Training in progress, step 8211, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 306619286
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6cda0b31f70d9aa1b6b0d2861a2bbcd9fb8c95012a6ab358768d46b28163374a
|
| 3 |
size 306619286
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 919972410
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:96f21eb8f0138de484c920cb8f7bdeb36ad541352260059030b3b39bc9b5fb3b
|
| 3 |
size 919972410
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1000
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b0aceef701350b39c9357a1bb5166b5f89159643c4a993e803dff7eb40b4b245
|
| 3 |
size 1000
|
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": 5000,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -3515,6 +3515,2253 @@
|
|
| 3515 |
"eval_samples_per_second": 2268.978,
|
| 3516 |
"eval_steps_per_second": 35.453,
|
| 3517 |
"step": 5000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3518 |
}
|
| 3519 |
],
|
| 3520 |
"logging_steps": 10,
|
|
@@ -3529,12 +5776,12 @@
|
|
| 3529 |
"should_evaluate": false,
|
| 3530 |
"should_log": false,
|
| 3531 |
"should_save": true,
|
| 3532 |
-
"should_training_stop":
|
| 3533 |
},
|
| 3534 |
"attributes": {}
|
| 3535 |
}
|
| 3536 |
},
|
| 3537 |
-
"total_flos":
|
| 3538 |
"train_batch_size": 16,
|
| 3539 |
"trial_name": null,
|
| 3540 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9998934469898775,
|
| 5 |
"eval_steps": 5000,
|
| 6 |
+
"global_step": 8211,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 3515 |
"eval_samples_per_second": 2268.978,
|
| 3516 |
"eval_steps_per_second": 35.453,
|
| 3517 |
"step": 5000
|
| 3518 |
+
},
|
| 3519 |
+
{
|
| 3520 |
+
"epoch": 0.610092092244463,
|
| 3521 |
+
"grad_norm": 511.75,
|
| 3522 |
+
"learning_rate": 9.904673654728358e-06,
|
| 3523 |
+
"loss": 22.1654,
|
| 3524 |
+
"step": 5010
|
| 3525 |
+
},
|
| 3526 |
+
{
|
| 3527 |
+
"epoch": 0.6113098409315777,
|
| 3528 |
+
"grad_norm": 349.25,
|
| 3529 |
+
"learning_rate": 9.904483382582109e-06,
|
| 3530 |
+
"loss": 22.3448,
|
| 3531 |
+
"step": 5020
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 0.6125275896186925,
|
| 3535 |
+
"grad_norm": 481.5,
|
| 3536 |
+
"learning_rate": 9.904293110435858e-06,
|
| 3537 |
+
"loss": 22.3357,
|
| 3538 |
+
"step": 5030
|
| 3539 |
+
},
|
| 3540 |
+
{
|
| 3541 |
+
"epoch": 0.6137453383058071,
|
| 3542 |
+
"grad_norm": 583.0,
|
| 3543 |
+
"learning_rate": 9.904102838289607e-06,
|
| 3544 |
+
"loss": 22.5857,
|
| 3545 |
+
"step": 5040
|
| 3546 |
+
},
