
{
|
| 28604 |
+
"epoch": 0.9068888888888889,
|
| 28605 |
+
"grad_norm": 0.06701880693435669,
|
| 28606 |
+
"learning_rate": 1.866369710467706e-05,
|
| 28607 |
+
"loss": 0.0177,
|
| 28608 |
+
"step": 4081
|
| 28609 |
+
},
|
| 28610 |
+
{
|
| 28611 |
+
"epoch": 0.9071111111111111,
|
| 28612 |
+
"grad_norm": 0.06481373310089111,
|
| 28613 |
+
"learning_rate": 1.8619153674832963e-05,
|
| 28614 |
+
"loss": 0.0178,
|
| 28615 |
+
"step": 4082
|
| 28616 |
+
},
|
| 28617 |
+
{
|
| 28618 |
+
"epoch": 0.9073333333333333,
|
| 28619 |
+
"grad_norm": 0.7761397361755371,
|
| 28620 |
+
"learning_rate": 1.8574610244988865e-05,
|
| 28621 |
+
"loss": 0.7757,
|
| 28622 |
+
"step": 4083
|
| 28623 |
+
},
|
| 28624 |
+
{
|
| 28625 |
+
"epoch": 0.9075555555555556,
|
| 28626 |
+
"grad_norm": 1.0291235446929932,
|
| 28627 |
+
"learning_rate": 1.8530066815144767e-05,
|
| 28628 |
+
"loss": 1.6664,
|
| 28629 |
+
"step": 4084
|
| 28630 |
+
},
|
| 28631 |
+
{
|
| 28632 |
+
"epoch": 0.9077777777777778,
|
| 28633 |
+
"grad_norm": 0.7274791598320007,
|
| 28634 |
+
"learning_rate": 1.848552338530067e-05,
|
| 28635 |
+
"loss": 0.979,
|
| 28636 |
+
"step": 4085
|
| 28637 |
+
},
|
| 28638 |
+
{
|
| 28639 |
+
"epoch": 0.908,
|
| 28640 |
+
"grad_norm": 0.6331042647361755,
|
| 28641 |
+
"learning_rate": 1.844097995545657e-05,
|
| 28642 |
+
"loss": 0.6799,
|
| 28643 |
+
"step": 4086
|
| 28644 |
+
},
|
| 28645 |
+
{
|
| 28646 |
+
"epoch": 0.9082222222222223,
|
| 28647 |
+
"grad_norm": 0.08008535206317902,
|
| 28648 |
+
"learning_rate": 1.8396436525612473e-05,
|
| 28649 |
+
"loss": 0.018,
|
| 28650 |
+
"step": 4087
|
| 28651 |
+
},
|
| 28652 |
+
{
|
| 28653 |
+
"epoch": 0.9084444444444445,
|
| 28654 |
+
"grad_norm": 0.08232392370700836,
|
| 28655 |
+
"learning_rate": 1.8351893095768375e-05,
|
| 28656 |
+
"loss": 0.018,
|
| 28657 |
+
"step": 4088
|
| 28658 |
+
},
|
| 28659 |
+
{
|
| 28660 |
+
"epoch": 0.9086666666666666,
|
| 28661 |
+
"grad_norm": 0.6026217341423035,
|
| 28662 |
+
"learning_rate": 1.830734966592428e-05,
|
| 28663 |
+
"loss": 0.8012,
|
| 28664 |
+
"step": 4089
|
| 28665 |
+
},
|
| 28666 |
+
{
|
| 28667 |
+
"epoch": 0.9088888888888889,
|
| 28668 |
+
"grad_norm": 1.0574473142623901,
|
| 28669 |
+
"learning_rate": 1.826280623608018e-05,
|
| 28670 |
+
"loss": 1.7651,
|
| 28671 |
+
"step": 4090
|
| 28672 |
+
},
|
| 28673 |
+
{
|
| 28674 |
+
"epoch": 0.9091111111111111,
|
| 28675 |
+
"grad_norm": 1.1249449253082275,
|
| 28676 |
+
"learning_rate": 1.821826280623608e-05,
|
| 28677 |
+
"loss": 1.6539,
|
| 28678 |
+
"step": 4091
|
| 28679 |
+
},
|
| 28680 |
+
{
|
| 28681 |
+
"epoch": 0.9093333333333333,
|
| 28682 |
+
"grad_norm": 0.7003470659255981,
|
| 28683 |
+
"learning_rate": 1.8173719376391986e-05,
|
| 28684 |
+
"loss": 0.764,
|
| 28685 |
+
"step": 4092
|
| 28686 |
+
},
|
| 28687 |
+
{
|
| 28688 |
+
"epoch": 0.9095555555555556,
|
| 28689 |
+
"grad_norm": 1.0299309492111206,
|
| 28690 |
+
"learning_rate": 1.8129175946547884e-05,
|
| 28691 |
+
"loss": 1.3027,
|
| 28692 |
+
"step": 4093
|
| 28693 |
+
},
|
| 28694 |
+
{
|
| 28695 |
+
"epoch": 0.9097777777777778,
|
| 28696 |
+
"grad_norm": 0.21282123029232025,
|
| 28697 |
+
"learning_rate": 1.8084632516703786e-05,
|
| 28698 |
+
"loss": 0.0309,
|
| 28699 |
+
"step": 4094
|
| 28700 |
+
},
|
| 28701 |
+
{
|
| 28702 |
+
"epoch": 0.91,
|
| 28703 |
+
"grad_norm": 1.1035081148147583,
|
| 28704 |
+
"learning_rate": 1.804008908685969e-05,
|
| 28705 |
+
"loss": 1.2723,
|
| 28706 |
+
"step": 4095
|
| 28707 |
+
},
|
| 28708 |
+
{
|
| 28709 |
+
"epoch": 0.9102222222222223,
|
| 28710 |
+
"grad_norm": 0.9910405874252319,
|
| 28711 |
+
"learning_rate": 1.799554565701559e-05,
|
| 28712 |
+
"loss": 1.0153,
|
| 28713 |
+
"step": 4096
|
| 28714 |
+
},
|
| 28715 |
+
{
|
| 28716 |
+
"epoch": 0.9104444444444444,
|
| 28717 |
+
"grad_norm": 1.0081919431686401,
|
| 28718 |
+
"learning_rate": 1.7951002227171492e-05,
|
| 28719 |
+
"loss": 1.0348,
|
| 28720 |
+
"step": 4097
|
| 28721 |
+
},
|
| 28722 |
+
{
|
| 28723 |
+
"epoch": 0.9106666666666666,
|
| 28724 |
+
"grad_norm": 0.1447010040283203,
|
| 28725 |
+
"learning_rate": 1.7906458797327394e-05,
|
| 28726 |
+
"loss": 0.0327,
|
| 28727 |
+
"step": 4098
|
| 28728 |
+
},
|
| 28729 |
+
{
|
| 28730 |
+
"epoch": 0.9108888888888889,
|
| 28731 |
+
"grad_norm": 1.1237828731536865,
|
| 28732 |
+
"learning_rate": 1.7861915367483296e-05,
|
| 28733 |
+
"loss": 1.332,
|
| 28734 |
+
"step": 4099
|
| 28735 |
+
},
|
| 28736 |
+
{
|
| 28737 |
+
"epoch": 0.9111111111111111,
|
| 28738 |
+
"grad_norm": 1.0061198472976685,
|
| 28739 |
+
"learning_rate": 1.7817371937639198e-05,
|
| 28740 |
+
"loss": 0.9762,
|
| 28741 |
+
"step": 4100
|
| 28742 |
+
},
|
| 28743 |
+
{
|
| 28744 |
+
"epoch": 0.9113333333333333,
|
| 28745 |
+
"grad_norm": 0.045394111424684525,
|
| 28746 |
+
"learning_rate": 1.77728285077951e-05,
|
| 28747 |
+
"loss": 0.0105,
|
| 28748 |
+
"step": 4101
|
| 28749 |
+
},
|
| 28750 |
+
{
|
| 28751 |
+
"epoch": 0.9115555555555556,
|
| 28752 |
+
"grad_norm": 0.6273143291473389,
|
| 28753 |
+
"learning_rate": 1.7728285077951006e-05,
|
| 28754 |
+
"loss": 0.8231,
|
| 28755 |
+
"step": 4102
|
| 28756 |
+
},
|
| 28757 |
+
{
|
| 28758 |
+
"epoch": 0.9117777777777778,
|
| 28759 |
+
"grad_norm": 0.5369709730148315,
|
| 28760 |
+
"learning_rate": 1.7683741648106904e-05,
|
| 28761 |
+
"loss": 1.0971,
|
| 28762 |
+
"step": 4103
|
| 28763 |
+
},
|
| 28764 |
+
{
|
| 28765 |
+
"epoch": 0.912,
|
| 28766 |
+
"grad_norm": 0.841785728931427,
|
| 28767 |
+
"learning_rate": 1.7639198218262806e-05,
|
| 28768 |
+
"loss": 2.1345,
|
| 28769 |
+
"step": 4104
|
| 28770 |
+
},
|
| 28771 |
+
{
|
| 28772 |
+
"epoch": 0.9122222222222223,
|
| 28773 |
+
"grad_norm": 0.5022440552711487,
|
| 28774 |
+
"learning_rate": 1.759465478841871e-05,
|
| 28775 |
+
"loss": 1.0847,
|
| 28776 |
+
"step": 4105
|
| 28777 |
+
},
|
| 28778 |
+
{
|
| 28779 |
+
"epoch": 0.9124444444444444,
|
| 28780 |
+
"grad_norm": 0.5736976265907288,
|
| 28781 |
+
"learning_rate": 1.755011135857461e-05,
|
| 28782 |
+
"loss": 1.0705,
|
| 28783 |
+
"step": 4106
|
| 28784 |
+
},
|
| 28785 |
+
{
|
| 28786 |
+
"epoch": 0.9126666666666666,
|
| 28787 |
+
"grad_norm": 0.7846779227256775,
|
| 28788 |
+
"learning_rate": 1.7505567928730512e-05,
|
| 28789 |
+
"loss": 2.2198,
|
| 28790 |
+
"step": 4107
|
| 28791 |
+
},
|
| 28792 |
+
{
|
| 28793 |
+
"epoch": 0.9128888888888889,
|
| 28794 |
+
"grad_norm": 0.04545416682958603,
|
| 28795 |
+
"learning_rate": 1.7461024498886417e-05,
|
| 28796 |
+
"loss": 0.0105,
|
| 28797 |
+
"step": 4108
|
| 28798 |
+
},
|
| 28799 |
+
{
|
| 28800 |
+
"epoch": 0.9131111111111111,
|
| 28801 |
+
"grad_norm": 0.8995314240455627,
|
| 28802 |
+
"learning_rate": 1.7416481069042316e-05,
|
| 28803 |
+
"loss": 2.0983,
|
| 28804 |
+
"step": 4109
|
| 28805 |
+
},
|
| 28806 |
+
{
|
| 28807 |
+
"epoch": 0.9133333333333333,
|
| 28808 |
+
"grad_norm": 0.08467597514390945,
|
| 28809 |
+
"learning_rate": 1.7371937639198218e-05,
|
| 28810 |
+
"loss": 0.0111,
|
| 28811 |
+
"step": 4110
|
| 28812 |
+
},
|
| 28813 |
+
{
|
| 28814 |
+
"epoch": 0.9135555555555556,
|
| 28815 |
+
"grad_norm": 0.06848177313804626,
|
| 28816 |
+
"learning_rate": 1.732739420935412e-05,
|
| 28817 |
+
"loss": 0.0109,
|
| 28818 |
+
"step": 4111
|
| 28819 |
+
},
|
| 28820 |
+
{
|
| 28821 |
+
"epoch": 0.9137777777777778,
|
| 28822 |
+
"grad_norm": 0.6615252494812012,
|
| 28823 |
+
"learning_rate": 1.7282850779510025e-05,
|
| 28824 |
+
"loss": 1.0423,
|
| 28825 |
+
"step": 4112
|
| 28826 |
+
},
|
| 28827 |
+
{
|
| 28828 |
+
"epoch": 0.914,
|
| 28829 |
+
"grad_norm": 0.8934789896011353,
|
| 28830 |
+
"learning_rate": 1.7238307349665924e-05,
|
| 28831 |
+
"loss": 1.8382,
|
| 28832 |
+
"step": 4113
|
| 28833 |
+
},
|
| 28834 |
+
{
|
| 28835 |
+
"epoch": 0.9142222222222223,
|
| 28836 |
+
"grad_norm": 0.8137645125389099,
|
| 28837 |
+
"learning_rate": 1.7193763919821826e-05,
|
| 28838 |
+
"loss": 1.9163,
|
| 28839 |
+
"step": 4114
|
| 28840 |
+
},
|
| 28841 |
+
{
|
| 28842 |
+
"epoch": 0.9144444444444444,
|
| 28843 |
+
"grad_norm": 0.8993197083473206,
|
| 28844 |
+
"learning_rate": 1.714922048997773e-05,
|
| 28845 |
+
"loss": 2.1383,
|
| 28846 |
+
"step": 4115
|
| 28847 |
+
},
|
| 28848 |
+
{
|
| 28849 |
+
"epoch": 0.9146666666666666,
|
| 28850 |
+
"grad_norm": 0.908676028251648,
|
| 28851 |
+
"learning_rate": 1.710467706013363e-05,
|
| 28852 |
+
"loss": 2.0524,
|
| 28853 |
+
"step": 4116
|
| 28854 |
+
},
|
| 28855 |
+
{
|
| 28856 |
+
"epoch": 0.9148888888888889,
|
| 28857 |
+
"grad_norm": 0.6348316669464111,
|
| 28858 |
+
"learning_rate": 1.7060133630289532e-05,
|
| 28859 |
+
"loss": 0.9211,
|
| 28860 |
+
"step": 4117
|
| 28861 |
+
},
|
| 28862 |
+
{
|
| 28863 |
+
"epoch": 0.9151111111111111,
|
| 28864 |
+
"grad_norm": 0.10803266614675522,
|
| 28865 |
+
"learning_rate": 1.7015590200445437e-05,
|
| 28866 |
+
"loss": 0.0177,
|
| 28867 |
+
"step": 4118
|
| 28868 |
+
},
|
| 28869 |
+
{
|
| 28870 |
+
"epoch": 0.9153333333333333,
|
| 28871 |
+
"grad_norm": 0.5778976678848267,
|
| 28872 |
+
"learning_rate": 1.6971046770601336e-05,
|
| 28873 |
+
"loss": 0.7947,
|
| 28874 |
+
"step": 4119
|
| 28875 |
+
},
|
| 28876 |
+
{
|
| 28877 |
+
"epoch": 0.9155555555555556,
|
| 28878 |
+
"grad_norm": 0.9023910164833069,
|
| 28879 |
+
"learning_rate": 1.6926503340757238e-05,
|
| 28880 |
+
"loss": 1.6568,
|
| 28881 |
+
"step": 4120
|
| 28882 |
+
},
|
| 28883 |
+
{
|
| 28884 |
+
"epoch": 0.9157777777777778,
|
| 28885 |
+
"grad_norm": 0.7427157759666443,
|
| 28886 |
+
"learning_rate": 1.6881959910913143e-05,
|
| 28887 |
+
"loss": 0.6653,
|
| 28888 |
+
"step": 4121
|
| 28889 |
+
},
|
| 28890 |
+
{
|
| 28891 |
+
"epoch": 0.916,
|
| 28892 |
+
"grad_norm": 0.8236956000328064,
|
| 28893 |
+
"learning_rate": 1.683741648106904e-05,
|
| 28894 |
+
"loss": 1.0835,
|
| 28895 |
+
"step": 4122
|
| 28896 |
+
},
|
| 28897 |
+
{
|
| 28898 |
+
"epoch": 0.9162222222222223,
|
| 28899 |
+
"grad_norm": 0.788445234298706,
|
| 28900 |
+
"learning_rate": 1.6792873051224944e-05,
|
| 28901 |
+
"loss": 1.1082,
|
| 28902 |
+
"step": 4123
|
| 28903 |
+
},
|
| 28904 |
+
{
|
| 28905 |
+
"epoch": 0.9164444444444444,
|
| 28906 |
+
"grad_norm": 0.7391776442527771,
|
| 28907 |
+
"learning_rate": 1.674832962138085e-05,
|
| 28908 |
+
"loss": 1.0042,
|
| 28909 |
+
"step": 4124
|
| 28910 |
+
},
|
| 28911 |
+
{
|
| 28912 |
+
"epoch": 0.9166666666666666,
|
| 28913 |
+
"grad_norm": 0.993009626865387,
|
| 28914 |
+
"learning_rate": 1.670378619153675e-05,
|
| 28915 |
+
"loss": 1.7091,
|
| 28916 |
+
"step": 4125
|
| 28917 |
+
},
|
| 28918 |
+
{
|
| 28919 |
+
"epoch": 0.9168888888888889,
|
| 28920 |
+
"grad_norm": 1.1671327352523804,
|
| 28921 |
+
"learning_rate": 1.665924276169265e-05,
|
| 28922 |
+
"loss": 1.8704,
|
| 28923 |
+
"step": 4126
|
| 28924 |
+
},
|
| 28925 |
+
{
|
| 28926 |
+
"epoch": 0.9171111111111111,
|
| 28927 |
+
"grad_norm": 0.9321463704109192,
|
| 28928 |
+
"learning_rate": 1.6614699331848555e-05,
|
| 28929 |
+
"loss": 1.3912,
|
| 28930 |
+
"step": 4127
|
| 28931 |
+
},
|
| 28932 |
+
{
|
| 28933 |
+
"epoch": 0.9173333333333333,
|
| 28934 |
+
"grad_norm": 0.7201982140541077,
|
| 28935 |
+
"learning_rate": 1.6570155902004457e-05,
|
| 28936 |
+
"loss": 1.0362,
|
| 28937 |
+
"step": 4128
|
| 28938 |
+
},
|
| 28939 |
+
{
|
| 28940 |
+
"epoch": 0.9175555555555556,
|
| 28941 |
+
"grad_norm": 0.06468725949525833,
|
| 28942 |
+
"learning_rate": 1.6525612472160355e-05,
|
| 28943 |
+
"loss": 0.0177,
|
| 28944 |
+
"step": 4129
|
| 28945 |
+
},
|
| 28946 |
+
{
|
| 28947 |
+
"epoch": 0.9177777777777778,
|
| 28948 |
+
"grad_norm": 0.06377862393856049,
|
| 28949 |
+
"learning_rate": 1.6481069042316257e-05,
|
| 28950 |
+
"loss": 0.0175,
|
| 28951 |
+
"step": 4130
|
| 28952 |
+
},
|
| 28953 |
+
{
|
| 28954 |
+
"epoch": 0.918,
|
| 28955 |
+
"grad_norm": 1.2018589973449707,
|
| 28956 |
+
"learning_rate": 1.6436525612472163e-05,
|
| 28957 |
+
"loss": 2.088,
|
| 28958 |
+
"step": 4131
|
| 28959 |
+
},
|
| 28960 |
+
{
|
| 28961 |
+
"epoch": 0.9182222222222223,
|
| 28962 |
+
"grad_norm": 0.08640787750482559,
|
| 28963 |
+
"learning_rate": 1.639198218262806e-05,
|
| 28964 |
+
"loss": 0.0178,
|
| 28965 |
+
"step": 4132
|
| 28966 |
+
},
|
| 28967 |
+
{
|
| 28968 |
+
"epoch": 0.9184444444444444,
|
| 28969 |
+
"grad_norm": 0.07626676559448242,
|
| 28970 |
+
"learning_rate": 1.6347438752783963e-05,
|
| 28971 |
+
"loss": 0.0179,
|
| 28972 |
+
"step": 4133
|
| 28973 |
+
},
|
| 28974 |
+
{
|
| 28975 |
+
"epoch": 0.9186666666666666,
|
| 28976 |
+
"grad_norm": 0.684622585773468,
|
| 28977 |
+
"learning_rate": 1.630289532293987e-05,
|
| 28978 |
+
"loss": 0.9346,
|
| 28979 |
+
"step": 4134
|
| 28980 |
+
},
|
| 28981 |
+
{
|
| 28982 |
+
"epoch": 0.9188888888888889,
|
| 28983 |
+
"grad_norm": 1.07980215549469,
|
| 28984 |
+
"learning_rate": 1.625835189309577e-05,
|
| 28985 |
+
"loss": 1.8093,
|
| 28986 |
+
"step": 4135
|
| 28987 |
+
},
|
| 28988 |
+
{
|
| 28989 |
+
"epoch": 0.9191111111111111,
|
| 28990 |
+
"grad_norm": 1.0103257894515991,
|
| 28991 |
+
"learning_rate": 1.621380846325167e-05,
|
| 28992 |
+
"loss": 1.3954,
|
| 28993 |
+
"step": 4136
|
| 28994 |
+
},
|
| 28995 |
+
{
|
| 28996 |
+
"epoch": 0.9193333333333333,
|
| 28997 |
+
"grad_norm": 1.0367659330368042,
|
| 28998 |
+
"learning_rate": 1.6169265033407574e-05,
|
| 28999 |
+
"loss": 1.665,
|
| 29000 |
+
"step": 4137
|
| 29001 |
+
},
|
| 29002 |
+
{
|
| 29003 |
+
"epoch": 0.9195555555555556,
|
| 29004 |
+
"grad_norm": 1.13039231300354,
|
| 29005 |
+
"learning_rate": 1.6124721603563476e-05,
|
| 29006 |
+
"loss": 1.3939,
|
| 29007 |
+
"step": 4138
|
| 29008 |
+
},
|
| 29009 |
+
{
|
| 29010 |
+
"epoch": 0.9197777777777778,
|
| 29011 |
+
"grad_norm": 1.2525602579116821,
|
| 29012 |
+
"learning_rate": 1.6080178173719375e-05,
|
| 29013 |
+
"loss": 1.7406,
|
| 29014 |
+
"step": 4139
|
| 29015 |
+
},
|
| 29016 |
+
{
|
| 29017 |
+
"epoch": 0.92,
|
| 29018 |
+
"grad_norm": 1.0886310338974,
|
| 29019 |
+
"learning_rate": 1.603563474387528e-05,
|
| 29020 |
+
"loss": 1.447,
|
| 29021 |
+
"step": 4140
|
| 29022 |
+
},
|
| 29023 |
+
{
|
| 29024 |
+
"epoch": 0.9202222222222223,
|
| 29025 |
+
"grad_norm": 1.0128674507141113,
|
| 29026 |
+
"learning_rate": 1.5991091314031182e-05,
|
| 29027 |
+
"loss": 1.3665,
|
| 29028 |
+
"step": 4141
|
| 29029 |
+
},
|
| 29030 |
+
{
|
| 29031 |
+
"epoch": 0.9204444444444444,
|
| 29032 |
+
"grad_norm": 1.087297797203064,
|
| 29033 |
+
"learning_rate": 1.594654788418708e-05,
|
| 29034 |
+
"loss": 1.2043,
|
| 29035 |
+
"step": 4142
|
| 29036 |
+
},
|
| 29037 |
+
{
|
| 29038 |
+
"epoch": 0.9206666666666666,
|
| 29039 |
+
"grad_norm": 1.0681723356246948,
|
| 29040 |
+
"learning_rate": 1.5902004454342986e-05,
|
| 29041 |
+
"loss": 1.3818,
|
| 29042 |
+
"step": 4143
|
| 29043 |
+
},
|
| 29044 |
+
{
|
| 29045 |
+
"epoch": 0.9208888888888889,
|
| 29046 |
+
"grad_norm": 0.8169934153556824,
|
| 29047 |
+
"learning_rate": 1.5857461024498888e-05,
|
| 29048 |
+
"loss": 0.7076,
|
| 29049 |
+
"step": 4144
|
| 29050 |
+
},
|
| 29051 |
+
{
|
| 29052 |
+
"epoch": 0.9211111111111111,
|
| 29053 |
+
"grad_norm": 1.1659146547317505,
|
| 29054 |
+
"learning_rate": 1.581291759465479e-05,
|
| 29055 |
+
"loss": 1.3053,
|
| 29056 |
+
"step": 4145
|
| 29057 |
+
},
|
| 29058 |
+
{
|
| 29059 |
+
"epoch": 0.9213333333333333,
|
| 29060 |
+
"grad_norm": 1.0384572744369507,
|
| 29061 |
+
"learning_rate": 1.5768374164810692e-05,
|
| 29062 |
+
"loss": 1.1864,
|
| 29063 |
+
"step": 4146
|
| 29064 |
+
},
|
| 29065 |
+
{
|
| 29066 |
+
"epoch": 0.9215555555555556,
|
| 29067 |
+
"grad_norm": 1.0471240282058716,
|
| 29068 |
+
"learning_rate": 1.5723830734966594e-05,
|
| 29069 |
+
"loss": 1.066,
|
| 29070 |
+
"step": 4147
|
| 29071 |
+
},
|
| 29072 |
+
{
|
| 29073 |
+
"epoch": 0.9217777777777778,
|
| 29074 |
+
"grad_norm": 0.7413065433502197,
|
| 29075 |
+
"learning_rate": 1.5679287305122496e-05,
|
| 29076 |
+
"loss": 0.5414,
|
| 29077 |
+
"step": 4148
|
| 29078 |
+
},
|
| 29079 |
+
{
|
| 29080 |
+
"epoch": 0.922,
|
| 29081 |
+
"grad_norm": 0.9853238463401794,
|
| 29082 |
+
"learning_rate": 1.5634743875278395e-05,
|
| 29083 |
+
"loss": 0.9216,
|
| 29084 |
+
"step": 4149
|
| 29085 |
+
},
|
| 29086 |
+
{
|
| 29087 |
+
"epoch": 0.9222222222222223,
|
| 29088 |
+
"grad_norm": 0.9708325266838074,
|
| 29089 |
+
"learning_rate": 1.55902004454343e-05,
|
| 29090 |
+
"loss": 0.8306,
|
| 29091 |
+
"step": 4150
|
| 29092 |
+
},
|
| 29093 |
+
{
|
| 29094 |
+
"epoch": 0.9224444444444444,
|
| 29095 |
+
"grad_norm": 0.565495491027832,
|
| 29096 |
+
"learning_rate": 1.5545657015590202e-05,
|
| 29097 |
+
"loss": 1.0385,
|
| 29098 |
+
"step": 4151
|
| 29099 |
+
},
|
| 29100 |
+
{
|
| 29101 |
+
"epoch": 0.9226666666666666,
|
| 29102 |
+
"grad_norm": 0.046501629054546356,
|
| 29103 |
+
"learning_rate": 1.55011135857461e-05,
|
| 29104 |
+
"loss": 0.0103,
|
| 29105 |
+
"step": 4152
|
| 29106 |
+
},
|
| 29107 |
+
{
|
| 29108 |
+
"epoch": 0.9228888888888889,
|
| 29109 |
+
"grad_norm": 0.5920565128326416,
|
| 29110 |
+
"learning_rate": 1.5456570155902006e-05,
|
| 29111 |
+
"loss": 1.1167,
|
| 29112 |
+
"step": 4153
|
| 29113 |
+
},
|
| 29114 |
+
{
|
| 29115 |
+
"epoch": 0.9231111111111111,
|
| 29116 |
+
"grad_norm": 0.6111573576927185,
|
| 29117 |
+
"learning_rate": 1.5412026726057908e-05,
|
| 29118 |
+
"loss": 0.9939,
|
| 29119 |
+
"step": 4154
|
| 29120 |
+
},
|
| 29121 |
+
{
|
| 29122 |
+
"epoch": 0.9233333333333333,
|
| 29123 |
+
"grad_norm": 0.5594994425773621,
|
| 29124 |
+
"learning_rate": 1.5367483296213807e-05,
|
| 29125 |
+
"loss": 1.1777,
|
| 29126 |
+
"step": 4155
|
| 29127 |
+
},
|
| 29128 |
+
{
|
| 29129 |
+
"epoch": 0.9235555555555556,
|
| 29130 |
+
"grad_norm": 0.05240090563893318,
|
| 29131 |
+
"learning_rate": 1.5322939866369712e-05,
|
| 29132 |
+
"loss": 0.0107,
|
| 29133 |
+
"step": 4156
|
| 29134 |
+
},
|
| 29135 |
+
{
|
| 29136 |
+
"epoch": 0.9237777777777778,
|
| 29137 |
+
"grad_norm": 0.8332452178001404,
|
| 29138 |
+
"learning_rate": 1.5278396436525614e-05,
|
| 29139 |
+
"loss": 2.1969,
|
| 29140 |
+
"step": 4157
|
| 29141 |
+
},
|
| 29142 |
+
{
|
| 29143 |
+
"epoch": 0.924,
|
| 29144 |
+
"grad_norm": 0.60703045129776,
|
| 29145 |
+
"learning_rate": 1.5233853006681514e-05,
|
| 29146 |
+
"loss": 0.8806,
|
| 29147 |
+
"step": 4158
|
| 29148 |
+
},
|
| 29149 |
+
{
|
| 29150 |
+
"epoch": 0.9242222222222222,
|
| 29151 |
+
"grad_norm": 0.0834372490644455,
|
| 29152 |
+
"learning_rate": 1.5189309576837418e-05,
|
| 29153 |
+
"loss": 0.0109,
|
| 29154 |
+
"step": 4159
|
| 29155 |
+
},
|
| 29156 |
+
{
|
| 29157 |
+
"epoch": 0.9244444444444444,
|
| 29158 |
+
"grad_norm": 0.06800226867198944,
|
| 29159 |
+
"learning_rate": 1.514476614699332e-05,
|
| 29160 |
+
"loss": 0.0111,
|
| 29161 |
+
"step": 4160
|
| 29162 |
+
},
|
| 29163 |
+
{
|
| 29164 |
+
"epoch": 0.9246666666666666,
|
| 29165 |
+
"grad_norm": 0.0833854079246521,
|
| 29166 |
+
"learning_rate": 1.510022271714922e-05,
|
| 29167 |
+
"loss": 0.0112,
|
| 29168 |
+
"step": 4161
|
| 29169 |
+
},
|
| 29170 |
+
{
|
| 29171 |
+
"epoch": 0.9248888888888889,
|
| 29172 |
+
"grad_norm": 0.6188771724700928,
|
| 29173 |
+
"learning_rate": 1.5055679287305124e-05,
|
| 29174 |
+
"loss": 0.9193,
|
| 29175 |
+
"step": 4162
|
| 29176 |
+
},
|
| 29177 |
+
{
|
| 29178 |
+
"epoch": 0.9251111111111111,
|
| 29179 |
+
"grad_norm": 0.9110550284385681,
|
| 29180 |
+
"learning_rate": 1.5011135857461026e-05,
|
| 29181 |
+
"loss": 1.8064,
|
| 29182 |
+
"step": 4163
|
| 29183 |
+
},
|
| 29184 |
+
{
|
| 29185 |
+
"epoch": 0.9253333333333333,
|
| 29186 |
+
"grad_norm": 0.9889512658119202,
|
| 29187 |
+
"learning_rate": 1.4966592427616926e-05,
|
| 29188 |
+
"loss": 2.1231,
|
| 29189 |
+
"step": 4164
|
| 29190 |
+
},
|
| 29191 |
+
{
|
| 29192 |
+
"epoch": 0.9255555555555556,
|
| 29193 |
+
"grad_norm": 0.700734555721283,
|
| 29194 |
+
"learning_rate": 1.4922048997772831e-05,
|
| 29195 |
+
"loss": 0.9772,
|
| 29196 |
+
"step": 4165
|
| 29197 |
+
},
|
| 29198 |
+
{
|
| 29199 |
+
"epoch": 0.9257777777777778,
|
| 29200 |
+
"grad_norm": 0.10837709903717041,
|
| 29201 |
+
"learning_rate": 1.4877505567928732e-05,
|
| 29202 |
+
"loss": 0.0171,
|
| 29203 |
+
"step": 4166
|
| 29204 |
+
},
|
| 29205 |
+
{
|
| 29206 |
+
"epoch": 0.926,
|
| 29207 |
+
"grad_norm": 0.11039458960294724,
|
| 29208 |
+
"learning_rate": 1.4832962138084634e-05,
|
| 29209 |
+
"loss": 0.017,
|
| 29210 |
+
"step": 4167
|
| 29211 |
+
},
|
| 29212 |
+
{
|
| 29213 |
+
"epoch": 0.9262222222222222,
|
| 29214 |
+
"grad_norm": 1.0548503398895264,
|
| 29215 |
+
"learning_rate": 1.4788418708240534e-05,
|
| 29216 |
+
"loss": 2.02,
|
| 29217 |
+
"step": 4168
|
| 29218 |
+
},
|
| 29219 |
+
{
|
| 29220 |
+
"epoch": 0.9264444444444444,
|
| 29221 |
+
"grad_norm": 1.027597427368164,
|
| 29222 |
+
"learning_rate": 1.4743875278396438e-05,
|
| 29223 |
+
"loss": 2.0137,
|
| 29224 |
+
"step": 4169
|
| 29225 |
+
},
|
| 29226 |
+
{
|
| 29227 |
+
"epoch": 0.9266666666666666,
|
| 29228 |
+
"grad_norm": 0.9742117524147034,
|
| 29229 |
+
"learning_rate": 1.469933184855234e-05,
|
| 29230 |
+
"loss": 1.6216,
|
| 29231 |
+
"step": 4170
|
| 29232 |
+
},
|
| 29233 |
+
{
|
| 29234 |
+
"epoch": 0.9268888888888889,
|
| 29235 |
+
"grad_norm": 1.1034184694290161,
|
| 29236 |
+
"learning_rate": 1.465478841870824e-05,
|
| 29237 |
+
"loss": 1.9359,
|
| 29238 |
+
"step": 4171
|
| 29239 |
+
},
|
| 29240 |
+
{
|
| 29241 |
+
"epoch": 0.9271111111111111,
|
| 29242 |
+
"grad_norm": 0.9604677557945251,
|
| 29243 |
+
"learning_rate": 1.4610244988864143e-05,
|
| 29244 |
+
"loss": 1.9562,
|
| 29245 |
+
"step": 4172
|
| 29246 |
+
},
|
| 29247 |
+
{
|
| 29248 |
+
"epoch": 0.9273333333333333,
|
| 29249 |
+
"grad_norm": 0.07002148032188416,
|
| 29250 |
+
"learning_rate": 1.4565701559020045e-05,
|
| 29251 |
+
"loss": 0.0148,
|
| 29252 |
+
"step": 4173
|
| 29253 |
+
},
|
| 29254 |
+
{
|
| 29255 |
+
"epoch": 0.9275555555555556,
|
| 29256 |
+
"grad_norm": 0.07596537470817566,
|
| 29257 |
+
"learning_rate": 1.4521158129175946e-05,
|
| 29258 |
+
"loss": 0.015,
|
| 29259 |
+
"step": 4174
|
| 29260 |
+
},
|
| 29261 |
+
{
|
| 29262 |
+
"epoch": 0.9277777777777778,
|
| 29263 |
+
"grad_norm": 0.07176879793405533,
|
| 29264 |
+
"learning_rate": 1.447661469933185e-05,
|
| 29265 |
+
"loss": 0.0153,
|
| 29266 |
+
"step": 4175
|
| 29267 |
+
},
|
| 29268 |
+
{
|
| 29269 |
+
"epoch": 0.928,
|
| 29270 |
+
"grad_norm": 0.6465597152709961,
|
| 29271 |
+
"learning_rate": 1.4432071269487751e-05,
|
| 29272 |
+
"loss": 0.8603,
|
| 29273 |
+
"step": 4176
|
| 29274 |
+
},
|
| 29275 |
+
{
|
| 29276 |
+
"epoch": 0.9282222222222222,
|
| 29277 |
+
"grad_norm": 0.7567986249923706,
|
| 29278 |
+
"learning_rate": 1.4387527839643652e-05,
|
| 29279 |
+
"loss": 0.974,
|
| 29280 |
+
"step": 4177
|
| 29281 |
+
},
|
| 29282 |
+
{
|
| 29283 |
+
"epoch": 0.9284444444444444,
|
| 29284 |
+
"grad_norm": 0.9745081067085266,
|
| 29285 |
+
"learning_rate": 1.4342984409799557e-05,
|
| 29286 |
+
"loss": 1.7431,
|
| 29287 |
+
"step": 4178
|
| 29288 |
+
},
|
| 29289 |
+
{
|
| 29290 |
+
"epoch": 0.9286666666666666,
|
| 29291 |
+
"grad_norm": 1.2606123685836792,
|
| 29292 |
+
"learning_rate": 1.4298440979955457e-05,
|
| 29293 |
+
"loss": 1.8718,
|
| 29294 |
+
"step": 4179
|
| 29295 |
+
},
|
| 29296 |
+
{
|
| 29297 |
+
"epoch": 0.9288888888888889,
|
| 29298 |
+
"grad_norm": 1.018596887588501,
|
| 29299 |
+
"learning_rate": 1.425389755011136e-05,
|
| 29300 |
+
"loss": 1.7782,
|
| 29301 |
+
"step": 4180
|
| 29302 |
+
},
|
| 29303 |
+
{
|
| 29304 |
+
"epoch": 0.9291111111111111,
|
| 29305 |
+
"grad_norm": 0.961146891117096,
|
| 29306 |
+
"learning_rate": 1.4209354120267263e-05,
|
| 29307 |
+
"loss": 1.6335,
|
| 29308 |
+
"step": 4181
|
| 29309 |
+
},
|
| 29310 |
+
{
|
| 29311 |
+
"epoch": 0.9293333333333333,
|
| 29312 |
+
"grad_norm": 0.9671225547790527,
|
| 29313 |
+
"learning_rate": 1.4164810690423163e-05,
|
| 29314 |
+
"loss": 1.507,
|
| 29315 |
+
"step": 4182
|
| 29316 |
+
},
|
| 29317 |
+
{
|
| 29318 |
+
"epoch": 0.9295555555555556,
|
| 29319 |
+
"grad_norm": 0.06404636800289154,
|
| 29320 |
+
"learning_rate": 1.4120267260579065e-05,
|
| 29321 |
+
"loss": 0.0178,
|
| 29322 |
+
"step": 4183
|
| 29323 |
+
},
|
| 29324 |
+
{
|
| 29325 |
+
"epoch": 0.9297777777777778,
|
| 29326 |
+
"grad_norm": 0.06602248549461365,
|
| 29327 |
+
"learning_rate": 1.4075723830734965e-05,
|
| 29328 |
+
"loss": 0.0179,
|
| 29329 |
+
"step": 4184
|
| 29330 |
+
},
|
| 29331 |
+
{
|
| 29332 |
+
"epoch": 0.93,
|
| 29333 |
+
"grad_norm": 0.06565021723508835,
|
| 29334 |
+
"learning_rate": 1.4031180400890869e-05,
|
| 29335 |
+
"loss": 0.0178,
|
| 29336 |
+
"step": 4185
|
| 29337 |
+
},
|
| 29338 |
+
{
|
| 29339 |
+
"epoch": 0.9302222222222222,
|
| 29340 |
+
"grad_norm": 1.20155930519104,
|
| 29341 |
+
"learning_rate": 1.3986636971046771e-05,