|
| 3547 |
+
{
|
| 3548 |
+
"epoch": 0.6149630869929218,
|
| 3549 |
+
"grad_norm": 388.0,
|
| 3550 |
+
"learning_rate": 9.903912566143355e-06,
|
| 3551 |
+
"loss": 22.3212,
|
| 3552 |
+
"step": 5050
|
| 3553 |
+
},
|
| 3554 |
+
{
|
| 3555 |
+
"epoch": 0.6161808356800366,
|
| 3556 |
+
"grad_norm": 305.0,
|
| 3557 |
+
"learning_rate": 9.903722293997104e-06,
|
| 3558 |
+
"loss": 22.3341,
|
| 3559 |
+
"step": 5060
|
| 3560 |
+
},
|
| 3561 |
+
{
|
| 3562 |
+
"epoch": 0.6173985843671512,
|
| 3563 |
+
"grad_norm": 264.75,
|
| 3564 |
+
"learning_rate": 9.903532021850855e-06,
|
| 3565 |
+
"loss": 22.5816,
|
| 3566 |
+
"step": 5070
|
| 3567 |
+
},
|
| 3568 |
+
{
|
| 3569 |
+
"epoch": 0.6186163330542659,
|
| 3570 |
+
"grad_norm": 311.75,
|
| 3571 |
+
"learning_rate": 9.903341749704604e-06,
|
| 3572 |
+
"loss": 22.5137,
|
| 3573 |
+
"step": 5080
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 0.6198340817413807,
|
| 3577 |
+
"grad_norm": 619.5,
|
| 3578 |
+
"learning_rate": 9.903151477558353e-06,
|
| 3579 |
+
"loss": 22.5041,
|
| 3580 |
+
"step": 5090
|
| 3581 |
+
},
|
| 3582 |
+
{
|
| 3583 |
+
"epoch": 0.6210518304284953,
|
| 3584 |
+
"grad_norm": 398.0,
|
| 3585 |
+
"learning_rate": 9.902961205412102e-06,
|
| 3586 |
+
"loss": 22.4265,
|
| 3587 |
+
"step": 5100
|
| 3588 |
+
},
|
| 3589 |
+
{
|
| 3590 |
+
"epoch": 0.62226957911561,
|
| 3591 |
+
"grad_norm": 456.5,
|
| 3592 |
+
"learning_rate": 9.90277093326585e-06,
|
| 3593 |
+
"loss": 22.3639,
|
| 3594 |
+
"step": 5110
|
| 3595 |
+
},
|
| 3596 |
+
{
|
| 3597 |
+
"epoch": 0.6234873278027248,
|
| 3598 |
+
"grad_norm": 300.0,
|
| 3599 |
+
"learning_rate": 9.902580661119601e-06,
|
| 3600 |
+
"loss": 22.2858,
|
| 3601 |
+
"step": 5120
|
| 3602 |
+
},
|
| 3603 |
+
{
|
| 3604 |
+
"epoch": 0.6247050764898394,
|
| 3605 |
+
"grad_norm": 342.75,
|
| 3606 |
+
"learning_rate": 9.90239038897335e-06,
|
| 3607 |
+
"loss": 22.2856,
|
| 3608 |
+
"step": 5130
|
| 3609 |
+
},
|
| 3610 |
+
{
|
| 3611 |
+
"epoch": 0.6259228251769541,
|
| 3612 |
+
"grad_norm": 679.0,
|
| 3613 |
+
"learning_rate": 9.902200116827099e-06,
|
| 3614 |
+
"loss": 22.4301,
|
| 3615 |
+
"step": 5140
|
| 3616 |
+
},
|
| 3617 |
+
{
|
| 3618 |
+
"epoch": 0.6271405738640689,
|
| 3619 |
+
"grad_norm": 250.75,
|
| 3620 |
+
"learning_rate": 9.902009844680848e-06,
|
| 3621 |
+
"loss": 22.1408,
|
| 3622 |
+
"step": 5150
|
| 3623 |
+
},
|
| 3624 |
+
{
|
| 3625 |
+
"epoch": 0.6283583225511835,
|
| 3626 |
+
"grad_norm": 480.75,
|
| 3627 |
+
"learning_rate": 9.901819572534596e-06,
|