|
| 29342 |
+
"loss": 1.839,
|
| 29343 |
+
"step": 4186
|
| 29344 |
+
},
|
| 29345 |
+
{
|
| 29346 |
+
"epoch": 0.9304444444444444,
|
| 29347 |
+
"grad_norm": 0.08105297386646271,
|
| 29348 |
+
"learning_rate": 1.3942093541202671e-05,
|
| 29349 |
+
"loss": 0.0171,
|
| 29350 |
+
"step": 4187
|
| 29351 |
+
},
|
| 29352 |
+
{
|
| 29353 |
+
"epoch": 0.9306666666666666,
|
| 29354 |
+
"grad_norm": 0.07562917470932007,
|
| 29355 |
+
"learning_rate": 1.3897550111358577e-05,
|
| 29356 |
+
"loss": 0.0173,
|
| 29357 |
+
"step": 4188
|
| 29358 |
+
},
|
| 29359 |
+
{
|
| 29360 |
+
"epoch": 0.9308888888888889,
|
| 29361 |
+
"grad_norm": 0.9316315054893494,
|
| 29362 |
+
"learning_rate": 1.3853006681514477e-05,
|
| 29363 |
+
"loss": 0.9604,
|
| 29364 |
+
"step": 4189
|
| 29365 |
+
},
|
| 29366 |
+
{
|
| 29367 |
+
"epoch": 0.9311111111111111,
|
| 29368 |
+
"grad_norm": 1.0534974336624146,
|
| 29369 |
+
"learning_rate": 1.3808463251670379e-05,
|
| 29370 |
+
"loss": 1.5912,
|
| 29371 |
+
"step": 4190
|
| 29372 |
+
},
|
| 29373 |
+
{
|
| 29374 |
+
"epoch": 0.9313333333333333,
|
| 29375 |
+
"grad_norm": 1.1811798810958862,
|
| 29376 |
+
"learning_rate": 1.3763919821826283e-05,
|
| 29377 |
+
"loss": 1.76,
|
| 29378 |
+
"step": 4191
|
| 29379 |
+
},
|
| 29380 |
+
{
|
| 29381 |
+
"epoch": 0.9315555555555556,
|
| 29382 |
+
"grad_norm": 1.158895492553711,
|
| 29383 |
+
"learning_rate": 1.3719376391982183e-05,
|
| 29384 |
+
"loss": 1.6761,
|
| 29385 |
+
"step": 4192
|
| 29386 |
+
},
|
| 29387 |
+
{
|
| 29388 |
+
"epoch": 0.9317777777777778,
|
| 29389 |
+
"grad_norm": 1.1721011400222778,
|
| 29390 |
+
"learning_rate": 1.3674832962138085e-05,
|
| 29391 |
+
"loss": 1.3907,
|
| 29392 |
+
"step": 4193
|
| 29393 |
+
},
|
| 29394 |
+
{
|
| 29395 |
+
"epoch": 0.932,
|
| 29396 |
+
"grad_norm": 0.180902361869812,
|
| 29397 |
+
"learning_rate": 1.3630289532293989e-05,
|
| 29398 |
+
"loss": 0.0291,
|
| 29399 |
+
"step": 4194
|
| 29400 |
+
},
|
| 29401 |
+
{
|
| 29402 |
+
"epoch": 0.9322222222222222,
|
| 29403 |
+
"grad_norm": 0.17803487181663513,
|
| 29404 |
+
"learning_rate": 1.3585746102449889e-05,
|
| 29405 |
+
"loss": 0.029,
|
| 29406 |
+
"step": 4195
|
| 29407 |
+
},
|
| 29408 |
+
{
|
| 29409 |
+
"epoch": 0.9324444444444444,
|
| 29410 |
+
"grad_norm": 1.2288343906402588,
|
| 29411 |
+
"learning_rate": 1.354120267260579e-05,
|
| 29412 |
+
"loss": 1.5034,
|
| 29413 |
+
"step": 4196
|
| 29414 |
+
},
|
| 29415 |
+
{
|
| 29416 |
+
"epoch": 0.9326666666666666,
|
| 29417 |
+
"grad_norm": 1.1452394723892212,
|
| 29418 |
+
"learning_rate": 1.3496659242761694e-05,
|
| 29419 |
+
"loss": 1.3718,
|
| 29420 |
+
"step": 4197
|
| 29421 |
+
},
|
| 29422 |
+
{
|
| 29423 |
+
"epoch": 0.9328888888888889,
|
| 29424 |
+
"grad_norm": 1.1846433877944946,
|
| 29425 |
+
"learning_rate": 1.3452115812917596e-05,
|
| 29426 |
+
"loss": 1.1493,
|
| 29427 |
+
"step": 4198
|
| 29428 |
+
},
|
| 29429 |
+
{
|
| 29430 |
+
"epoch": 0.9331111111111111,
|
| 29431 |
+
"grad_norm": 1.074758529663086,
|
| 29432 |
+
"learning_rate": 1.3407572383073497e-05,
|
| 29433 |
+
"loss": 1.3701,
|
| 29434 |
+
"step": 4199
|
| 29435 |
+
},
|
| 29436 |
+
{
|
| 29437 |
+
"epoch": 0.9333333333333333,
|
| 29438 |
+
"grad_norm": 1.0819156169891357,
|
| 29439 |
+
"learning_rate": 1.33630289532294e-05,
|
| 29440 |
+
"loss": 0.4173,
|
| 29441 |
+
"step": 4200
|
| 29442 |
+
},
|
| 29443 |
+
{
|
| 29444 |
+
"epoch": 0.9335555555555556,
|
| 29445 |
+
"grad_norm": 0.04741929471492767,
|
| 29446 |
+
"learning_rate": 1.3318485523385302e-05,
|
| 29447 |
+
"loss": 0.0103,
|
| 29448 |
+
"step": 4201
|
| 29449 |
+
},
|
| 29450 |
+
{
|
| 29451 |
+
"epoch": 0.9337777777777778,
|
| 29452 |
+
"grad_norm": 0.5368507504463196,
|
| 29453 |
+
"learning_rate": 1.3273942093541203e-05,
|
| 29454 |
+
"loss": 0.9904,
|
| 29455 |
+
"step": 4202
|
| 29456 |
+
},
|
| 29457 |
+
{
|
| 29458 |
+
"epoch": 0.934,
|
| 29459 |
+
"grad_norm": 0.9232761859893799,
|
| 29460 |
+
"learning_rate": 1.3229398663697105e-05,
|
| 29461 |
+
"loss": 2.0784,
|
| 29462 |
+
"step": 4203
|
| 29463 |
+
},
|
| 29464 |
+
{
|
| 29465 |
+
"epoch": 0.9342222222222222,
|
| 29466 |
+
"grad_norm": 0.8134653568267822,
|
| 29467 |
+
"learning_rate": 1.3184855233853008e-05,
|
| 29468 |
+
"loss": 1.8798,
|
| 29469 |
+
"step": 4204
|
| 29470 |
+
},
|
| 29471 |
+
{
|
| 29472 |
+
"epoch": 0.9344444444444444,
|
| 29473 |
+
"grad_norm": 0.6048870086669922,
|
| 29474 |
+
"learning_rate": 1.3140311804008909e-05,
|
| 29475 |
+
"loss": 1.0963,
|
| 29476 |
+
"step": 4205
|
| 29477 |
+
},
|
| 29478 |
+
{
|
| 29479 |
+
"epoch": 0.9346666666666666,
|
| 29480 |
+
"grad_norm": 0.6360740065574646,
|
| 29481 |
+
"learning_rate": 1.309576837416481e-05,
|
| 29482 |
+
"loss": 1.0909,
|
| 29483 |
+
"step": 4206
|
| 29484 |
+
},
|
| 29485 |
+
{
|
| 29486 |
+
"epoch": 0.9348888888888889,
|
| 29487 |
+
"grad_norm": 0.6409539580345154,
|
| 29488 |
+
"learning_rate": 1.3051224944320714e-05,
|
| 29489 |
+
"loss": 0.9653,
|
| 29490 |
+
"step": 4207
|
| 29491 |
+
},
|
| 29492 |
+
{
|
| 29493 |
+
"epoch": 0.9351111111111111,
|
| 29494 |
+
"grad_norm": 0.8436147570610046,
|
| 29495 |
+
"learning_rate": 1.3006681514476614e-05,
|
| 29496 |
+
"loss": 2.0953,
|
| 29497 |
+
"step": 4208
|
| 29498 |
+
},
|
| 29499 |
+
{
|
| 29500 |
+
"epoch": 0.9353333333333333,
|
| 29501 |
+
"grad_norm": 0.06498900055885315,
|
| 29502 |
+
"learning_rate": 1.2962138084632516e-05,
|
| 29503 |
+
"loss": 0.0108,
|
| 29504 |
+
"step": 4209
|
| 29505 |
+
},
|
| 29506 |
+
{
|
| 29507 |
+
"epoch": 0.9355555555555556,
|
| 29508 |
+
"grad_norm": 0.06997820734977722,
|
| 29509 |
+
"learning_rate": 1.291759465478842e-05,
|
| 29510 |
+
"loss": 0.0104,
|
| 29511 |
+
"step": 4210
|
| 29512 |
+
},
|
| 29513 |
+
{
|
| 29514 |
+
"epoch": 0.9357777777777778,
|
| 29515 |
+
"grad_norm": 0.07132133841514587,
|
| 29516 |
+
"learning_rate": 1.2873051224944322e-05,
|
| 29517 |
+
"loss": 0.0105,
|
| 29518 |
+
"step": 4211
|
| 29519 |
+
},
|
| 29520 |
+
{
|
| 29521 |
+
"epoch": 0.936,
|
| 29522 |
+
"grad_norm": 0.880042552947998,
|
| 29523 |
+
"learning_rate": 1.2828507795100222e-05,
|
| 29524 |
+
"loss": 2.0466,
|
| 29525 |
+
"step": 4212
|
| 29526 |
+
},
|
| 29527 |
+
{
|
| 29528 |
+
"epoch": 0.9362222222222222,
|
| 29529 |
+
"grad_norm": 0.9125528931617737,
|
| 29530 |
+
"learning_rate": 1.2783964365256126e-05,
|
| 29531 |
+
"loss": 1.4838,
|
| 29532 |
+
"step": 4213
|
| 29533 |
+
},
|
| 29534 |
+
{
|
| 29535 |
+
"epoch": 0.9364444444444444,
|
| 29536 |
+
"grad_norm": 0.8474193811416626,
|
| 29537 |
+
"learning_rate": 1.2739420935412028e-05,
|
| 29538 |
+
"loss": 1.9134,
|
| 29539 |
+
"step": 4214
|
| 29540 |
+
},
|
| 29541 |
+
{
|
| 29542 |
+
"epoch": 0.9366666666666666,
|
| 29543 |
+
"grad_norm": 0.9230953454971313,
|
| 29544 |
+
"learning_rate": 1.2694877505567928e-05,
|
| 29545 |
+
"loss": 1.9544,
|
| 29546 |
+
"step": 4215
|
| 29547 |
+
},
|
| 29548 |
+
{
|
| 29549 |
+
"epoch": 0.9368888888888889,
|
| 29550 |
+
"grad_norm": 0.9458275437355042,
|
| 29551 |
+
"learning_rate": 1.2650334075723832e-05,
|
| 29552 |
+
"loss": 1.9736,
|
| 29553 |
+
"step": 4216
|
| 29554 |
+
},
|
| 29555 |
+
{
|
| 29556 |
+
"epoch": 0.9371111111111111,
|
| 29557 |
+
"grad_norm": 1.3373515605926514,
|
| 29558 |
+
"learning_rate": 1.2605790645879734e-05,
|
| 29559 |
+
"loss": 2.2127,
|
| 29560 |
+
"step": 4217
|
| 29561 |
+
},
|
| 29562 |
+
{
|
| 29563 |
+
"epoch": 0.9373333333333334,
|
| 29564 |
+
"grad_norm": 0.09605936706066132,
|
| 29565 |
+
"learning_rate": 1.2561247216035634e-05,
|
| 29566 |
+
"loss": 0.0156,
|
| 29567 |
+
"step": 4218
|
| 29568 |
+
},
|
| 29569 |
+
{
|
| 29570 |
+
"epoch": 0.9375555555555556,
|
| 29571 |
+
"grad_norm": 0.9786912798881531,
|
| 29572 |
+
"learning_rate": 1.251670378619154e-05,
|
| 29573 |
+
"loss": 1.8977,
|
| 29574 |
+
"step": 4219
|
| 29575 |
+
},
|
| 29576 |
+
{
|
| 29577 |
+
"epoch": 0.9377777777777778,
|
| 29578 |
+
"grad_norm": 0.9807332754135132,
|
| 29579 |
+
"learning_rate": 1.247216035634744e-05,
|
| 29580 |
+
"loss": 1.8869,
|
| 29581 |
+
"step": 4220
|
| 29582 |
+
},
|
| 29583 |
+
{
|
| 29584 |
+
"epoch": 0.938,
|
| 29585 |
+
"grad_norm": 0.8643897175788879,
|
| 29586 |
+
"learning_rate": 1.2427616926503342e-05,
|
| 29587 |
+
"loss": 1.6852,
|
| 29588 |
+
"step": 4221
|
| 29589 |
+
},
|
| 29590 |
+
{
|
| 29591 |
+
"epoch": 0.9382222222222222,
|
| 29592 |
+
"grad_norm": 1.1636970043182373,
|
| 29593 |
+
"learning_rate": 1.2383073496659244e-05,
|
| 29594 |
+
"loss": 1.5879,
|
| 29595 |
+
"step": 4222
|
| 29596 |
+
},
|
| 29597 |
+
{
|
| 29598 |
+
"epoch": 0.9384444444444444,
|
| 29599 |
+
"grad_norm": 0.0700439065694809,
|
| 29600 |
+
"learning_rate": 1.2338530066815146e-05,
|
| 29601 |
+
"loss": 0.015,
|
| 29602 |
+
"step": 4223
|
| 29603 |
+
},
|
| 29604 |
+
{
|
| 29605 |
+
"epoch": 0.9386666666666666,
|
| 29606 |
+
"grad_norm": 0.07256254553794861,
|
| 29607 |
+
"learning_rate": 1.2293986636971048e-05,
|
| 29608 |
+
"loss": 0.015,
|
| 29609 |
+
"step": 4224
|
| 29610 |
+
},
|
| 29611 |
+
{
|
| 29612 |
+
"epoch": 0.9388888888888889,
|
| 29613 |
+
"grad_norm": 0.6184701919555664,
|
| 29614 |
+
"learning_rate": 1.224944320712695e-05,
|
| 29615 |
+
"loss": 0.8797,
|
| 29616 |
+
"step": 4225
|
| 29617 |
+
},
|
| 29618 |
+
{
|
| 29619 |
+
"epoch": 0.9391111111111111,
|
| 29620 |
+
"grad_norm": 0.9730594158172607,
|
| 29621 |
+
"learning_rate": 1.2204899777282852e-05,
|
| 29622 |
+
"loss": 1.6952,
|
| 29623 |
+
"step": 4226
|
| 29624 |
+
},
|
| 29625 |
+
{
|
| 29626 |
+
"epoch": 0.9393333333333334,
|
| 29627 |
+
"grad_norm": 0.8093335628509521,
|
| 29628 |
+
"learning_rate": 1.2160356347438754e-05,
|
| 29629 |
+
"loss": 1.7324,
|
| 29630 |
+
"step": 4227
|
| 29631 |
+
},
|
| 29632 |
+
{
|
| 29633 |
+
"epoch": 0.9395555555555556,
|
| 29634 |
+
"grad_norm": 1.0205293893814087,
|
| 29635 |
+
"learning_rate": 1.2115812917594656e-05,
|
| 29636 |
+
"loss": 1.6117,
|
| 29637 |
+
"step": 4228
|
| 29638 |
+
},
|
| 29639 |
+
{
|
| 29640 |
+
"epoch": 0.9397777777777778,
|
| 29641 |
+
"grad_norm": 0.9411073327064514,
|
| 29642 |
+
"learning_rate": 1.2071269487750557e-05,
|
| 29643 |
+
"loss": 1.6911,
|
| 29644 |
+
"step": 4229
|
| 29645 |
+
},
|
| 29646 |
+
{
|
| 29647 |
+
"epoch": 0.94,
|
| 29648 |
+
"grad_norm": 1.1085401773452759,
|
| 29649 |
+
"learning_rate": 1.202672605790646e-05,
|
| 29650 |
+
"loss": 1.822,
|
| 29651 |
+
"step": 4230
|
| 29652 |
+
},
|
| 29653 |
+
{
|
| 29654 |
+
"epoch": 0.9402222222222222,
|
| 29655 |
+
"grad_norm": 1.0736027956008911,
|
| 29656 |
+
"learning_rate": 1.1982182628062361e-05,
|
| 29657 |
+
"loss": 1.7295,
|
| 29658 |
+
"step": 4231
|
| 29659 |
+
},
|
| 29660 |
+
{
|
| 29661 |
+
"epoch": 0.9404444444444444,
|
| 29662 |
+
"grad_norm": 1.0651301145553589,
|
| 29663 |
+
"learning_rate": 1.1937639198218263e-05,
|
| 29664 |
+
"loss": 1.6569,
|
| 29665 |
+
"step": 4232
|
| 29666 |
+
},
|
| 29667 |
+
{
|
| 29668 |
+
"epoch": 0.9406666666666667,
|
| 29669 |
+
"grad_norm": 0.9908804297447205,
|
| 29670 |
+
"learning_rate": 1.1893095768374165e-05,
|
| 29671 |
+
"loss": 1.7012,
|
| 29672 |
+
"step": 4233
|
| 29673 |
+
},
|
| 29674 |
+
{
|
| 29675 |
+
"epoch": 0.9408888888888889,
|
| 29676 |
+
"grad_norm": 0.9733399748802185,
|
| 29677 |
+
"learning_rate": 1.1848552338530067e-05,
|
| 29678 |
+
"loss": 1.7654,
|
| 29679 |
+
"step": 4234
|
| 29680 |
+
},
|
| 29681 |
+
{
|
| 29682 |
+
"epoch": 0.9411111111111111,
|
| 29683 |
+
"grad_norm": 0.8500748872756958,
|
| 29684 |
+
"learning_rate": 1.180400890868597e-05,
|
| 29685 |
+
"loss": 0.8944,
|
| 29686 |
+
"step": 4235
|
| 29687 |
+
},
|
| 29688 |
+
{
|
| 29689 |
+
"epoch": 0.9413333333333334,
|
| 29690 |
+
"grad_norm": 0.7698972821235657,
|
| 29691 |
+
"learning_rate": 1.1759465478841871e-05,
|
| 29692 |
+
"loss": 0.6535,
|
| 29693 |
+
"step": 4236
|
| 29694 |
+
},
|
| 29695 |
+
{
|
| 29696 |
+
"epoch": 0.9415555555555556,
|
| 29697 |
+
"grad_norm": 0.976285994052887,
|
| 29698 |
+
"learning_rate": 1.1714922048997773e-05,
|
| 29699 |
+
"loss": 1.5464,
|
| 29700 |
+
"step": 4237
|
| 29701 |
+
},
|
| 29702 |
+
{
|
| 29703 |
+
"epoch": 0.9417777777777778,
|
| 29704 |
+
"grad_norm": 0.07256097346544266,
|
| 29705 |
+
"learning_rate": 1.1670378619153675e-05,
|
| 29706 |
+
"loss": 0.0177,
|
| 29707 |
+
"step": 4238
|
| 29708 |
+
},
|
| 29709 |
+
{
|
| 29710 |
+
"epoch": 0.942,
|
| 29711 |
+
"grad_norm": 0.8034363389015198,
|
| 29712 |
+
"learning_rate": 1.1625835189309577e-05,
|
| 29713 |
+
"loss": 0.7093,
|
| 29714 |
+
"step": 4239
|
| 29715 |
+
},
|
| 29716 |
+
{
|
| 29717 |
+
"epoch": 0.9422222222222222,
|
| 29718 |
+
"grad_norm": 1.131561279296875,
|
| 29719 |
+
"learning_rate": 1.158129175946548e-05,
|
| 29720 |
+
"loss": 1.6093,
|
| 29721 |
+
"step": 4240
|
| 29722 |
+
},
|
| 29723 |
+
{
|
| 29724 |
+
"epoch": 0.9424444444444444,
|
| 29725 |
+
"grad_norm": 0.6806484460830688,
|
| 29726 |
+
"learning_rate": 1.1536748329621381e-05,
|
| 29727 |
+
"loss": 0.5989,
|
| 29728 |
+
"step": 4241
|
| 29729 |
+
},
|
| 29730 |
+
{
|
| 29731 |
+
"epoch": 0.9426666666666667,
|
| 29732 |
+
"grad_norm": 0.7716354727745056,
|
| 29733 |
+
"learning_rate": 1.1492204899777285e-05,
|
| 29734 |
+
"loss": 0.8757,
|
| 29735 |
+
"step": 4242
|
| 29736 |
+
},
|
| 29737 |
+
{
|
| 29738 |
+
"epoch": 0.9428888888888889,
|
| 29739 |
+
"grad_norm": 1.0691800117492676,
|
| 29740 |
+
"learning_rate": 1.1447661469933185e-05,
|
| 29741 |
+
"loss": 1.525,
|
| 29742 |
+
"step": 4243
|
| 29743 |
+
},
|
| 29744 |
+
{
|
| 29745 |
+
"epoch": 0.9431111111111111,
|
| 29746 |
+
"grad_norm": 1.1259671449661255,
|
| 29747 |
+
"learning_rate": 1.1403118040089087e-05,
|
| 29748 |
+
"loss": 1.3549,
|
| 29749 |
+
"step": 4244
|
| 29750 |
+
},
|
| 29751 |
+
{
|
| 29752 |
+
"epoch": 0.9433333333333334,
|
| 29753 |
+
"grad_norm": 1.147953987121582,
|
| 29754 |
+
"learning_rate": 1.1358574610244989e-05,
|
| 29755 |
+
"loss": 1.5969,
|
| 29756 |
+
"step": 4245
|
| 29757 |
+
},
|
| 29758 |
+
{
|
| 29759 |
+
"epoch": 0.9435555555555556,
|
| 29760 |
+
"grad_norm": 0.932826578617096,
|
| 29761 |
+
"learning_rate": 1.1314031180400891e-05,
|
| 29762 |
+
"loss": 1.3558,
|
| 29763 |
+
"step": 4246
|
| 29764 |
+
},
|
| 29765 |
+
{
|
| 29766 |
+
"epoch": 0.9437777777777778,
|
| 29767 |
+
"grad_norm": 1.2732888460159302,
|
| 29768 |
+
"learning_rate": 1.1269487750556795e-05,
|
| 29769 |
+
"loss": 1.2856,
|
| 29770 |
+
"step": 4247
|
| 29771 |
+
},
|
| 29772 |
+
{
|
| 29773 |
+
"epoch": 0.944,
|
| 29774 |
+
"grad_norm": 1.4992557764053345,
|
| 29775 |
+
"learning_rate": 1.1224944320712695e-05,
|
| 29776 |
+
"loss": 1.1909,
|
| 29777 |
+
"step": 4248
|
| 29778 |
+
},
|
| 29779 |
+
{
|
| 29780 |
+
"epoch": 0.9442222222222222,
|
| 29781 |
+
"grad_norm": 0.13996893167495728,
|
| 29782 |
+
"learning_rate": 1.1180400890868597e-05,
|
| 29783 |
+
"loss": 0.0324,
|
| 29784 |
+
"step": 4249
|
| 29785 |
+
},
|
| 29786 |
+
{
|
| 29787 |
+
"epoch": 0.9444444444444444,
|
| 29788 |
+
"grad_norm": 1.2993894815444946,
|
| 29789 |
+
"learning_rate": 1.11358574610245e-05,
|
| 29790 |
+
"loss": 1.1957,
|
| 29791 |
+
"step": 4250
|
| 29792 |
+
},
|
| 29793 |
+
{
|
| 29794 |
+
"epoch": 0.9446666666666667,
|
| 29795 |
+
"grad_norm": 0.5980258584022522,
|
| 29796 |
+
"learning_rate": 1.10913140311804e-05,
|
| 29797 |
+
"loss": 1.0878,
|
| 29798 |
+
"step": 4251
|
| 29799 |
+
},
|
| 29800 |
+
{
|
| 29801 |
+
"epoch": 0.9448888888888889,
|
| 29802 |
+
"grad_norm": 0.044246070086956024,
|
| 29803 |
+
"learning_rate": 1.1046770601336305e-05,
|
| 29804 |
+
"loss": 0.0103,
|
| 29805 |
+
"step": 4252
|
| 29806 |
+
},
|
| 29807 |
+
{
|
| 29808 |
+
"epoch": 0.9451111111111111,
|
| 29809 |
+
"grad_norm": 0.04438783973455429,
|
| 29810 |
+
"learning_rate": 1.1002227171492205e-05,
|
| 29811 |
+
"loss": 0.0103,
|
| 29812 |
+
"step": 4253
|
| 29813 |
+
},
|
| 29814 |
+
{
|
| 29815 |
+
"epoch": 0.9453333333333334,
|
| 29816 |
+
"grad_norm": 0.5702335834503174,
|
| 29817 |
+
"learning_rate": 1.0957683741648107e-05,
|
| 29818 |
+
"loss": 1.2407,
|
| 29819 |
+
"step": 4254
|
| 29820 |
+
},
|
| 29821 |
+
{
|
| 29822 |
+
"epoch": 0.9455555555555556,
|
| 29823 |
+
"grad_norm": 0.5994575023651123,
|
| 29824 |
+
"learning_rate": 1.091314031180401e-05,
|
| 29825 |
+
"loss": 0.974,
|
| 29826 |
+
"step": 4255
|
| 29827 |
+
},
|
| 29828 |
+
{
|
| 29829 |
+
"epoch": 0.9457777777777778,
|
| 29830 |
+
"grad_norm": 0.916534423828125,
|
| 29831 |
+
"learning_rate": 1.086859688195991e-05,
|
| 29832 |
+
"loss": 2.1087,
|
| 29833 |
+
"step": 4256
|
| 29834 |
+
},
|
| 29835 |
+
{
|
| 29836 |
+
"epoch": 0.946,
|
| 29837 |
+
"grad_norm": 0.8901073336601257,
|
| 29838 |
+
"learning_rate": 1.0824053452115813e-05,
|
| 29839 |
+
"loss": 2.4281,
|
| 29840 |
+
"step": 4257
|
| 29841 |
+
},
|
| 29842 |
+
{
|
| 29843 |
+
"epoch": 0.9462222222222222,
|
| 29844 |
+
"grad_norm": 0.9570392966270447,
|
| 29845 |
+
"learning_rate": 1.0779510022271716e-05,
|
| 29846 |
+
"loss": 1.9183,
|
| 29847 |
+
"step": 4258
|
| 29848 |
+
},
|
| 29849 |
+
{
|
| 29850 |
+
"epoch": 0.9464444444444444,
|
| 29851 |
+
"grad_norm": 0.880129873752594,
|
| 29852 |
+
"learning_rate": 1.0734966592427617e-05,
|
| 29853 |
+
"loss": 2.15,
|
| 29854 |
+
"step": 4259
|
| 29855 |
+
},
|
| 29856 |
+
{
|
| 29857 |
+
"epoch": 0.9466666666666667,
|
| 29858 |
+
"grad_norm": 0.06885567307472229,
|
| 29859 |
+
"learning_rate": 1.069042316258352e-05,
|
| 29860 |
+
"loss": 0.0106,
|
| 29861 |
+
"step": 4260
|
| 29862 |
+
},
|
| 29863 |
+
{
|
| 29864 |
+
"epoch": 0.9468888888888889,
|
| 29865 |
+
"grad_norm": 0.8540828824043274,
|
| 29866 |
+
"learning_rate": 1.064587973273942e-05,
|
| 29867 |
+
"loss": 1.9726,
|
| 29868 |
+
"step": 4261
|
| 29869 |
+
},
|
| 29870 |
+
{
|
| 29871 |
+
"epoch": 0.9471111111111111,
|
| 29872 |
+
"grad_norm": 0.9908187985420227,
|
| 29873 |
+
"learning_rate": 1.0601336302895323e-05,
|
| 29874 |
+
"loss": 2.4167,
|
| 29875 |
+
"step": 4262
|
| 29876 |
+
},
|
| 29877 |
+
{
|
| 29878 |
+
"epoch": 0.9473333333333334,
|
| 29879 |
+
"grad_norm": 0.9318075180053711,
|
| 29880 |
+
"learning_rate": 1.0556792873051226e-05,
|
| 29881 |
+
"loss": 1.9713,
|
| 29882 |
+
"step": 4263
|
| 29883 |
+
},
|
| 29884 |
+
{
|
| 29885 |
+
"epoch": 0.9475555555555556,
|
| 29886 |
+
"grad_norm": 0.9172812700271606,
|
| 29887 |
+
"learning_rate": 1.0512249443207126e-05,
|
| 29888 |
+
"loss": 1.9912,
|
| 29889 |
+
"step": 4264
|
| 29890 |
+
},
|
| 29891 |
+
{
|
| 29892 |
+
"epoch": 0.9477777777777778,
|
| 29893 |
+
"grad_norm": 0.6781771779060364,
|
| 29894 |
+
"learning_rate": 1.046770601336303e-05,
|
| 29895 |
+
"loss": 1.0824,
|
| 29896 |
+
"step": 4265
|
| 29897 |
+
},
|
| 29898 |
+
{
|
| 29899 |
+
"epoch": 0.948,
|
| 29900 |
+
"grad_norm": 0.7084238529205322,
|
| 29901 |
+
"learning_rate": 1.0423162583518932e-05,
|
| 29902 |
+
"loss": 0.9626,
|
| 29903 |
+
"step": 4266
|
| 29904 |
+
},
|
| 29905 |
+
{
|
| 29906 |
+
"epoch": 0.9482222222222222,
|
| 29907 |
+
"grad_norm": 1.27622389793396,
|
| 29908 |
+
"learning_rate": 1.0378619153674832e-05,
|
| 29909 |
+
"loss": 1.8587,
|
| 29910 |
+
"step": 4267
|
| 29911 |
+
},
|
| 29912 |
+
{
|
| 29913 |
+
"epoch": 0.9484444444444444,
|
| 29914 |
+
"grad_norm": 0.9670615792274475,
|
| 29915 |
+
"learning_rate": 1.0334075723830736e-05,
|
| 29916 |
+
"loss": 1.6965,
|
| 29917 |
+
"step": 4268
|
| 29918 |
+
},
|
| 29919 |
+
{
|
| 29920 |
+
"epoch": 0.9486666666666667,
|
| 29921 |
+
"grad_norm": 0.07310232520103455,
|
| 29922 |
+
"learning_rate": 1.0289532293986638e-05,
|
| 29923 |
+
"loss": 0.0157,
|
| 29924 |
+
"step": 4269
|
| 29925 |
+
},
|
| 29926 |
+
{
|
| 29927 |
+
"epoch": 0.9488888888888889,
|
| 29928 |
+
"grad_norm": 0.9369098544120789,
|
| 29929 |
+
"learning_rate": 1.024498886414254e-05,
|
| 29930 |
+
"loss": 1.7478,
|
| 29931 |
+
"step": 4270
|
| 29932 |
+
},
|
| 29933 |
+
{
|
| 29934 |
+
"epoch": 0.9491111111111111,
|
| 29935 |
+
"grad_norm": 0.7294790148735046,
|
| 29936 |
+
"learning_rate": 1.0200445434298442e-05,
|
| 29937 |
+
"loss": 1.1187,
|
| 29938 |
+
"step": 4271
|
| 29939 |
+
},
|
| 29940 |
+
{
|
| 29941 |
+
"epoch": 0.9493333333333334,
|
| 29942 |
+
"grad_norm": 0.9501145482063293,
|
| 29943 |
+
"learning_rate": 1.0155902004454342e-05,
|
| 29944 |
+
"loss": 1.6452,
|
| 29945 |
+
"step": 4272
|
| 29946 |
+
},
|
| 29947 |
+
{
|
| 29948 |
+
"epoch": 0.9495555555555556,
|
| 29949 |
+
"grad_norm": 1.1477452516555786,
|
| 29950 |
+
"learning_rate": 1.0111358574610246e-05,
|
| 29951 |
+
"loss": 1.8428,
|
| 29952 |
+
"step": 4273
|
| 29953 |
+
},
|
| 29954 |
+
{
|
| 29955 |
+
"epoch": 0.9497777777777778,
|
| 29956 |
+
"grad_norm": 0.9523744583129883,
|
| 29957 |
+
"learning_rate": 1.0066815144766148e-05,
|
| 29958 |
+
"loss": 1.8126,
|
| 29959 |
+
"step": 4274
|
| 29960 |
+
},
|
| 29961 |
+
{
|
| 29962 |
+
"epoch": 0.95,
|
| 29963 |
+
"grad_norm": 0.9534154534339905,
|
| 29964 |
+
"learning_rate": 1.002227171492205e-05,
|
| 29965 |
+
"loss": 1.7667,
|
| 29966 |
+
"step": 4275
|
| 29967 |
+
},
|
| 29968 |
+
{
|
| 29969 |
+
"epoch": 0.9502222222222222,
|
| 29970 |
+
"grad_norm": 1.2276133298873901,
|
| 29971 |
+
"learning_rate": 9.977728285077952e-06,
|
| 29972 |
+
"loss": 1.901,
|
| 29973 |
+
"step": 4276
|
| 29974 |
+
},
|
| 29975 |
+
{
|
| 29976 |
+
"epoch": 0.9504444444444444,
|
| 29977 |
+
"grad_norm": 0.7759425044059753,
|
| 29978 |
+
"learning_rate": 9.933184855233854e-06,
|
| 29979 |
+
"loss": 0.7691,
|
| 29980 |
+
"step": 4277
|
| 29981 |
+
},
|
| 29982 |
+
{
|
| 29983 |
+
"epoch": 0.9506666666666667,
|
| 29984 |
+
"grad_norm": 0.06685524433851242,
|
| 29985 |
+
"learning_rate": 9.888641425389756e-06,
|
| 29986 |
+
"loss": 0.0175,
|
| 29987 |
+
"step": 4278
|
| 29988 |
+
},
|
| 29989 |
+
{
|
| 29990 |
+
"epoch": 0.9508888888888889,
|
| 29991 |
+
"grad_norm": 0.7249411940574646,
|
| 29992 |
+
"learning_rate": 9.844097995545658e-06,
|
| 29993 |
+
"loss": 0.983,
|
| 29994 |
+
"step": 4279
|
| 29995 |
+
},
|
| 29996 |
+
{
|
| 29997 |
+
"epoch": 0.9511111111111111,
|
| 29998 |
+
"grad_norm": 0.6953089237213135,
|
| 29999 |
+
"learning_rate": 9.79955456570156e-06,
|
| 30000 |
+
"loss": 0.9221,
|
| 30001 |
+
"step": 4280
|
| 30002 |
+
},
|
| 30003 |
+
{
|
| 30004 |
+
"epoch": 0.9513333333333334,
|
| 30005 |
+
"grad_norm": 1.0122225284576416,
|
| 30006 |
+
"learning_rate": 9.755011135857462e-06,
|
| 30007 |
+
"loss": 1.6027,
|
| 30008 |
+
"step": 4281
|
| 30009 |
+
},
|
| 30010 |
+
{
|
| 30011 |
+
"epoch": 0.9515555555555556,
|
| 30012 |
+
"grad_norm": 1.0647549629211426,
|
| 30013 |
+
"learning_rate": 9.710467706013364e-06,
|
| 30014 |
+
"loss": 1.6956,
|
| 30015 |
+
"step": 4282
|
| 30016 |
+
},
|
| 30017 |
+
{
|
| 30018 |
+
"epoch": 0.9517777777777777,
|
| 30019 |
+
"grad_norm": 1.0427286624908447,
|
| 30020 |
+
"learning_rate": 9.665924276169266e-06,
|
| 30021 |
+
"loss": 1.6593,
|
| 30022 |
+
"step": 4283
|
| 30023 |
+
},
|
| 30024 |
+
{
|
| 30025 |
+
"epoch": 0.952,
|
| 30026 |
+
"grad_norm": 0.07722879201173782,
|
| 30027 |
+
"learning_rate": 9.621380846325168e-06,
|
| 30028 |
+
"loss": 0.0186,
|
| 30029 |
+
"step": 4284
|
| 30030 |
+
},
|
| 30031 |
+
{
|
| 30032 |
+
"epoch": 0.9522222222222222,
|
| 30033 |
+
"grad_norm": 0.7623983025550842,
|
| 30034 |
+
"learning_rate": 9.57683741648107e-06,
|
| 30035 |
+
"loss": 0.9881,
|
| 30036 |
+
"step": 4285
|
| 30037 |
+
},
|
| 30038 |
+
{
|
| 30039 |
+
"epoch": 0.9524444444444444,
|
| 30040 |
+
"grad_norm": 1.0861274003982544,
|
| 30041 |
+
"learning_rate": 9.532293986636972e-06,
|
| 30042 |
+
"loss": 1.5431,
|
| 30043 |
+
"step": 4286
|
| 30044 |
+
},
|
| 30045 |
+
{
|
| 30046 |
+
"epoch": 0.9526666666666667,
|
| 30047 |
+
"grad_norm": 0.9466423988342285,
|
| 30048 |
+
"learning_rate": 9.487750556792873e-06,
|
| 30049 |
+
"loss": 1.5866,
|
| 30050 |
+
"step": 4287
|
| 30051 |
+
},
|
| 30052 |
+
{
|
| 30053 |
+
"epoch": 0.9528888888888889,
|
| 30054 |
+
"grad_norm": 0.7006486058235168,
|
| 30055 |
+
"learning_rate": 9.443207126948775e-06,
|
| 30056 |
+
"loss": 0.8612,
|
| 30057 |
+
"step": 4288
|
| 30058 |
+
},
|
| 30059 |
+
{
|
| 30060 |
+
"epoch": 0.9531111111111111,
|
| 30061 |
+
"grad_norm": 0.9445701241493225,
|
| 30062 |
+
"learning_rate": 9.398663697104677e-06,
|
| 30063 |
+
"loss": 1.4541,
|
| 30064 |
+
"step": 4289
|
| 30065 |
+
},
|
| 30066 |
+
{
|
| 30067 |
+
"epoch": 0.9533333333333334,
|
| 30068 |
+
"grad_norm": 1.2460566759109497,
|
| 30069 |
+
"learning_rate": 9.35412026726058e-06,
|
| 30070 |
+
"loss": 1.6395,
|
| 30071 |
+
"step": 4290
|
| 30072 |
+
},
|
| 30073 |
+
{
|
| 30074 |
+
"epoch": 0.9535555555555556,
|
| 30075 |
+
"grad_norm": 1.0960421562194824,
|
| 30076 |
+
"learning_rate": 9.309576837416481e-06,
|
| 30077 |
+
"loss": 1.3166,
|
| 30078 |
+
"step": 4291
|
| 30079 |
+
},
|
| 30080 |
+
{
|
| 30081 |