| 3628 |
+
"loss": 22.5274,
|
| 3629 |
+
"step": 5160
|
| 3630 |
+
},
|
| 3631 |
+
{
|
| 3632 |
+
"epoch": 0.6295760712382982,
|
| 3633 |
+
"grad_norm": 495.75,
|
| 3634 |
+
"learning_rate": 9.901629300388347e-06,
|
| 3635 |
+
"loss": 22.3002,
|
| 3636 |
+
"step": 5170
|
| 3637 |
+
},
|
| 3638 |
+
{
|
| 3639 |
+
"epoch": 0.630793819925413,
|
| 3640 |
+
"grad_norm": 370.75,
|
| 3641 |
+
"learning_rate": 9.901439028242096e-06,
|
| 3642 |
+
"loss": 22.5818,
|
| 3643 |
+
"step": 5180
|
| 3644 |
+
},
|
| 3645 |
+
{
|
| 3646 |
+
"epoch": 0.6320115686125276,
|
| 3647 |
+
"grad_norm": 633.0,
|
| 3648 |
+
"learning_rate": 9.901248756095845e-06,
|
| 3649 |
+
"loss": 21.9686,
|
| 3650 |
+
"step": 5190
|
| 3651 |
+
},
|
| 3652 |
+
{
|
| 3653 |
+
"epoch": 0.6332293172996423,
|
| 3654 |
+
"grad_norm": 317.25,
|
| 3655 |
+
"learning_rate": 9.901058483949594e-06,
|
| 3656 |
+
"loss": 22.3873,
|
| 3657 |
+
"step": 5200
|
| 3658 |
+
},
|
| 3659 |
+
{
|
| 3660 |
+
"epoch": 0.6344470659867569,
|
| 3661 |
+
"grad_norm": 508.0,
|
| 3662 |
+
"learning_rate": 9.900868211803342e-06,
|
| 3663 |
+
"loss": 22.2581,
|
| 3664 |
+
"step": 5210
|
| 3665 |
+
},
|
| 3666 |
+
{
|
| 3667 |
+
"epoch": 0.6356648146738717,
|
| 3668 |
+
"grad_norm": 443.0,
|
| 3669 |
+
"learning_rate": 9.900677939657093e-06,
|
| 3670 |
+
"loss": 22.2598,
|
| 3671 |
+
"step": 5220
|
| 3672 |
+
},
|
| 3673 |
+
{
|
| 3674 |
+
"epoch": 0.6368825633609864,
|
| 3675 |
+
"grad_norm": 540.5,
|
| 3676 |
+
"learning_rate": 9.900487667510842e-06,
|
| 3677 |
+
"loss": 22.0666,
|
| 3678 |
+
"step": 5230
|
| 3679 |
+
},
|
| 3680 |
+
{
|
| 3681 |
+
"epoch": 0.638100312048101,
|
| 3682 |
+
"grad_norm": 499.0,
|
| 3683 |
+
"learning_rate": 9.90029739536459e-06,
|
| 3684 |
+
"loss": 22.5543,
|
| 3685 |
+
"step": 5240
|
| 3686 |
+
},
|
| 3687 |
+
{
|
| 3688 |
+
"epoch": 0.6393180607352158,
|
| 3689 |
+
"grad_norm": 410.25,
|
| 3690 |
+
"learning_rate": 9.90010712321834e-06,
|
| 3691 |
+
"loss": 22.2988,
|
| 3692 |
+
"step": 5250
|
| 3693 |
+
},
|
| 3694 |
+
{
|
| 3695 |
+
"epoch": 0.6405358094223305,
|
| 3696 |
+
"grad_norm": 316.25,
|
| 3697 |
+
"learning_rate": 9.899916851072089e-06,
|
| 3698 |
+
"loss": 22.3868,
|
| 3699 |
+
"step": 5260
|
| 3700 |
+
},
|
| 3701 |
+
{
|
| 3702 |
+
"epoch": 0.6417535581094451,
|
| 3703 |
+
"grad_norm": 310.25,
|
| 3704 |
+
"learning_rate": 9.899726578925837e-06,
|
| 3705 |
+
"loss": 22.3954,
|
| 3706 |
+
"step": 5270
|
| 3707 |
+
},
|
| 3708 |
+
{
|
| 3709 |
+
"epoch": 0.6429713067965599,
|
| 3710 |
+