+
"epoch": 0.9537777777777777,
|
| 30082 |
+
"grad_norm": 1.22000253200531,
|
| 30083 |
+
"learning_rate": 9.265033407572383e-06,
|
| 30084 |
+
"loss": 1.4362,
|
| 30085 |
+
"step": 4292
|
| 30086 |
+
},
|
| 30087 |
+
{
|
| 30088 |
+
"epoch": 0.954,
|
| 30089 |
+
"grad_norm": 1.0577735900878906,
|
| 30090 |
+
"learning_rate": 9.220489977728285e-06,
|
| 30091 |
+
"loss": 1.2362,
|
| 30092 |
+
"step": 4293
|
| 30093 |
+
},
|
| 30094 |
+
{
|
| 30095 |
+
"epoch": 0.9542222222222222,
|
| 30096 |
+
"grad_norm": 1.261118769645691,
|
| 30097 |
+
"learning_rate": 9.175946547884187e-06,
|
| 30098 |
+
"loss": 1.5433,
|
| 30099 |
+
"step": 4294
|
| 30100 |
+
},
|
| 30101 |
+
{
|
| 30102 |
+
"epoch": 0.9544444444444444,
|
| 30103 |
+
"grad_norm": 1.0835603475570679,
|
| 30104 |
+
"learning_rate": 9.13140311804009e-06,
|
| 30105 |
+
"loss": 1.0706,
|
| 30106 |
+
"step": 4295
|
| 30107 |
+
},
|
| 30108 |
+
{
|
| 30109 |
+
"epoch": 0.9546666666666667,
|
| 30110 |
+
"grad_norm": 0.1414426565170288,
|
| 30111 |
+
"learning_rate": 9.086859688195993e-06,
|
| 30112 |
+
"loss": 0.0324,
|
| 30113 |
+
"step": 4296
|
| 30114 |
+
},
|
| 30115 |
+
{
|
| 30116 |
+
"epoch": 0.9548888888888889,
|
| 30117 |
+
"grad_norm": 0.15208975970745087,
|
| 30118 |
+
"learning_rate": 9.042316258351893e-06,
|
| 30119 |
+
"loss": 0.0327,
|
| 30120 |
+
"step": 4297
|
| 30121 |
+
},
|
| 30122 |
+
{
|
| 30123 |
+
"epoch": 0.9551111111111111,
|
| 30124 |
+
"grad_norm": 0.942937970161438,
|
| 30125 |
+
"learning_rate": 8.997772828507795e-06,
|
| 30126 |
+
"loss": 0.9045,
|
| 30127 |
+
"step": 4298
|
| 30128 |
+
},
|
| 30129 |
+
{
|
| 30130 |
+
"epoch": 0.9553333333333334,
|
| 30131 |
+
"grad_norm": 0.9884275197982788,
|
| 30132 |
+
"learning_rate": 8.953229398663697e-06,
|
| 30133 |
+
"loss": 0.9895,
|
| 30134 |
+
"step": 4299
|
| 30135 |
+
},
|
| 30136 |
+
{
|
| 30137 |
+
"epoch": 0.9555555555555556,
|
| 30138 |
+
"grad_norm": 0.936667263507843,
|
| 30139 |
+
"learning_rate": 8.908685968819599e-06,
|
| 30140 |
+
"loss": 0.776,
|
| 30141 |
+
"step": 4300
|
| 30142 |
+
},
|
| 30143 |
+
{
|
| 30144 |
+
"epoch": 0.9557777777777777,
|
| 30145 |
+
"grad_norm": 0.6337212324142456,
|
| 30146 |
+
"learning_rate": 8.864142538975503e-06,
|
| 30147 |
+
"loss": 1.1042,
|
| 30148 |
+
"step": 4301
|
| 30149 |
+
},
|
| 30150 |
+
{
|
| 30151 |
+
"epoch": 0.956,
|
| 30152 |
+
"grad_norm": 0.8414755463600159,
|
| 30153 |
+
"learning_rate": 8.819599109131403e-06,
|
| 30154 |
+
"loss": 1.9703,
|
| 30155 |
+
"step": 4302
|
| 30156 |
+
},
|
| 30157 |
+
{
|
| 30158 |
+
"epoch": 0.9562222222222222,
|
| 30159 |
+
"grad_norm": 0.5726562738418579,
|
| 30160 |
+
"learning_rate": 8.775055679287305e-06,
|
| 30161 |
+
"loss": 0.8849,
|
| 30162 |
+
"step": 4303
|
| 30163 |
+
},
|
| 30164 |
+
{
|
| 30165 |
+
"epoch": 0.9564444444444444,
|
| 30166 |
+
"grad_norm": 0.6391728520393372,
|
| 30167 |
+
"learning_rate": 8.730512249443209e-06,
|
| 30168 |
+
"loss": 1.1577,
|
| 30169 |
+
"step": 4304
|
| 30170 |
+
},
|
| 30171 |
+
{
|
| 30172 |
+
"epoch": 0.9566666666666667,
|
| 30173 |
+
"grad_norm": 0.5593711137771606,
|
| 30174 |
+
"learning_rate": 8.685968819599109e-06,
|
| 30175 |
+
"loss": 0.8922,
|
| 30176 |
+
"step": 4305
|
| 30177 |
+
},
|
| 30178 |
+
{
|
| 30179 |
+
"epoch": 0.9568888888888889,
|
| 30180 |
+
"grad_norm": 0.7331346869468689,
|
| 30181 |
+
"learning_rate": 8.641425389755013e-06,
|
| 30182 |
+
"loss": 1.1386,
|
| 30183 |
+
"step": 4306
|
| 30184 |
+
},
|
| 30185 |
+
{
|
| 30186 |
+
"epoch": 0.9571111111111111,
|
| 30187 |
+
"grad_norm": 0.5739585757255554,
|
| 30188 |
+
"learning_rate": 8.596881959910913e-06,
|
| 30189 |
+
"loss": 0.8375,
|
| 30190 |
+
"step": 4307
|
| 30191 |
+
},
|
| 30192 |
+
{
|
| 30193 |
+
"epoch": 0.9573333333333334,
|
| 30194 |
+
"grad_norm": 0.07684678584337234,
|
| 30195 |
+
"learning_rate": 8.552338530066815e-06,
|
| 30196 |
+
"loss": 0.011,
|
| 30197 |
+
"step": 4308
|
| 30198 |
+
},
|
| 30199 |
+
{
|
| 30200 |
+
"epoch": 0.9575555555555556,
|
| 30201 |
+
"grad_norm": 0.07993515580892563,
|
| 30202 |
+
"learning_rate": 8.507795100222719e-06,
|
| 30203 |
+
"loss": 0.0111,
|
| 30204 |
+
"step": 4309
|
| 30205 |
+
},
|
| 30206 |
+
{
|
| 30207 |
+
"epoch": 0.9577777777777777,
|
| 30208 |
+
"grad_norm": 0.6196415424346924,
|
| 30209 |
+
"learning_rate": 8.463251670378619e-06,
|
| 30210 |
+
"loss": 1.0166,
|
| 30211 |
+
"step": 4310
|
| 30212 |
+
},
|
| 30213 |
+
{
|
| 30214 |
+
"epoch": 0.958,
|
| 30215 |
+
"grad_norm": 0.8932988047599792,
|
| 30216 |
+
"learning_rate": 8.41870824053452e-06,
|
| 30217 |
+
"loss": 2.3441,
|
| 30218 |
+
"step": 4311
|
| 30219 |
+
},
|
| 30220 |
+
{
|
| 30221 |
+
"epoch": 0.9582222222222222,
|
| 30222 |
+
"grad_norm": 1.2204405069351196,
|
| 30223 |
+
"learning_rate": 8.374164810690424e-06,
|
| 30224 |
+
"loss": 2.0743,
|
| 30225 |
+
"step": 4312
|
| 30226 |
+
},
|
| 30227 |
+
{
|
| 30228 |
+
"epoch": 0.9584444444444444,
|
| 30229 |
+
"grad_norm": 0.9031944870948792,
|
| 30230 |
+
"learning_rate": 8.329621380846325e-06,
|
| 30231 |
+
"loss": 1.9911,
|
| 30232 |
+
"step": 4313
|
| 30233 |
+
},
|
| 30234 |
+
{
|
| 30235 |
+
"epoch": 0.9586666666666667,
|
| 30236 |
+
"grad_norm": 1.0615592002868652,
|
| 30237 |
+
"learning_rate": 8.285077951002228e-06,
|
| 30238 |
+
"loss": 2.3094,
|
| 30239 |
+
"step": 4314
|
| 30240 |
+
},
|
| 30241 |
+
{
|
| 30242 |
+
"epoch": 0.9588888888888889,
|
| 30243 |
+
"grad_norm": 0.931339681148529,
|
| 30244 |
+
"learning_rate": 8.240534521158129e-06,
|
| 30245 |
+
"loss": 1.8745,
|
| 30246 |
+
"step": 4315
|
| 30247 |
+
},
|
| 30248 |
+
{
|
| 30249 |
+
"epoch": 0.9591111111111111,
|
| 30250 |
+
"grad_norm": 1.068681240081787,
|
| 30251 |
+
"learning_rate": 8.19599109131403e-06,
|
| 30252 |
+
"loss": 2.1088,
|
| 30253 |
+
"step": 4316
|
| 30254 |
+
},
|
| 30255 |
+
{
|
| 30256 |
+
"epoch": 0.9593333333333334,
|
| 30257 |
+
"grad_norm": 0.9199005365371704,
|
| 30258 |
+
"learning_rate": 8.151447661469934e-06,
|
| 30259 |
+
"loss": 1.9663,
|
| 30260 |
+
"step": 4317
|
| 30261 |
+
},
|
| 30262 |
+
{
|
| 30263 |
+
"epoch": 0.9595555555555556,
|
| 30264 |
+
"grad_norm": 0.9643719792366028,
|
| 30265 |
+
"learning_rate": 8.106904231625835e-06,
|
| 30266 |
+
"loss": 2.2083,
|
| 30267 |
+
"step": 4318
|
| 30268 |
+
},
|
| 30269 |
+
{
|
| 30270 |
+
"epoch": 0.9597777777777777,
|
| 30271 |
+
"grad_norm": 1.026021122932434,
|
| 30272 |
+
"learning_rate": 8.062360801781738e-06,
|
| 30273 |
+
"loss": 1.688,
|
| 30274 |
+
"step": 4319
|
| 30275 |
+
},
|
| 30276 |
+
{
|
| 30277 |
+
"epoch": 0.96,
|
| 30278 |
+
"grad_norm": 0.9671936631202698,
|
| 30279 |
+
"learning_rate": 8.01781737193764e-06,
|
| 30280 |
+
"loss": 2.0219,
|
| 30281 |
+
"step": 4320
|
| 30282 |
+
},
|
| 30283 |
+
{
|
| 30284 |
+
"epoch": 0.9602222222222222,
|
| 30285 |
+
"grad_norm": 0.9827919006347656,
|
| 30286 |
+
"learning_rate": 7.97327394209354e-06,
|
| 30287 |
+
"loss": 1.6401,
|
| 30288 |
+
"step": 4321
|
| 30289 |
+
},
|
| 30290 |
+
{
|
| 30291 |
+
"epoch": 0.9604444444444444,
|
| 30292 |
+
"grad_norm": 0.07093919813632965,
|
| 30293 |
+
"learning_rate": 7.928730512249444e-06,
|
| 30294 |
+
"loss": 0.0157,
|
| 30295 |
+
"step": 4322
|
| 30296 |
+
},
|
| 30297 |
+
{
|
| 30298 |
+
"epoch": 0.9606666666666667,
|
| 30299 |
+
"grad_norm": 0.07090691477060318,
|
| 30300 |
+
"learning_rate": 7.884187082405346e-06,
|
| 30301 |
+
"loss": 0.0154,
|
| 30302 |
+
"step": 4323
|
| 30303 |
+
},
|
| 30304 |
+
{
|
| 30305 |
+
"epoch": 0.9608888888888889,
|
| 30306 |
+
"grad_norm": 0.06990091502666473,
|
| 30307 |
+
"learning_rate": 7.839643652561248e-06,
|
| 30308 |
+
"loss": 0.0153,
|
| 30309 |
+
"step": 4324
|
| 30310 |
+
},
|
| 30311 |
+
{
|
| 30312 |
+
"epoch": 0.9611111111111111,
|
| 30313 |
+
"grad_norm": 0.06884946674108505,
|
| 30314 |
+
"learning_rate": 7.79510022271715e-06,
|
| 30315 |
+
"loss": 0.0152,
|
| 30316 |
+
"step": 4325
|
| 30317 |
+
},
|
| 30318 |
+
{
|
| 30319 |
+
"epoch": 0.9613333333333334,
|
| 30320 |
+
"grad_norm": 0.8022354245185852,
|
| 30321 |
+
"learning_rate": 7.75055679287305e-06,
|
| 30322 |
+
"loss": 0.8116,
|
| 30323 |
+
"step": 4326
|
| 30324 |
+
},
|
| 30325 |
+
{
|
| 30326 |
+
"epoch": 0.9615555555555556,
|
| 30327 |
+
"grad_norm": 1.062786340713501,
|
| 30328 |
+
"learning_rate": 7.706013363028954e-06,
|
| 30329 |
+
"loss": 1.7179,
|
| 30330 |
+
"step": 4327
|
| 30331 |
+
},
|
| 30332 |
+
{
|
| 30333 |
+
"epoch": 0.9617777777777777,
|
| 30334 |
+
"grad_norm": 1.0791099071502686,
|
| 30335 |
+
"learning_rate": 7.661469933184856e-06,
|
| 30336 |
+
"loss": 1.7474,
|
| 30337 |
+
"step": 4328
|
| 30338 |
+
},
|
| 30339 |
+
{
|
| 30340 |
+
"epoch": 0.962,
|
| 30341 |
+
"grad_norm": 1.0296615362167358,
|
| 30342 |
+
"learning_rate": 7.616926503340757e-06,
|
| 30343 |
+
"loss": 1.6771,
|
| 30344 |
+
"step": 4329
|
| 30345 |
+
},
|
| 30346 |
+
{
|
| 30347 |
+
"epoch": 0.9622222222222222,
|
| 30348 |
+
"grad_norm": 1.076456904411316,
|
| 30349 |
+
"learning_rate": 7.57238307349666e-06,
|
| 30350 |
+
"loss": 1.8226,
|
| 30351 |
+
"step": 4330
|
| 30352 |
+
},
|
| 30353 |
+
{
|
| 30354 |
+
"epoch": 0.9624444444444444,
|
| 30355 |
+
"grad_norm": 0.9866617321968079,
|
| 30356 |
+
"learning_rate": 7.527839643652562e-06,
|
| 30357 |
+
"loss": 1.887,
|
| 30358 |
+
"step": 4331
|
| 30359 |
+
},
|
| 30360 |
+
{
|
| 30361 |
+
"epoch": 0.9626666666666667,
|
| 30362 |
+
"grad_norm": 0.06448253244161606,
|
| 30363 |
+
"learning_rate": 7.483296213808463e-06,
|
| 30364 |
+
"loss": 0.0173,
|
| 30365 |
+
"step": 4332
|
| 30366 |
+
},
|
| 30367 |
+
{
|
| 30368 |
+
"epoch": 0.9628888888888889,
|
| 30369 |
+
"grad_norm": 0.06699193269014359,
|
| 30370 |
+
"learning_rate": 7.438752783964366e-06,
|
| 30371 |
+
"loss": 0.0175,
|
| 30372 |
+
"step": 4333
|
| 30373 |
+
},
|
| 30374 |
+
{
|
| 30375 |
+
"epoch": 0.9631111111111111,
|
| 30376 |
+
"grad_norm": 0.7605263590812683,
|
| 30377 |
+
"learning_rate": 7.394209354120267e-06,
|
| 30378 |
+
"loss": 0.8152,
|
| 30379 |
+
"step": 4334
|
| 30380 |
+
},
|
| 30381 |
+
{
|
| 30382 |
+
"epoch": 0.9633333333333334,
|
| 30383 |
+
"grad_norm": 1.077652096748352,
|
| 30384 |
+
"learning_rate": 7.34966592427617e-06,
|
| 30385 |
+
"loss": 1.8169,
|
| 30386 |
+
"step": 4335
|
| 30387 |
+
},
|
| 30388 |
+
{
|
| 30389 |
+
"epoch": 0.9635555555555556,
|
| 30390 |
+
"grad_norm": 0.07532133162021637,
|
| 30391 |
+
"learning_rate": 7.305122494432072e-06,
|
| 30392 |
+
"loss": 0.0182,
|
| 30393 |
+
"step": 4336
|
| 30394 |
+
},
|
| 30395 |
+
{
|
| 30396 |
+
"epoch": 0.9637777777777777,
|
| 30397 |
+
"grad_norm": 0.07628657668828964,
|
| 30398 |
+
"learning_rate": 7.260579064587973e-06,
|
| 30399 |
+
"loss": 0.0177,
|
| 30400 |
+
"step": 4337
|
| 30401 |
+
},
|
| 30402 |
+
{
|
| 30403 |
+
"epoch": 0.964,
|
| 30404 |
+
"grad_norm": 0.07550018280744553,
|
| 30405 |
+
"learning_rate": 7.216035634743876e-06,
|
| 30406 |
+
"loss": 0.0181,
|
| 30407 |
+
"step": 4338
|
| 30408 |
+
},
|
| 30409 |
+
{
|
| 30410 |
+
"epoch": 0.9642222222222222,
|
| 30411 |
+
"grad_norm": 0.9071139097213745,
|
| 30412 |
+
"learning_rate": 7.1714922048997785e-06,
|
| 30413 |
+
"loss": 0.8412,
|
| 30414 |
+
"step": 4339
|
| 30415 |
+
},
|
| 30416 |
+
{
|
| 30417 |
+
"epoch": 0.9644444444444444,
|
| 30418 |
+
"grad_norm": 0.9595382213592529,
|
| 30419 |
+
"learning_rate": 7.12694877505568e-06,
|
| 30420 |
+
"loss": 1.4946,
|
| 30421 |
+
"step": 4340
|
| 30422 |
+
},
|
| 30423 |
+
{
|
| 30424 |
+
"epoch": 0.9646666666666667,
|
| 30425 |
+
"grad_norm": 1.0608892440795898,
|
| 30426 |
+
"learning_rate": 7.082405345211582e-06,
|
| 30427 |
+
"loss": 1.6542,
|
| 30428 |
+
"step": 4341
|
| 30429 |
+
},
|
| 30430 |
+
{
|
| 30431 |
+
"epoch": 0.9648888888888889,
|
| 30432 |
+
"grad_norm": 1.1697642803192139,
|
| 30433 |
+
"learning_rate": 7.037861915367483e-06,
|
| 30434 |
+
"loss": 1.7757,
|
| 30435 |
+
"step": 4342
|
| 30436 |
+
},
|
| 30437 |
+
{
|
| 30438 |
+
"epoch": 0.9651111111111111,
|
| 30439 |
+
"grad_norm": 1.0154902935028076,
|
| 30440 |
+
"learning_rate": 6.9933184855233855e-06,
|
| 30441 |
+
"loss": 1.2555,
|
| 30442 |
+
"step": 4343
|
| 30443 |
+
},
|
| 30444 |
+
{
|
| 30445 |
+
"epoch": 0.9653333333333334,
|
| 30446 |
+
"grad_norm": 1.1377027034759521,
|
| 30447 |
+
"learning_rate": 6.948775055679288e-06,
|
| 30448 |
+
"loss": 1.7557,
|
| 30449 |
+
"step": 4344
|
| 30450 |
+
},
|
| 30451 |
+
{
|
| 30452 |
+
"epoch": 0.9655555555555555,
|
| 30453 |
+
"grad_norm": 1.1074367761611938,
|
| 30454 |
+
"learning_rate": 6.9042316258351895e-06,
|
| 30455 |
+
"loss": 1.5758,
|
| 30456 |
+
"step": 4345
|
| 30457 |
+
},
|
| 30458 |
+
{
|
| 30459 |
+
"epoch": 0.9657777777777777,
|
| 30460 |
+
"grad_norm": 0.6881236433982849,
|
| 30461 |
+
"learning_rate": 6.8596881959910914e-06,
|
| 30462 |
+
"loss": 0.7293,
|
| 30463 |
+
"step": 4346
|
| 30464 |
+
},
|
| 30465 |
+
{
|
| 30466 |
+
"epoch": 0.966,
|
| 30467 |
+
"grad_norm": 0.18515631556510925,
|
| 30468 |
+
"learning_rate": 6.815144766146994e-06,
|
| 30469 |
+
"loss": 0.0304,
|
| 30470 |
+
"step": 4347
|
| 30471 |
+
},
|
| 30472 |
+
{
|
| 30473 |
+
"epoch": 0.9662222222222222,
|
| 30474 |
+
"grad_norm": 1.1112456321716309,
|
| 30475 |
+
"learning_rate": 6.770601336302895e-06,
|
| 30476 |
+
"loss": 1.2363,
|
| 30477 |
+
"step": 4348
|
| 30478 |
+
},
|
| 30479 |
+
{
|
| 30480 |
+
"epoch": 0.9664444444444444,
|
| 30481 |
+
"grad_norm": 0.8793594837188721,
|
| 30482 |
+
"learning_rate": 6.726057906458798e-06,
|
| 30483 |
+
"loss": 0.8094,
|
| 30484 |
+
"step": 4349
|
| 30485 |
+
},
|
| 30486 |
+
{
|
| 30487 |
+
"epoch": 0.9666666666666667,
|
| 30488 |
+
"grad_norm": 0.9809714555740356,
|
| 30489 |
+
"learning_rate": 6.6815144766147e-06,
|
| 30490 |
+
"loss": 0.7478,
|
| 30491 |
+
"step": 4350
|
| 30492 |
+
},
|
| 30493 |
+
{
|
| 30494 |
+
"epoch": 0.9668888888888889,
|
| 30495 |
+
"grad_norm": 0.6749547123908997,
|
| 30496 |
+
"learning_rate": 6.636971046770601e-06,
|
| 30497 |
+
"loss": 1.0962,
|
| 30498 |
+
"step": 4351
|
| 30499 |
+
},
|
| 30500 |
+
{
|
| 30501 |
+
"epoch": 0.9671111111111111,
|
| 30502 |
+
"grad_norm": 0.8518489599227905,
|
| 30503 |
+
"learning_rate": 6.592427616926504e-06,
|
| 30504 |
+
"loss": 2.0981,
|
| 30505 |
+
"step": 4352
|
| 30506 |
+
},
|
| 30507 |
+
{
|
| 30508 |
+
"epoch": 0.9673333333333334,
|
| 30509 |
+
"grad_norm": 0.5904682874679565,
|
| 30510 |
+
"learning_rate": 6.547884187082405e-06,
|
| 30511 |
+
"loss": 0.8928,
|
| 30512 |
+
"step": 4353
|
| 30513 |
+
},
|
| 30514 |
+
{
|
| 30515 |
+
"epoch": 0.9675555555555555,
|
| 30516 |
+
"grad_norm": 0.6970412731170654,
|
| 30517 |
+
"learning_rate": 6.503340757238307e-06,
|
| 30518 |
+
"loss": 1.1904,
|
| 30519 |
+
"step": 4354
|
| 30520 |
+
},
|
| 30521 |
+
{
|
| 30522 |
+
"epoch": 0.9677777777777777,
|
| 30523 |
+
"grad_norm": 0.8849335312843323,
|
| 30524 |
+
"learning_rate": 6.45879732739421e-06,
|
| 30525 |
+
"loss": 1.9197,
|
| 30526 |
+
"step": 4355
|
| 30527 |
+
},
|
| 30528 |
+
{
|
| 30529 |
+
"epoch": 0.968,
|
| 30530 |
+
"grad_norm": 0.7975565791130066,
|
| 30531 |
+
"learning_rate": 6.414253897550111e-06,
|
| 30532 |
+
"loss": 2.0422,
|
| 30533 |
+
"step": 4356
|
| 30534 |
+
},
|
| 30535 |
+
{
|
| 30536 |
+
"epoch": 0.9682222222222222,
|
| 30537 |
+
"grad_norm": 0.898343026638031,
|
| 30538 |
+
"learning_rate": 6.369710467706014e-06,
|
| 30539 |
+
"loss": 2.0284,
|
| 30540 |
+
"step": 4357
|
| 30541 |
+
},
|
| 30542 |
+
{
|
| 30543 |
+
"epoch": 0.9684444444444444,
|
| 30544 |
+
"grad_norm": 0.08013878762722015,
|
| 30545 |
+
"learning_rate": 6.325167037861916e-06,
|
| 30546 |
+
"loss": 0.011,
|
| 30547 |
+
"step": 4358
|
| 30548 |
+
},
|
| 30549 |
+
{
|
| 30550 |
+
"epoch": 0.9686666666666667,
|
| 30551 |
+
"grad_norm": 0.0778137668967247,
|
| 30552 |
+
"learning_rate": 6.280623608017817e-06,
|
| 30553 |
+
"loss": 0.011,
|
| 30554 |
+
"step": 4359
|
| 30555 |
+
},
|
| 30556 |
+
{
|
| 30557 |
+
"epoch": 0.9688888888888889,
|
| 30558 |
+
"grad_norm": 0.075262151658535,
|
| 30559 |
+
"learning_rate": 6.23608017817372e-06,
|
| 30560 |
+
"loss": 0.0111,
|
| 30561 |
+
"step": 4360
|
| 30562 |
+
},
|
| 30563 |
+
{
|
| 30564 |
+
"epoch": 0.9691111111111111,
|
| 30565 |
+
"grad_norm": 0.8569869995117188,
|
| 30566 |
+
"learning_rate": 6.191536748329622e-06,
|
| 30567 |
+
"loss": 2.1552,
|
| 30568 |
+
"step": 4361
|
| 30569 |
+
},
|
| 30570 |
+
{
|
| 30571 |
+
"epoch": 0.9693333333333334,
|
| 30572 |
+
"grad_norm": 0.7958348393440247,
|
| 30573 |
+
"learning_rate": 6.146993318485524e-06,
|
| 30574 |
+
"loss": 2.1074,
|
| 30575 |
+
"step": 4362
|
| 30576 |
+
},
|
| 30577 |
+
{
|
| 30578 |
+
"epoch": 0.9695555555555555,
|
| 30579 |
+
"grad_norm": 0.861589789390564,
|
| 30580 |
+
"learning_rate": 6.102449888641426e-06,
|
| 30581 |
+
"loss": 1.8603,
|
| 30582 |
+
"step": 4363
|
| 30583 |
+
},
|
| 30584 |
+
{
|
| 30585 |
+
"epoch": 0.9697777777777777,
|
| 30586 |
+
"grad_norm": 0.870924174785614,
|
| 30587 |
+
"learning_rate": 6.057906458797328e-06,
|
| 30588 |
+
"loss": 1.6482,
|
| 30589 |
+
"step": 4364
|
| 30590 |
+
},
|
| 30591 |
+
{
|
| 30592 |
+
"epoch": 0.97,
|
| 30593 |
+
"grad_norm": 0.6811211109161377,
|
| 30594 |
+
"learning_rate": 6.01336302895323e-06,
|
| 30595 |
+
"loss": 0.9431,
|
| 30596 |
+
"step": 4365
|
| 30597 |
+
},
|
| 30598 |
+
{
|
| 30599 |
+
"epoch": 0.9702222222222222,
|
| 30600 |
+
"grad_norm": 1.2624512910842896,
|
| 30601 |
+
"learning_rate": 5.968819599109132e-06,
|
| 30602 |
+
"loss": 1.2395,
|
| 30603 |
+
"step": 4366
|
| 30604 |
+
},
|
| 30605 |
+
{
|
| 30606 |
+
"epoch": 0.9704444444444444,
|
| 30607 |
+
"grad_norm": 0.6082854866981506,
|
| 30608 |
+
"learning_rate": 5.924276169265034e-06,
|
| 30609 |
+
"loss": 0.9055,
|
| 30610 |
+
"step": 4367
|
| 30611 |
+
},
|
| 30612 |
+
{
|
| 30613 |
+
"epoch": 0.9706666666666667,
|
| 30614 |
+
"grad_norm": 0.990861713886261,
|
| 30615 |
+
"learning_rate": 5.879732739420936e-06,
|
| 30616 |
+
"loss": 1.96,
|
| 30617 |
+
"step": 4368
|
| 30618 |
+
},
|
| 30619 |
+
{
|
| 30620 |
+
"epoch": 0.9708888888888889,
|
| 30621 |
+
"grad_norm": 0.99041748046875,
|
| 30622 |
+
"learning_rate": 5.835189309576838e-06,
|
| 30623 |
+
"loss": 1.6842,
|
| 30624 |
+
"step": 4369
|
| 30625 |
+
},
|
| 30626 |
+
{
|
| 30627 |
+
"epoch": 0.9711111111111111,
|
| 30628 |
+
"grad_norm": 0.9319810271263123,
|
| 30629 |
+
"learning_rate": 5.79064587973274e-06,
|
| 30630 |
+
"loss": 1.9125,
|
| 30631 |
+
"step": 4370
|
| 30632 |
+
},
|
| 30633 |
+
{
|
| 30634 |
+
"epoch": 0.9713333333333334,
|
| 30635 |
+
"grad_norm": 1.01008141040802,
|
| 30636 |
+
"learning_rate": 5.746102449888642e-06,
|
| 30637 |
+
"loss": 1.9553,
|
| 30638 |
+
"step": 4371
|
| 30639 |
+
},
|
| 30640 |
+
{
|
| 30641 |
+
"epoch": 0.9715555555555555,
|
| 30642 |
+
"grad_norm": 0.7167505025863647,
|
| 30643 |
+
"learning_rate": 5.7015590200445435e-06,
|
| 30644 |
+
"loss": 0.972,
|
| 30645 |
+
"step": 4372
|
| 30646 |
+
},
|
| 30647 |
+
{
|
| 30648 |
+
"epoch": 0.9717777777777777,
|
| 30649 |
+
"grad_norm": 0.07200965285301208,
|
| 30650 |
+
"learning_rate": 5.6570155902004455e-06,
|
| 30651 |
+
"loss": 0.0151,
|
| 30652 |
+
"step": 4373
|
| 30653 |
+
},
|
| 30654 |
+
{
|
| 30655 |
+
"epoch": 0.972,
|
| 30656 |
+
"grad_norm": 0.6893488764762878,
|
| 30657 |
+
"learning_rate": 5.6124721603563475e-06,
|
| 30658 |
+
"loss": 0.8677,
|
| 30659 |
+
"step": 4374
|
| 30660 |
+
},
|
| 30661 |
+
{
|
| 30662 |
+
"epoch": 0.9722222222222222,
|
| 30663 |
+
"grad_norm": 0.12305945158004761,
|
| 30664 |
+
"learning_rate": 5.56792873051225e-06,
|
| 30665 |
+
"loss": 0.0204,
|
| 30666 |
+
"step": 4375
|
| 30667 |
+
},
|
| 30668 |
+
{
|
| 30669 |
+
"epoch": 0.9724444444444444,
|
| 30670 |
+
"grad_norm": 0.9432768821716309,
|
| 30671 |
+
"learning_rate": 5.523385300668152e-06,
|
| 30672 |
+
"loss": 1.7875,
|
| 30673 |
+
"step": 4376
|
| 30674 |
+
},
|
| 30675 |
+
{
|
| 30676 |
+
"epoch": 0.9726666666666667,
|
| 30677 |
+
"grad_norm": 1.0131165981292725,
|
| 30678 |
+
"learning_rate": 5.478841870824053e-06,
|
| 30679 |
+
"loss": 1.6501,
|
| 30680 |
+
"step": 4377
|
| 30681 |
+
},
|
| 30682 |
+
{
|
| 30683 |
+
"epoch": 0.9728888888888889,
|
| 30684 |
+
"grad_norm": 1.0048753023147583,
|
| 30685 |
+
"learning_rate": 5.434298440979955e-06,
|
| 30686 |
+
"loss": 1.6584,
|
| 30687 |
+
"step": 4378
|
| 30688 |
+
},
|
| 30689 |
+
{
|
| 30690 |
+
"epoch": 0.9731111111111111,
|
| 30691 |
+
"grad_norm": 0.9927910566329956,
|
| 30692 |
+
"learning_rate": 5.389755011135858e-06,
|
| 30693 |
+
"loss": 1.9151,
|
| 30694 |
+
"step": 4379
|
| 30695 |
+
},
|
| 30696 |
+
{
|
| 30697 |
+
"epoch": 0.9733333333333334,
|
| 30698 |
+
"grad_norm": 0.7287546396255493,
|
| 30699 |
+
"learning_rate": 5.34521158129176e-06,
|
| 30700 |
+
"loss": 0.7935,
|
| 30701 |
+
"step": 4380
|
| 30702 |
+
},
|
| 30703 |
+
{
|
| 30704 |
+
"epoch": 0.9735555555555555,
|
| 30705 |
+
"grad_norm": 0.06438437104225159,
|
| 30706 |
+
"learning_rate": 5.300668151447661e-06,
|
| 30707 |
+
"loss": 0.0174,
|
| 30708 |
+
"step": 4381
|
| 30709 |
+
},
|
| 30710 |
+
{
|
| 30711 |
+
"epoch": 0.9737777777777777,
|
| 30712 |
+
"grad_norm": 0.7063573002815247,
|
| 30713 |
+
"learning_rate": 5.256124721603563e-06,
|
| 30714 |
+
"loss": 0.794,
|
| 30715 |
+
"step": 4382
|
| 30716 |
+
},
|
| 30717 |
+
{
|
| 30718 |
+
"epoch": 0.974,
|
| 30719 |
+
"grad_norm": 0.07574823498725891,
|
| 30720 |
+
"learning_rate": 5.211581291759466e-06,
|
| 30721 |
+
"loss": 0.0178,
|
| 30722 |
+
"step": 4383
|
| 30723 |
+
},
|
| 30724 |
+
{
|
| 30725 |
+
"epoch": 0.9742222222222222,
|
| 30726 |
+
"grad_norm": 0.0726298987865448,
|
| 30727 |
+
"learning_rate": 5.167037861915368e-06,
|
| 30728 |
+
"loss": 0.0177,
|
| 30729 |
+
"step": 4384
|
| 30730 |
+
},
|
| 30731 |
+
{
|
| 30732 |
+
"epoch": 0.9744444444444444,
|
| 30733 |
+
"grad_norm": 0.7331129312515259,
|
| 30734 |
+
"learning_rate": 5.12249443207127e-06,
|
| 30735 |
+
"loss": 0.7402,
|
| 30736 |
+
"step": 4385
|
| 30737 |
+
},
|
| 30738 |
+
{
|
| 30739 |
+
"epoch": 0.9746666666666667,
|
| 30740 |
+
"grad_norm": 1.0017316341400146,
|
| 30741 |
+
"learning_rate": 5.077951002227171e-06,
|
| 30742 |
+
"loss": 1.6145,
|
| 30743 |
+
"step": 4386
|
| 30744 |
+
},
|
| 30745 |
+
{
|
| 30746 |
+
"epoch": 0.9748888888888889,
|
| 30747 |
+
"grad_norm": 0.9680055379867554,
|
| 30748 |
+
"learning_rate": 5.033407572383074e-06,
|
| 30749 |
+
"loss": 1.4337,
|
| 30750 |
+
"step": 4387
|
| 30751 |
+
},
|
| 30752 |
+
{
|
| 30753 |
+
"epoch": 0.9751111111111112,
|
| 30754 |
+
"grad_norm": 1.0141950845718384,
|
| 30755 |
+
"learning_rate": 4.988864142538976e-06,
|
| 30756 |
+
"loss": 1.549,
|
| 30757 |
+
"step": 4388
|
| 30758 |
+
},
|
| 30759 |
+
{
|
| 30760 |
+
"epoch": 0.9753333333333334,
|
| 30761 |
+
"grad_norm": 0.7489122748374939,
|
| 30762 |
+
"learning_rate": 4.944320712694878e-06,
|
| 30763 |
+
"loss": 0.6078,
|
| 30764 |
+
"step": 4389
|
| 30765 |
+
},
|
| 30766 |
+
{
|
| 30767 |
+
"epoch": 0.9755555555555555,
|
| 30768 |
+
"grad_norm": 0.9311794638633728,
|
| 30769 |
+
"learning_rate": 4.89977728285078e-06,
|
| 30770 |
+
"loss": 0.6076,
|
| 30771 |
+
"step": 4390
|
| 30772 |
+
},
|
| 30773 |
+
{
|
| 30774 |
+
"epoch": 0.9757777777777777,
|
| 30775 |
+
"grad_norm": 1.1416817903518677,
|
| 30776 |
+
"learning_rate": 4.855233853006682e-06,
|
| 30777 |
+
"loss": 1.7661,
|
| 30778 |
+
"step": 4391
|
| 30779 |
+
},
|
| 30780 |
+
{
|
| 30781 |
+
"epoch": 0.976,
|
| 30782 |
+
"grad_norm": 1.3029440641403198,
|
| 30783 |
+
"learning_rate": 4.810690423162584e-06,
|
| 30784 |
+
"loss": 1.5154,
|
| 30785 |
+
"step": 4392
|
| 30786 |
+
},
|
| 30787 |
+
{
|
| 30788 |
+
"epoch": 0.9762222222222222,
|
| 30789 |
+
"grad_norm": 1.1306506395339966,
|
| 30790 |
+
"learning_rate": 4.766146993318486e-06,
|
| 30791 |
+
"loss": 1.382,
|
| 30792 |
+
"step": 4393
|
| 30793 |
+
},
|
| 30794 |
+
{
|
| 30795 |
+
"epoch": 0.9764444444444444,
|
| 30796 |
+
"grad_norm": 0.9537327885627747,
|
| 30797 |
+
"learning_rate": 4.721603563474388e-06,
|
| 30798 |
+
"loss": 1.432,
|
| 30799 |
+
"step": 4394
|
| 30800 |
+
},
|
| 30801 |
+
{
|
| 30802 |
+
"epoch": 0.9766666666666667,
|
| 30803 |
+
"grad_norm": 0.9183233976364136,
|
| 30804 |
+
"learning_rate": 4.67706013363029e-06,
|
| 30805 |
+
"loss": 1.214,
|
| 30806 |
+
"step": 4395
|
| 30807 |
+
},
|
| 30808 |
+
{
|
| 30809 |
+
"epoch": 0.9768888888888889,
|
| 30810 |
+
"grad_norm": 0.9410824775695801,
|
| 30811 |
+
"learning_rate": 4.632516703786192e-06,
|
| 30812 |
+
"loss": 1.1194,
|
| 30813 |
+
"step": 4396
|
| 30814 |
+
},
|
| 30815 |
+
{
|
| 30816 |
+
"epoch": 0.9771111111111112,
|
| 30817 |
+
"grad_norm": 1.0940196514129639,
|
| 30818 |
+
"learning_rate": 4.587973273942094e-06,
|
| 30819 |
+
"loss": 1.1115,
|
| 30820 |
+
"step": 4397
|
| 30821 |
+
},
|
| 30822 |