"grad_norm": 303.5,
|
| 3711 |
+
"learning_rate": 9.899536306779588e-06,
|
| 3712 |
+
"loss": 21.8775,
|
| 3713 |
+
"step": 5280
|
| 3714 |
+
},
|
| 3715 |
+
{
|
| 3716 |
+
"epoch": 0.6441890554836746,
|
| 3717 |
+
"grad_norm": 326.25,
|
| 3718 |
+
"learning_rate": 9.899346034633337e-06,
|
| 3719 |
+
"loss": 22.2514,
|
| 3720 |
+
"step": 5290
|
| 3721 |
+
},
|
| 3722 |
+
{
|
| 3723 |
+
"epoch": 0.6454068041707892,
|
| 3724 |
+
"grad_norm": 408.75,
|
| 3725 |
+
"learning_rate": 9.899155762487086e-06,
|
| 3726 |
+
"loss": 22.3362,
|
| 3727 |
+
"step": 5300
|
| 3728 |
+
},
|
| 3729 |
+
{
|
| 3730 |
+
"epoch": 0.646624552857904,
|
| 3731 |
+
"grad_norm": 364.25,
|
| 3732 |
+
"learning_rate": 9.898965490340835e-06,
|
| 3733 |
+
"loss": 22.2638,
|
| 3734 |
+
"step": 5310
|
| 3735 |
+
},
|
| 3736 |
+
{
|
| 3737 |
+
"epoch": 0.6478423015450187,
|
| 3738 |
+
"grad_norm": 468.0,
|
| 3739 |
+
"learning_rate": 9.898775218194583e-06,
|
| 3740 |
+
"loss": 22.3181,
|
| 3741 |
+
"step": 5320
|
| 3742 |
+
},
|
| 3743 |
+
{
|
| 3744 |
+
"epoch": 0.6490600502321333,
|
| 3745 |
+
"grad_norm": 464.25,
|
| 3746 |
+
"learning_rate": 9.898584946048334e-06,
|
| 3747 |
+
"loss": 22.1543,
|
| 3748 |
+
"step": 5330
|
| 3749 |
+
},
|
| 3750 |
+
{
|
| 3751 |
+
"epoch": 0.6502777989192481,
|
| 3752 |
+
"grad_norm": 550.5,
|
| 3753 |
+
"learning_rate": 9.898394673902083e-06,
|
| 3754 |
+
"loss": 22.3125,
|
| 3755 |
+
"step": 5340
|
| 3756 |
+
},
|
| 3757 |
+
{
|
| 3758 |
+
"epoch": 0.6514955476063627,
|
| 3759 |
+
"grad_norm": 721.5,
|
| 3760 |
+
"learning_rate": 9.898204401755832e-06,
|
| 3761 |
+
"loss": 22.4355,
|
| 3762 |
+
"step": 5350
|
| 3763 |
+
},
|
| 3764 |
+
{
|
| 3765 |
+
"epoch": 0.6527132962934774,
|
| 3766 |
+
"grad_norm": 878.5,
|
| 3767 |
+
"learning_rate": 9.89801412960958e-06,
|
| 3768 |
+
"loss": 22.2877,
|
| 3769 |
+
"step": 5360
|
| 3770 |
+
},
|
| 3771 |
+
{
|
| 3772 |
+
"epoch": 0.6539310449805922,
|
| 3773 |
+
"grad_norm": 378.25,
|
| 3774 |
+
"learning_rate": 9.89782385746333e-06,
|
| 3775 |
+
"loss": 22.2965,
|
| 3776 |
+
"step": 5370
|
| 3777 |
+
},
|
| 3778 |
+
{
|
| 3779 |
+
"epoch": 0.6551487936677068,
|
| 3780 |
+
"grad_norm": 444.0,
|
| 3781 |
+
"learning_rate": 9.89763358531708e-06,
|
| 3782 |
+
"loss": 22.4212,
|
| 3783 |
+
"step": 5380
|
| 3784 |
+
},
|
| 3785 |
+
{
|
| 3786 |
+
"epoch": 0.6563665423548215,
|
| 3787 |
+
"grad_norm": 1159.0,
|
| 3788 |
+
"learning_rate": 9.897443313170829e-06,
|
| 3789 |
+
"loss": 22.283,
|
| 3790 |
+
"step": 5390
|
| 3791 |
+
},
|
| 3792 |
+
{
|
| 3793 |
+
"epoch": 0.6575842910419363,
|
| 3794 |
+
"grad_norm": 463.25,
|
| 3795 |
+
"learning_rate": 9.897253041024578e-06,
|
| 3796 |
+
"loss": 22.3946,
|
| 3797 |
+
"step": 5400
|
| 3798 |
+
},
|
| 3799 |
+
{
|
| 3800 |
+
"epoch": 0.6588020397290509,
|
| 3801 |
+
"grad_norm": 508.5,
|
| 3802 |
+
"learning_rate": 9.897062768878327e-06,
|
| 3803 |
+
"loss": 22.2301,
|
| 3804 |
+
"step": 5410
|
| 3805 |
+
},
|
| 3806 |
+
{
|
| 3807 |
+
"epoch": 0.6600197884161656,
|
| 3808 |
+
"grad_norm": 330.0,
|
| 3809 |
+
"learning_rate": 9.896872496732076e-06,
|
| 3810 |
+
"loss": 22.1951,
|
| 3811 |
+
"step": 5420
|
| 3812 |
+
},
|
| 3813 |
+
{
|
| 3814 |
+
"epoch": 0.6612375371032803,
|
| 3815 |
+
"grad_norm": 509.5,
|
| 3816 |
+
"learning_rate": 9.896682224585826e-06,
|
| 3817 |
+
"loss": 22.2835,
|
| 3818 |
+
"step": 5430
|
| 3819 |
+
},
|
| 3820 |
+
{
|
| 3821 |
+
"epoch": 0.662455285790395,
|
| 3822 |
+
"grad_norm": 983.0,
|
| 3823 |
+
"learning_rate": 9.896491952439575e-06,
|
| 3824 |
+
"loss": 22.0716,
|
| 3825 |
+
"step": 5440
|
| 3826 |
+
},
|
| 3827 |
+
{
|
| 3828 |
+
"epoch": 0.6636730344775097,
|
| 3829 |
+
"grad_norm": 367.25,
|
| 3830 |
+
"learning_rate": 9.896301680293324e-06,
|
| 3831 |
+
"loss": 22.0852,
|
| 3832 |
+
"step": 5450
|
| 3833 |
+
},
|
| 3834 |
+
{
|
| 3835 |
+
"epoch": 0.6648907831646244,
|
| 3836 |
+
"grad_norm": 320.0,
|
| 3837 |
+
"learning_rate": 9.896111408147073e-06,
|
| 3838 |
+
"loss": 22.0447,
|
| 3839 |
+
"step": 5460
|
| 3840 |
+
},
|
| 3841 |
+
{
|
| 3842 |
+
"epoch": 0.6661085318517391,
|
| 3843 |
+
"grad_norm": 443.25,
|
| 3844 |
+
"learning_rate": 9.895921136000822e-06,
|
| 3845 |
+
"loss": 22.1223,
|
| 3846 |
+
"step": 5470
|
| 3847 |
+
},
|
| 3848 |
+
{
|
| 3849 |
+
"epoch": 0.6673262805388538,
|
| 3850 |
+
"grad_norm": 737.0,
|
| 3851 |
+
"learning_rate": 9.895730863854572e-06,
|
| 3852 |
+
"loss": 22.3605,
|
| 3853 |
+
"step": 5480
|
| 3854 |
+
},
|
| 3855 |
+
{
|
| 3856 |
+
"epoch": 0.6685440292259685,
|
| 3857 |
+
"grad_norm": 370.25,
|
| 3858 |
+
"learning_rate": 9.895540591708321e-06,
|
| 3859 |
+
"loss": 22.3988,
|
| 3860 |
+
"step": 5490
|
| 3861 |
+
},
|
| 3862 |
+
{
|
| 3863 |
+
"epoch": 0.6697617779130832,
|
| 3864 |
+
"grad_norm": 445.25,
|
| 3865 |
+
"learning_rate": 9.89535031956207e-06,
|
| 3866 |
+
"loss": 22.2857,
|
| 3867 |
+
"step": 5500
|
| 3868 |
+
},
|
| 3869 |
+
{
|
| 3870 |
+
"epoch": 0.6709795266001979,
|
| 3871 |
+
"grad_norm": 376.5,
|
| 3872 |
+
"learning_rate": 9.895160047415819e-06,
|
| 3873 |