+
{
|
| 30823 |
+
"epoch": 0.9773333333333334,
|
| 30824 |
+
"grad_norm": 1.0624735355377197,
|
| 30825 |
+
"learning_rate": 4.5434298440979965e-06,
|
| 30826 |
+
"loss": 1.0935,
|
| 30827 |
+
"step": 4398
|
| 30828 |
+
},
|
| 30829 |
+
{
|
| 30830 |
+
"epoch": 0.9775555555555555,
|
| 30831 |
+
"grad_norm": 1.0532846450805664,
|
| 30832 |
+
"learning_rate": 4.498886414253898e-06,
|
| 30833 |
+
"loss": 1.1252,
|
| 30834 |
+
"step": 4399
|
| 30835 |
+
},
|
| 30836 |
+
{
|
| 30837 |
+
"epoch": 0.9777777777777777,
|
| 30838 |
+
"grad_norm": 1.0047916173934937,
|
| 30839 |
+
"learning_rate": 4.4543429844097995e-06,
|
| 30840 |
+
"loss": 0.6994,
|
| 30841 |
+
"step": 4400
|
| 30842 |
+
},
|
| 30843 |
+
{
|
| 30844 |
+
"epoch": 0.978,
|
| 30845 |
+
"grad_norm": 0.7360401153564453,
|
| 30846 |
+
"learning_rate": 4.4097995545657015e-06,
|
| 30847 |
+
"loss": 1.1339,
|
| 30848 |
+
"step": 4401
|
| 30849 |
+
},
|
| 30850 |
+
{
|
| 30851 |
+
"epoch": 0.9782222222222222,
|
| 30852 |
+
"grad_norm": 0.8057011961936951,
|
| 30853 |
+
"learning_rate": 4.365256124721604e-06,
|
| 30854 |
+
"loss": 2.1264,
|
| 30855 |
+
"step": 4402
|
| 30856 |
+
},
|
| 30857 |
+
{
|
| 30858 |
+
"epoch": 0.9784444444444444,
|
| 30859 |
+
"grad_norm": 0.6433674097061157,
|
| 30860 |
+
"learning_rate": 4.320712694877506e-06,
|
| 30861 |
+
"loss": 1.1829,
|
| 30862 |
+
"step": 4403
|
| 30863 |
+
},
|
| 30864 |
+
{
|
| 30865 |
+
"epoch": 0.9786666666666667,
|
| 30866 |
+
"grad_norm": 0.5455031991004944,
|
| 30867 |
+
"learning_rate": 4.2761692650334074e-06,
|
| 30868 |
+
"loss": 1.2065,
|
| 30869 |
+
"step": 4404
|
| 30870 |
+
},
|
| 30871 |
+
{
|
| 30872 |
+
"epoch": 0.9788888888888889,
|
| 30873 |
+
"grad_norm": 0.883698582649231,
|
| 30874 |
+
"learning_rate": 4.231625835189309e-06,
|
| 30875 |
+
"loss": 2.0593,
|
| 30876 |
+
"step": 4405
|
| 30877 |
+
},
|
| 30878 |
+
{
|
| 30879 |
+
"epoch": 0.9791111111111112,
|
| 30880 |
+
"grad_norm": 0.6390405297279358,
|
| 30881 |
+
"learning_rate": 4.187082405345212e-06,
|
| 30882 |
+
"loss": 0.9819,
|
| 30883 |
+
"step": 4406
|
| 30884 |
+
},
|
| 30885 |
+
{
|
| 30886 |
+
"epoch": 0.9793333333333333,
|
| 30887 |
+
"grad_norm": 0.8261483311653137,
|
| 30888 |
+
"learning_rate": 4.142538975501114e-06,
|
| 30889 |
+
"loss": 2.308,
|
| 30890 |
+
"step": 4407
|
| 30891 |
+
},
|
| 30892 |
+
{
|
| 30893 |
+
"epoch": 0.9795555555555555,
|
| 30894 |
+
"grad_norm": 0.6975874900817871,
|
| 30895 |
+
"learning_rate": 4.097995545657015e-06,
|
| 30896 |
+
"loss": 1.0839,
|
| 30897 |
+
"step": 4408
|
| 30898 |
+
},
|
| 30899 |
+
{
|
| 30900 |
+
"epoch": 0.9797777777777777,
|
| 30901 |
+
"grad_norm": 0.6761125326156616,
|
| 30902 |
+
"learning_rate": 4.053452115812917e-06,
|
| 30903 |
+
"loss": 0.9341,
|
| 30904 |
+
"step": 4409
|
| 30905 |
+
},
|
| 30906 |
+
{
|
| 30907 |
+
"epoch": 0.98,
|
| 30908 |
+
"grad_norm": 0.7890470027923584,
|
| 30909 |
+
"learning_rate": 4.00890868596882e-06,
|
| 30910 |
+
"loss": 1.901,
|
| 30911 |
+
"step": 4410
|
| 30912 |
+
},
|
| 30913 |
+
{
|
| 30914 |
+
"epoch": 0.9802222222222222,
|
| 30915 |
+
"grad_norm": 0.9101024866104126,
|
| 30916 |
+
"learning_rate": 3.964365256124722e-06,
|
| 30917 |
+
"loss": 1.9368,
|
| 30918 |
+
"step": 4411
|
| 30919 |
+
},
|
| 30920 |
+
{
|
| 30921 |
+
"epoch": 0.9804444444444445,
|
| 30922 |
+
"grad_norm": 0.9786936640739441,
|
| 30923 |
+
"learning_rate": 3.919821826280624e-06,
|
| 30924 |
+
"loss": 2.1794,
|
| 30925 |
+
"step": 4412
|
| 30926 |
+
},
|
| 30927 |
+
{
|
| 30928 |
+
"epoch": 0.9806666666666667,
|
| 30929 |
+
"grad_norm": 0.7119241952896118,
|
| 30930 |
+
"learning_rate": 3.875278396436525e-06,
|
| 30931 |
+
"loss": 1.0641,
|
| 30932 |
+
"step": 4413
|
| 30933 |
+
},
|
| 30934 |
+
{
|
| 30935 |
+
"epoch": 0.9808888888888889,
|
| 30936 |
+
"grad_norm": 0.09762410819530487,
|
| 30937 |
+
"learning_rate": 3.830734966592428e-06,
|
| 30938 |
+
"loss": 0.0164,
|
| 30939 |
+
"step": 4414
|
| 30940 |
+
},
|
| 30941 |
+
{
|
| 30942 |
+
"epoch": 0.9811111111111112,
|
| 30943 |
+
"grad_norm": 0.7070305943489075,
|
| 30944 |
+
"learning_rate": 3.78619153674833e-06,
|
| 30945 |
+
"loss": 0.9271,
|
| 30946 |
+
"step": 4415
|
| 30947 |
+
},
|
| 30948 |
+
{
|
| 30949 |
+
"epoch": 0.9813333333333333,
|
| 30950 |
+
"grad_norm": 0.9111929535865784,
|
| 30951 |
+
"learning_rate": 3.7416481069042315e-06,
|
| 30952 |
+
"loss": 1.7253,
|
| 30953 |
+
"step": 4416
|
| 30954 |
+
},
|
| 30955 |
+
{
|
| 30956 |
+
"epoch": 0.9815555555555555,
|
| 30957 |
+
"grad_norm": 1.0224978923797607,
|
| 30958 |
+
"learning_rate": 3.6971046770601335e-06,
|
| 30959 |
+
"loss": 2.0652,
|
| 30960 |
+
"step": 4417
|
| 30961 |
+
},
|
| 30962 |
+
{
|
| 30963 |
+
"epoch": 0.9817777777777777,
|
| 30964 |
+
"grad_norm": 1.2484158277511597,
|
| 30965 |
+
"learning_rate": 3.652561247216036e-06,
|
| 30966 |
+
"loss": 2.263,
|
| 30967 |
+
"step": 4418
|
| 30968 |
+
},
|
| 30969 |
+
{
|
| 30970 |
+
"epoch": 0.982,
|
| 30971 |
+
"grad_norm": 0.6740500926971436,
|
| 30972 |
+
"learning_rate": 3.608017817371938e-06,
|
| 30973 |
+
"loss": 0.8257,
|
| 30974 |
+
"step": 4419
|
| 30975 |
+
},
|
| 30976 |
+
{
|
| 30977 |
+
"epoch": 0.9822222222222222,
|
| 30978 |
+
"grad_norm": 0.06990643590688705,
|
| 30979 |
+
"learning_rate": 3.56347438752784e-06,
|
| 30980 |
+
"loss": 0.0152,
|
| 30981 |
+
"step": 4420
|
| 30982 |
+
},
|
| 30983 |
+
{
|
| 30984 |
+
"epoch": 0.9824444444444445,
|
| 30985 |
+
"grad_norm": 0.728216826915741,
|
| 30986 |
+
"learning_rate": 3.5189309576837414e-06,
|
| 30987 |
+
"loss": 0.8901,
|
| 30988 |
+
"step": 4421
|
| 30989 |
+
},
|
| 30990 |
+
{
|
| 30991 |
+
"epoch": 0.9826666666666667,
|
| 30992 |
+
"grad_norm": 0.06907333433628082,
|
| 30993 |
+
"learning_rate": 3.474387527839644e-06,
|
| 30994 |
+
"loss": 0.0152,
|
| 30995 |
+
"step": 4422
|
| 30996 |
+
},
|
| 30997 |
+
{
|
| 30998 |
+
"epoch": 0.9828888888888889,
|
| 30999 |
+
"grad_norm": 0.792972981929779,
|
| 31000 |
+
"learning_rate": 3.4298440979955457e-06,
|
| 31001 |
+
"loss": 0.8383,
|
| 31002 |
+
"step": 4423
|
| 31003 |
+
},
|
| 31004 |
+
{
|
| 31005 |
+
"epoch": 0.9831111111111112,
|
| 31006 |
+
"grad_norm": 0.9240522384643555,
|
| 31007 |
+
"learning_rate": 3.3853006681514477e-06,
|
| 31008 |
+
"loss": 1.9004,
|
| 31009 |
+
"step": 4424
|
| 31010 |
+
},
|
| 31011 |
+
{
|
| 31012 |
+
"epoch": 0.9833333333333333,
|
| 31013 |
+
"grad_norm": 0.9684634208679199,
|
| 31014 |
+
"learning_rate": 3.34075723830735e-06,
|
| 31015 |
+
"loss": 1.7412,
|
| 31016 |
+
"step": 4425
|
| 31017 |
+
},
|
| 31018 |
+
{
|
| 31019 |
+
"epoch": 0.9835555555555555,
|
| 31020 |
+
"grad_norm": 1.075197696685791,
|
| 31021 |
+
"learning_rate": 3.296213808463252e-06,
|
| 31022 |
+
"loss": 1.4785,
|
| 31023 |
+
"step": 4426
|
| 31024 |
+
},
|
| 31025 |
+
{
|
| 31026 |
+
"epoch": 0.9837777777777778,
|
| 31027 |
+
"grad_norm": 0.9526484608650208,
|
| 31028 |
+
"learning_rate": 3.2516703786191536e-06,
|
| 31029 |
+
"loss": 1.6998,
|
| 31030 |
+
"step": 4427
|
| 31031 |
+
},
|
| 31032 |
+
{
|
| 31033 |
+
"epoch": 0.984,
|
| 31034 |
+
"grad_norm": 0.995002269744873,
|
| 31035 |
+
"learning_rate": 3.2071269487750556e-06,
|
| 31036 |
+
"loss": 1.5562,
|
| 31037 |
+
"step": 4428
|
| 31038 |
+
},
|
| 31039 |
+
{
|
| 31040 |
+
"epoch": 0.9842222222222222,
|
| 31041 |
+
"grad_norm": 1.0168581008911133,
|
| 31042 |
+
"learning_rate": 3.162583518930958e-06,
|
| 31043 |
+
"loss": 1.722,
|
| 31044 |
+
"step": 4429
|
| 31045 |
+
},
|
| 31046 |
+
{
|
| 31047 |
+
"epoch": 0.9844444444444445,
|
| 31048 |
+
"grad_norm": 0.06682226806879044,
|
| 31049 |
+
"learning_rate": 3.11804008908686e-06,
|
| 31050 |
+
"loss": 0.0176,
|
| 31051 |
+
"step": 4430
|
| 31052 |
+
},
|
| 31053 |
+
{
|
| 31054 |
+
"epoch": 0.9846666666666667,
|
| 31055 |
+
"grad_norm": 0.06983762979507446,
|
| 31056 |
+
"learning_rate": 3.073496659242762e-06,
|
| 31057 |
+
"loss": 0.0173,
|
| 31058 |
+
"step": 4431
|
| 31059 |
+
},
|
| 31060 |
+
{
|
| 31061 |
+
"epoch": 0.9848888888888889,
|
| 31062 |
+
"grad_norm": 0.06633459031581879,
|
| 31063 |
+
"learning_rate": 3.028953229398664e-06,
|
| 31064 |
+
"loss": 0.0174,
|
| 31065 |
+
"step": 4432
|
| 31066 |
+
},
|
| 31067 |
+
{
|
| 31068 |
+
"epoch": 0.9851111111111112,
|
| 31069 |
+
"grad_norm": 0.06410173326730728,
|
| 31070 |
+
"learning_rate": 2.984409799554566e-06,
|
| 31071 |
+
"loss": 0.0175,
|
| 31072 |
+
"step": 4433
|
| 31073 |
+
},
|
| 31074 |
+
{
|
| 31075 |
+
"epoch": 0.9853333333333333,
|
| 31076 |
+
"grad_norm": 0.8998127579689026,
|
| 31077 |
+
"learning_rate": 2.939866369710468e-06,
|
| 31078 |
+
"loss": 1.6707,
|
| 31079 |
+
"step": 4434
|
| 31080 |
+
},
|
| 31081 |
+
{
|
| 31082 |
+
"epoch": 0.9855555555555555,
|
| 31083 |
+
"grad_norm": 0.08308030664920807,
|
| 31084 |
+
"learning_rate": 2.89532293986637e-06,
|
| 31085 |
+
"loss": 0.0179,
|
| 31086 |
+
"step": 4435
|
| 31087 |
+
},
|
| 31088 |
+
{
|
| 31089 |
+
"epoch": 0.9857777777777778,
|
| 31090 |
+
"grad_norm": 0.8417572379112244,
|
| 31091 |
+
"learning_rate": 2.8507795100222718e-06,
|
| 31092 |
+
"loss": 0.966,
|
| 31093 |
+
"step": 4436
|
| 31094 |
+
},
|
| 31095 |
+
{
|
| 31096 |
+
"epoch": 0.986,
|
| 31097 |
+
"grad_norm": 1.395193338394165,
|
| 31098 |
+
"learning_rate": 2.8062360801781737e-06,
|
| 31099 |
+
"loss": 1.8783,
|
| 31100 |
+
"step": 4437
|
| 31101 |
+
},
|
| 31102 |
+
{
|
| 31103 |
+
"epoch": 0.9862222222222222,
|
| 31104 |
+
"grad_norm": 0.9416733384132385,
|
| 31105 |
+
"learning_rate": 2.761692650334076e-06,
|
| 31106 |
+
"loss": 1.5385,
|
| 31107 |
+
"step": 4438
|
| 31108 |
+
},
|
| 31109 |
+
{
|
| 31110 |
+
"epoch": 0.9864444444444445,
|
| 31111 |
+
"grad_norm": 1.100425362586975,
|
| 31112 |
+
"learning_rate": 2.7171492204899777e-06,
|
| 31113 |
+
"loss": 1.7733,
|
| 31114 |
+
"step": 4439
|
| 31115 |
+
},
|
| 31116 |
+
{
|
| 31117 |
+
"epoch": 0.9866666666666667,
|
| 31118 |
+
"grad_norm": 0.7649857401847839,
|
| 31119 |
+
"learning_rate": 2.67260579064588e-06,
|
| 31120 |
+
"loss": 0.7851,
|
| 31121 |
+
"step": 4440
|
| 31122 |
+
},
|
| 31123 |
+
{
|
| 31124 |
+
"epoch": 0.9868888888888889,
|
| 31125 |
+
"grad_norm": 1.1875056028366089,
|
| 31126 |
+
"learning_rate": 2.6280623608017816e-06,
|
| 31127 |
+
"loss": 1.6325,
|
| 31128 |
+
"step": 4441
|
| 31129 |
+
},
|
| 31130 |
+
{
|
| 31131 |
+
"epoch": 0.9871111111111112,
|
| 31132 |
+
"grad_norm": 1.1401832103729248,
|
| 31133 |
+
"learning_rate": 2.583518930957684e-06,
|
| 31134 |
+
"loss": 1.6937,
|
| 31135 |
+
"step": 4442
|
| 31136 |
+
},
|
| 31137 |
+
{
|
| 31138 |
+
"epoch": 0.9873333333333333,
|
| 31139 |
+
"grad_norm": 1.1035478115081787,
|
| 31140 |
+
"learning_rate": 2.5389755011135856e-06,
|
| 31141 |
+
"loss": 1.526,
|
| 31142 |
+
"step": 4443
|
| 31143 |
+
},
|
| 31144 |
+
{
|
| 31145 |
+
"epoch": 0.9875555555555555,
|
| 31146 |
+
"grad_norm": 0.8037136793136597,
|
| 31147 |
+
"learning_rate": 2.494432071269488e-06,
|
| 31148 |
+
"loss": 0.7856,
|
| 31149 |
+
"step": 4444
|
| 31150 |
+
},
|
| 31151 |
+
{
|
| 31152 |
+
"epoch": 0.9877777777777778,
|
| 31153 |
+
"grad_norm": 1.0584372282028198,
|
| 31154 |
+
"learning_rate": 2.44988864142539e-06,
|
| 31155 |
+
"loss": 1.3084,
|
| 31156 |
+
"step": 4445
|
| 31157 |
+
},
|
| 31158 |
+
{
|
| 31159 |
+
"epoch": 0.988,
|
| 31160 |
+
"grad_norm": 0.1836099475622177,
|
| 31161 |
+
"learning_rate": 2.405345211581292e-06,
|
| 31162 |
+
"loss": 0.0299,
|
| 31163 |
+
"step": 4446
|
| 31164 |
+
},
|
| 31165 |
+
{
|
| 31166 |
+
"epoch": 0.9882222222222222,
|
| 31167 |
+
"grad_norm": 1.108872413635254,
|
| 31168 |
+
"learning_rate": 2.360801781737194e-06,
|
| 31169 |
+
"loss": 1.0455,
|
| 31170 |
+
"step": 4447
|
| 31171 |
+
},
|
| 31172 |
+
{
|
| 31173 |
+
"epoch": 0.9884444444444445,
|
| 31174 |
+
"grad_norm": 0.6207655072212219,
|
| 31175 |
+
"learning_rate": 2.316258351893096e-06,
|
| 31176 |
+
"loss": 0.4939,
|
| 31177 |
+
"step": 4448
|
| 31178 |
+
},
|
| 31179 |
+
{
|
| 31180 |
+
"epoch": 0.9886666666666667,
|
| 31181 |
+
"grad_norm": 0.14554363489151,
|
| 31182 |
+
"learning_rate": 2.2717149220489982e-06,
|
| 31183 |
+
"loss": 0.0318,
|
| 31184 |
+
"step": 4449
|
| 31185 |
+
},
|
| 31186 |
+
{
|
| 31187 |
+
"epoch": 0.9888888888888889,
|
| 31188 |
+
"grad_norm": 1.2572603225708008,
|
| 31189 |
+
"learning_rate": 2.2271714922048998e-06,
|
| 31190 |
+
"loss": 1.1098,
|
| 31191 |
+
"step": 4450
|
| 31192 |
+
},
|
| 31193 |
+
{
|
| 31194 |
+
"epoch": 0.9891111111111112,
|
| 31195 |
+
"grad_norm": 0.04474545270204544,
|
| 31196 |
+
"learning_rate": 2.182628062360802e-06,
|
| 31197 |
+
"loss": 0.0103,
|
| 31198 |
+
"step": 4451
|
| 31199 |
+
},
|
| 31200 |
+
{
|
| 31201 |
+
"epoch": 0.9893333333333333,
|
| 31202 |
+
"grad_norm": 0.04636682942509651,
|
| 31203 |
+
"learning_rate": 2.1380846325167037e-06,
|
| 31204 |
+
"loss": 0.0101,
|
| 31205 |
+
"step": 4452
|
| 31206 |
+
},
|
| 31207 |
+
{
|
| 31208 |
+
"epoch": 0.9895555555555555,
|
| 31209 |
+
"grad_norm": 0.5653097033500671,
|
| 31210 |
+
"learning_rate": 2.093541202672606e-06,
|
| 31211 |
+
"loss": 1.0041,
|
| 31212 |
+
"step": 4453
|
| 31213 |
+
},
|
| 31214 |
+
{
|
| 31215 |
+
"epoch": 0.9897777777777778,
|
| 31216 |
+
"grad_norm": 0.4789440333843231,
|
| 31217 |
+
"learning_rate": 2.0489977728285077e-06,
|
| 31218 |
+
"loss": 0.979,
|
| 31219 |
+
"step": 4454
|
| 31220 |
+
},
|
| 31221 |
+
{
|
| 31222 |
+
"epoch": 0.99,
|
| 31223 |
+
"grad_norm": 0.8047142028808594,
|
| 31224 |
+
"learning_rate": 2.00445434298441e-06,
|
| 31225 |
+
"loss": 2.0886,
|
| 31226 |
+
"step": 4455
|
| 31227 |
+
},
|
| 31228 |
+
{
|
| 31229 |
+
"epoch": 0.9902222222222222,
|
| 31230 |
+
"grad_norm": 0.8989213109016418,
|
| 31231 |
+
"learning_rate": 1.959910913140312e-06,
|
| 31232 |
+
"loss": 2.1387,
|
| 31233 |
+
"step": 4456
|
| 31234 |
+
},
|
| 31235 |
+
{
|
| 31236 |
+
"epoch": 0.9904444444444445,
|
| 31237 |
+
"grad_norm": 0.06995019316673279,
|
| 31238 |
+
"learning_rate": 1.915367483296214e-06,
|
| 31239 |
+
"loss": 0.0109,
|
| 31240 |
+
"step": 4457
|
| 31241 |
+
},
|
| 31242 |
+
{
|
| 31243 |
+
"epoch": 0.9906666666666667,
|
| 31244 |
+
"grad_norm": 0.07215920835733414,
|
| 31245 |
+
"learning_rate": 1.8708240534521158e-06,
|
| 31246 |
+
"loss": 0.0108,
|
| 31247 |
+
"step": 4458
|
| 31248 |
+
},
|
| 31249 |
+
{
|
| 31250 |
+
"epoch": 0.9908888888888889,
|
| 31251 |
+
"grad_norm": 0.07202310115098953,
|
| 31252 |
+
"learning_rate": 1.826280623608018e-06,
|
| 31253 |
+
"loss": 0.0109,
|
| 31254 |
+
"step": 4459
|
| 31255 |
+
},
|
| 31256 |
+
{
|
| 31257 |
+
"epoch": 0.9911111111111112,
|
| 31258 |
+
"grad_norm": 0.9508035778999329,
|
| 31259 |
+
"learning_rate": 1.78173719376392e-06,
|
| 31260 |
+
"loss": 2.2415,
|
| 31261 |
+
"step": 4460
|
| 31262 |
+
},
|
| 31263 |
+
{
|
| 31264 |
+
"epoch": 0.9913333333333333,
|
| 31265 |
+
"grad_norm": 0.891727864742279,
|
| 31266 |
+
"learning_rate": 1.737193763919822e-06,
|
| 31267 |
+
"loss": 1.9116,
|
| 31268 |
+
"step": 4461
|
| 31269 |
+
},
|
| 31270 |
+
{
|
| 31271 |
+
"epoch": 0.9915555555555555,
|
| 31272 |
+
"grad_norm": 1.0234503746032715,
|
| 31273 |
+
"learning_rate": 1.6926503340757238e-06,
|
| 31274 |
+
"loss": 2.0408,
|
| 31275 |
+
"step": 4462
|
| 31276 |
+
},
|
| 31277 |
+
{
|
| 31278 |
+
"epoch": 0.9917777777777778,
|
| 31279 |
+
"grad_norm": 0.8998834490776062,
|
| 31280 |
+
"learning_rate": 1.648106904231626e-06,
|
| 31281 |
+
"loss": 2.0895,
|
| 31282 |
+
"step": 4463
|
| 31283 |
+
},
|
| 31284 |
+
{
|
| 31285 |
+
"epoch": 0.992,
|
| 31286 |
+
"grad_norm": 0.9309079051017761,
|
| 31287 |
+
"learning_rate": 1.6035634743875278e-06,
|
| 31288 |
+
"loss": 1.9546,
|
| 31289 |
+
"step": 4464
|
| 31290 |
+
},
|
| 31291 |
+
{
|
| 31292 |
+
"epoch": 0.9922222222222222,
|
| 31293 |
+
"grad_norm": 0.903396725654602,
|
| 31294 |
+
"learning_rate": 1.55902004454343e-06,
|
| 31295 |
+
"loss": 1.0776,
|
| 31296 |
+
"step": 4465
|
| 31297 |
+
},
|
| 31298 |
+
{
|
| 31299 |
+
"epoch": 0.9924444444444445,
|
| 31300 |
+
"grad_norm": 1.0036734342575073,
|
| 31301 |
+
"learning_rate": 1.514476614699332e-06,
|
| 31302 |
+
"loss": 1.7439,
|
| 31303 |
+
"step": 4466
|
| 31304 |
+
},
|
| 31305 |
+
{
|
| 31306 |
+
"epoch": 0.9926666666666667,
|
| 31307 |
+
"grad_norm": 0.9246737957000732,
|
| 31308 |
+
"learning_rate": 1.469933184855234e-06,
|
| 31309 |
+
"loss": 1.7637,
|
| 31310 |
+
"step": 4467
|
| 31311 |
+
},
|
| 31312 |
+
{
|
| 31313 |
+
"epoch": 0.9928888888888889,
|
| 31314 |
+
"grad_norm": 1.0618118047714233,
|
| 31315 |
+
"learning_rate": 1.4253897550111359e-06,
|
| 31316 |
+
"loss": 1.9589,
|
| 31317 |
+
"step": 4468
|
| 31318 |
+
},
|
| 31319 |
+
{
|
| 31320 |
+
"epoch": 0.9931111111111111,
|
| 31321 |
+
"grad_norm": 1.1122076511383057,
|
| 31322 |
+
"learning_rate": 1.380846325167038e-06,
|
| 31323 |
+
"loss": 1.9023,
|
| 31324 |
+
"step": 4469
|
| 31325 |
+
},
|
| 31326 |
+
{
|
| 31327 |
+
"epoch": 0.9933333333333333,
|
| 31328 |
+
"grad_norm": 1.027601957321167,
|
| 31329 |
+
"learning_rate": 1.33630289532294e-06,
|
| 31330 |
+
"loss": 1.9814,
|
| 31331 |
+
"step": 4470
|
| 31332 |
+
},
|
| 31333 |
+
{
|
| 31334 |
+
"epoch": 0.9935555555555555,
|
| 31335 |
+
"grad_norm": 0.06850501894950867,
|
| 31336 |
+
"learning_rate": 1.291759465478842e-06,
|
| 31337 |
+
"loss": 0.0152,
|
| 31338 |
+
"step": 4471
|
| 31339 |
+
},
|
| 31340 |
+
{
|
| 31341 |
+
"epoch": 0.9937777777777778,
|
| 31342 |
+
"grad_norm": 0.067985400557518,
|
| 31343 |
+
"learning_rate": 1.247216035634744e-06,
|
| 31344 |
+
"loss": 0.0152,
|
| 31345 |
+
"step": 4472
|
| 31346 |
+
},
|
| 31347 |
+
{
|
| 31348 |
+
"epoch": 0.994,
|
| 31349 |
+
"grad_norm": 1.0229130983352661,
|
| 31350 |
+
"learning_rate": 1.202672605790646e-06,
|
| 31351 |
+
"loss": 1.6158,
|
| 31352 |
+
"step": 4473
|
| 31353 |
+
},
|
| 31354 |
+
{
|
| 31355 |
+
"epoch": 0.9942222222222222,
|
| 31356 |
+
"grad_norm": 0.6642321944236755,
|
| 31357 |
+
"learning_rate": 1.158129175946548e-06,
|
| 31358 |
+
"loss": 0.7592,
|
| 31359 |
+
"step": 4474
|
| 31360 |
+
},
|
| 31361 |
+
{
|
| 31362 |
+
"epoch": 0.9944444444444445,
|
| 31363 |
+
"grad_norm": 1.025769829750061,
|
| 31364 |
+
"learning_rate": 1.1135857461024499e-06,
|
| 31365 |
+
"loss": 1.8864,
|
| 31366 |
+
"step": 4475
|
| 31367 |
+
},
|
| 31368 |
+
{
|
| 31369 |
+
"epoch": 0.9946666666666667,
|
| 31370 |
+
"grad_norm": 1.1777735948562622,
|
| 31371 |
+
"learning_rate": 1.0690423162583519e-06,
|
| 31372 |
+
"loss": 1.9098,
|
| 31373 |
+
"step": 4476
|
| 31374 |
+
},
|
| 31375 |
+
{
|
| 31376 |
+
"epoch": 0.9948888888888889,
|
| 31377 |
+
"grad_norm": 1.0232651233673096,
|
| 31378 |
+
"learning_rate": 1.0244988864142538e-06,
|
| 31379 |
+
"loss": 1.6622,
|
| 31380 |
+
"step": 4477
|
| 31381 |
+
},
|
| 31382 |
+
{
|
| 31383 |
+
"epoch": 0.9951111111111111,
|
| 31384 |
+
"grad_norm": 1.0267844200134277,
|
| 31385 |
+
"learning_rate": 9.79955456570156e-07,
|
| 31386 |
+
"loss": 1.8175,
|
| 31387 |
+
"step": 4478
|
| 31388 |
+
},
|
| 31389 |
+
{
|
| 31390 |
+
"epoch": 0.9953333333333333,
|
| 31391 |
+
"grad_norm": 0.7749679684638977,
|
| 31392 |
+
"learning_rate": 9.354120267260579e-07,
|
| 31393 |
+
"loss": 0.906,
|
| 31394 |
+
"step": 4479
|
| 31395 |
+
},
|
| 31396 |
+
{
|
| 31397 |
+
"epoch": 0.9955555555555555,
|
| 31398 |
+
"grad_norm": 0.06536448746919632,
|
| 31399 |
+
"learning_rate": 8.9086859688196e-07,
|
| 31400 |
+
"loss": 0.0173,
|
| 31401 |
+
"step": 4480
|
| 31402 |
+
},
|
| 31403 |
+
{
|
| 31404 |
+
"epoch": 0.9957777777777778,
|
| 31405 |
+
"grad_norm": 0.6798564195632935,
|
| 31406 |
+
"learning_rate": 8.463251670378619e-07,
|
| 31407 |
+
"loss": 0.7955,
|
| 31408 |
+
"step": 4481
|
| 31409 |
+
},
|
| 31410 |
+
{
|
| 31411 |
+
"epoch": 0.996,
|
| 31412 |
+
"grad_norm": 0.06655056774616241,
|
| 31413 |
+
"learning_rate": 8.017817371937639e-07,
|
| 31414 |
+
"loss": 0.0176,
|
| 31415 |
+
"step": 4482
|
| 31416 |
+
},
|
| 31417 |
+
{
|
| 31418 |
+
"epoch": 0.9962222222222222,
|
| 31419 |
+
"grad_norm": 0.7525641918182373,
|
| 31420 |
+
"learning_rate": 7.57238307349666e-07,
|
| 31421 |
+
"loss": 0.8103,
|
| 31422 |
+
"step": 4483
|
| 31423 |
+
},
|
| 31424 |
+
{
|
| 31425 |
+
"epoch": 0.9964444444444445,
|
| 31426 |
+
"grad_norm": 0.6724408268928528,
|
| 31427 |
+
"learning_rate": 7.126948775055679e-07,
|
| 31428 |
+
"loss": 0.7957,
|
| 31429 |
+
"step": 4484
|
| 31430 |
+
},
|
| 31431 |
+
{
|
| 31432 |
+
"epoch": 0.9966666666666667,
|
| 31433 |
+
"grad_norm": 0.99349445104599,
|
| 31434 |
+
"learning_rate": 6.6815144766147e-07,
|
| 31435 |
+
"loss": 1.691,
|
| 31436 |
+
"step": 4485
|
| 31437 |
+
},
|
| 31438 |
+
{
|
| 31439 |
+
"epoch": 0.9968888888888889,
|
| 31440 |
+
"grad_norm": 1.0608917474746704,
|
| 31441 |
+
"learning_rate": 6.23608017817372e-07,
|
| 31442 |
+
"loss": 1.6244,
|
| 31443 |
+
"step": 4486
|
| 31444 |
+
},
|
| 31445 |
+
{
|
| 31446 |
+
"epoch": 0.9971111111111111,
|
| 31447 |
+
"grad_norm": 0.07752467691898346,
|
| 31448 |
+
"learning_rate": 5.79064587973274e-07,
|
| 31449 |
+
"loss": 0.018,
|
| 31450 |
+
"step": 4487
|
| 31451 |
+
},
|
| 31452 |
+
{
|
| 31453 |
+
"epoch": 0.9973333333333333,
|
| 31454 |
+
"grad_norm": 0.7708075046539307,
|
| 31455 |
+
"learning_rate": 5.345211581291759e-07,
|
| 31456 |
+
"loss": 0.8414,
|
| 31457 |
+
"step": 4488
|
| 31458 |
+
},
|
| 31459 |
+
{
|
| 31460 |
+
"epoch": 0.9975555555555555,
|
| 31461 |
+
"grad_norm": 0.6976569890975952,
|
| 31462 |
+
"learning_rate": 4.89977728285078e-07,
|
| 31463 |
+
"loss": 0.7637,
|
| 31464 |
+
"step": 4489
|
| 31465 |
+
},
|
| 31466 |
+
{
|
| 31467 |
+
"epoch": 0.9977777777777778,
|
| 31468 |
+
"grad_norm": 1.0548564195632935,
|
| 31469 |
+
"learning_rate": 4.4543429844098e-07,
|
| 31470 |
+
"loss": 1.4826,
|
| 31471 |
+
"step": 4490
|
| 31472 |
+
},
|
| 31473 |
+
{
|
| 31474 |
+
"epoch": 0.998,
|
| 31475 |
+
"grad_norm": 0.6447573900222778,
|
| 31476 |
+
"learning_rate": 4.0089086859688195e-07,
|
| 31477 |
+
"loss": 0.6006,
|
| 31478 |
+
"step": 4491
|
| 31479 |
+
},
|
| 31480 |
+
{
|
| 31481 |
+
"epoch": 0.9982222222222222,
|
| 31482 |
+
"grad_norm": 1.2821402549743652,
|
| 31483 |
+
"learning_rate": 3.5634743875278397e-07,
|
| 31484 |
+
"loss": 1.8731,
|
| 31485 |
+
"step": 4492
|
| 31486 |
+
},
|
| 31487 |
+
{
|
| 31488 |
+
"epoch": 0.9984444444444445,
|
| 31489 |
+
"grad_norm": 1.1518702507019043,
|
| 31490 |
+
"learning_rate": 3.11804008908686e-07,
|
| 31491 |
+
"loss": 1.5884,
|
| 31492 |
+
"step": 4493
|
| 31493 |
+
},
|
| 31494 |
+
{
|
| 31495 |
+
"epoch": 0.9986666666666667,
|
| 31496 |
+
"grad_norm": 1.11997389793396,
|
| 31497 |
+
"learning_rate": 2.6726057906458796e-07,
|
| 31498 |
+
"loss": 1.4486,
|
| 31499 |
+
"step": 4494
|
| 31500 |
+
},
|
| 31501 |
+
{
|
| 31502 |
+
"epoch": 0.9988888888888889,
|
| 31503 |
+
"grad_norm": 1.1092532873153687,
|
| 31504 |
+
"learning_rate": 2.2271714922049e-07,
|
| 31505 |
+
"loss": 1.3308,
|
| 31506 |
+
"step": 4495
|
| 31507 |
+
},
|
| 31508 |
+
{
|
| 31509 |
+
"epoch": 0.9991111111111111,
|
| 31510 |
+
"grad_norm": 0.17926262319087982,
|
| 31511 |
+
"learning_rate": 1.7817371937639199e-07,
|
| 31512 |
+
"loss": 0.0297,
|
| 31513 |
+
"step": 4496
|
| 31514 |
+
},
|
| 31515 |
+
{
|
| 31516 |
+
"epoch": 0.9993333333333333,
|
| 31517 |
+
"grad_norm": 1.144982933998108,
|
| 31518 |
+
"learning_rate": 1.3363028953229398e-07,
|
| 31519 |
+
"loss": 1.2413,
|
| 31520 |
+
"step": 4497
|
| 31521 |
+
},
|
| 31522 |
+
{
|
| 31523 |
+
"epoch": 0.9995555555555555,
|
| 31524 |
+
"grad_norm": 1.1863489151000977,
|
| 31525 |
+
"learning_rate": 8.908685968819599e-08,
|
| 31526 |
+
"loss": 1.1693,
|
| 31527 |
+
"step": 4498
|
| 31528 |
+
},
|
| 31529 |
+
{
|
| 31530 |
+
"epoch": 0.9997777777777778,
|
| 31531 |
+
"grad_norm": 0.7812955975532532,
|
| 31532 |
+
"learning_rate": 4.4543429844097996e-08,
|
| 31533 |
+
"loss": 0.5973,
|
| 31534 |
+
"step": 4499
|
| 31535 |
+
},
|
| 31536 |
+
{
|
| 31537 |
+
"epoch": 1.0,
|
| 31538 |
+
"grad_norm": 1.0512616634368896,
|
| 31539 |
+
"learning_rate": 0.0,
|
| 31540 |
+
"loss": 0.8343,
|
| 31541 |
+
"step": 4500
|
| 31542 |
+
},
|
| 31543 |
+
{
|
| 31544 |
+
"epoch": 1.0,
|
| 31545 |
+
"eval_loss": 1.1682192087173462,
|
| 31546 |
+
"eval_runtime": 240.8872,
|
| 31547 |
+
"eval_samples_per_second": 4.151,
|
| 31548 |
+
"eval_steps_per_second": 4.151,
|
| 31549 |
+
"step": 4500
|
| 31550 |
}
|
| 31551 |
],
|
| 31552 |
"logging_steps": 1,
|
|
|
|
| 31561 |
"should_evaluate": false,
|
| 31562 |
"should_log": false,
|
| 31563 |
"should_save": true,
|
| 31564 |
+
"should_training_stop": true
|
| 31565 |
},
|
| 31566 |
"attributes": {}
|
| 31567 |
}
|
| 31568 |
},
|
| 31569 |
+
"total_flos": 4.847769692985754e+16,
|
| 31570 |
"train_batch_size": 1,
|
| 31571 |
"trial_name": null,
|
| 31572 |
"trial_params": null
|