+
"loss": 22.3568,
|
| 3874 |
+
"step": 5510
|
| 3875 |
+
},
|
| 3876 |
+
{
|
| 3877 |
+
"epoch": 0.6721972752873125,
|
| 3878 |
+
"grad_norm": 570.0,
|
| 3879 |
+
"learning_rate": 9.894969775269568e-06,
|
| 3880 |
+
"loss": 22.2735,
|
| 3881 |
+
"step": 5520
|
| 3882 |
+
},
|
| 3883 |
+
{
|
| 3884 |
+
"epoch": 0.6734150239744273,
|
| 3885 |
+
"grad_norm": 336.25,
|
| 3886 |
+
"learning_rate": 9.894779503123318e-06,
|
| 3887 |
+
"loss": 22.1921,
|
| 3888 |
+
"step": 5530
|
| 3889 |
+
},
|
| 3890 |
+
{
|
| 3891 |
+
"epoch": 0.674632772661542,
|
| 3892 |
+
"grad_norm": 321.75,
|
| 3893 |
+
"learning_rate": 9.894589230977067e-06,
|
| 3894 |
+
"loss": 22.1618,
|
| 3895 |
+
"step": 5540
|
| 3896 |
+
},
|
| 3897 |
+
{
|
| 3898 |
+
"epoch": 0.6758505213486566,
|
| 3899 |
+
"grad_norm": 397.5,
|
| 3900 |
+
"learning_rate": 9.894398958830816e-06,
|
| 3901 |
+
"loss": 22.183,
|
| 3902 |
+
"step": 5550
|
| 3903 |
+
},
|
| 3904 |
+
{
|
| 3905 |
+
"epoch": 0.6770682700357714,
|
| 3906 |
+
"grad_norm": 389.75,
|
| 3907 |
+
"learning_rate": 9.894208686684565e-06,
|
| 3908 |
+
"loss": 22.3434,
|
| 3909 |
+
"step": 5560
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 0.6782860187228861,
|
| 3913 |
+
"grad_norm": 353.0,
|
| 3914 |
+
"learning_rate": 9.894018414538314e-06,
|
| 3915 |
+
"loss": 22.3097,
|
| 3916 |
+
"step": 5570
|
| 3917 |
+
},
|
| 3918 |
+
{
|
| 3919 |
+
"epoch": 0.6795037674100007,
|
| 3920 |
+
"grad_norm": 637.5,
|
| 3921 |
+
"learning_rate": 9.893828142392064e-06,
|
| 3922 |
+
"loss": 22.36,
|
| 3923 |
+
"step": 5580
|
| 3924 |
+
},
|
| 3925 |
+
{
|
| 3926 |
+
"epoch": 0.6807215160971155,
|
| 3927 |
+
"grad_norm": 800.5,
|
| 3928 |
+
"learning_rate": 9.893637870245813e-06,
|
| 3929 |
+
"loss": 22.2326,
|
| 3930 |
+
"step": 5590
|
| 3931 |
+
},
|
| 3932 |
+
{
|
| 3933 |
+
"epoch": 0.6819392647842302,
|
| 3934 |
+
"grad_norm": 381.75,
|
| 3935 |
+
"learning_rate": 9.893447598099562e-06,
|
| 3936 |
+
"loss": 22.0298,
|
| 3937 |
+
"step": 5600
|
| 3938 |
+
},
|
| 3939 |
+
{
|
| 3940 |
+
"epoch": 0.6831570134713448,
|
| 3941 |
+
"grad_norm": 405.75,
|
| 3942 |
+
"learning_rate": 9.893257325953311e-06,
|
| 3943 |
+
"loss": 22.0199,
|
| 3944 |
+
"step": 5610
|
| 3945 |
+
},
|
| 3946 |
+
{
|
| 3947 |
+
"epoch": 0.6843747621584595,
|
| 3948 |
+
"grad_norm": 803.5,
|
| 3949 |
+
"learning_rate": 9.89306705380706e-06,
|
| 3950 |
+
"loss": 21.978,
|
| 3951 |
+
"step": 5620
|
| 3952 |
+
},
|
| 3953 |
+
{
|
| 3954 |
+
"epoch": 0.6855925108455743,
|
| 3955 |
+
"grad_norm": 345.5,
|
| 3956 |
+
"learning_rate": 9.89287678166081e-06,
|
| 3957 |
+
"loss": 22.1337,
|
| 3958 |
+
"step": 5630
|
| 3959 |
+
},
|
| 3960 |
+
{
|
| 3961 |
+
"epoch": 0.6868102595326889,
|
| 3962 |
+
"grad_norm": 505.5,
|
| 3963 |
+
"learning_rate": 9.89268650951456e-06,
|
| 3964 |
+
"loss": 22.1635,
|
| 3965 |
+
"step": 5640
|
| 3966 |
+
},
|
| 3967 |
+
{
|
| 3968 |
+
"epoch": 0.6880280082198036,
|
| 3969 |
+
"grad_norm": 328.75,
|
| 3970 |
+
"learning_rate": 9.892496237368308e-06,
|
| 3971 |
+
"loss": 22.2217,
|
| 3972 |
+
"step": 5650
|
| 3973 |
+
},
|
| 3974 |
+
{
|
| 3975 |
+
"epoch": 0.6892457569069184,
|
| 3976 |
+
"grad_norm": 267.75,
|
| 3977 |
+
"learning_rate": 9.892305965222057e-06,
|
| 3978 |
+
"loss": 21.8837,
|
| 3979 |
+
"step": 5660
|
| 3980 |
+
},
|
| 3981 |
+
{
|
| 3982 |
+
"epoch": 0.690463505594033,
|
| 3983 |
+
"grad_norm": 377.5,
|
| 3984 |
+
"learning_rate": 9.892115693075806e-06,
|
| 3985 |
+
"loss": 22.2993,
|
| 3986 |
+
"step": 5670
|
| 3987 |
+
},
|
| 3988 |
+
{
|
| 3989 |
+
"epoch": 0.6916812542811477,
|
| 3990 |
+
"grad_norm": 419.5,
|
| 3991 |
+
"learning_rate": 9.891925420929556e-06,
|
| 3992 |
+
"loss": 22.2408,
|
| 3993 |
+
"step": 5680
|
| 3994 |
+
},
|
| 3995 |
+
{
|
| 3996 |
+
"epoch": 0.6928990029682625,
|
| 3997 |
+
"grad_norm": 353.25,
|
| 3998 |
+
"learning_rate": 9.891735148783305e-06,
|
| 3999 |
+
"loss": 22.3364,
|
| 4000 |
+
"step": 5690
|
| 4001 |
+
},
|
| 4002 |
+
{
|
| 4003 |
+
"epoch": 0.6941167516553771,
|
| 4004 |
+
"grad_norm": 337.5,
|
| 4005 |
+
"learning_rate": 9.891544876637054e-06,
|
| 4006 |
+
"loss": 22.2121,
|
| 4007 |
+
"step": 5700
|
| 4008 |
+
},
|
| 4009 |
+
{
|
| 4010 |
+
"epoch": 0.6953345003424918,
|
| 4011 |
+
"grad_norm": 381.5,
|
| 4012 |
+
"learning_rate": 9.891354604490803e-06,
|
| 4013 |
+
"loss": 22.3864,
|
| 4014 |
+
"step": 5710
|
| 4015 |
+
},
|
| 4016 |
+
{
|
| 4017 |
+
"epoch": 0.6965522490296066,
|
| 4018 |
+
"grad_norm": 311.5,
|
| 4019 |
+
"learning_rate": 9.891164332344554e-06,
|
| 4020 |
+
"loss": 22.0068,
|
| 4021 |
+
"step": 5720
|
| 4022 |
+
},
|
| 4023 |
+
{
|
| 4024 |
+
"epoch": 0.6977699977167212,
|
| 4025 |
+
"grad_norm": 364.75,
|
| 4026 |
+
"learning_rate": 9.890974060198302e-06,
|
| 4027 |
+
"loss": 22.2089,
|
| 4028 |
+
"step": 5730
|
| 4029 |
+
},
|
| 4030 |
+
{
|
| 4031 |
+
"epoch": 0.6989877464038359,
|
| 4032 |
+
"grad_norm": 469.75,
|
| 4033 |
+
"learning_rate": 9.890783788052051e-06,
|
| 4034 |
+
"loss": 22.3107,
|
| 4035 |
+
"step": 5740
|
| 4036 |
+
},
|
| 4037 |
+
{
|
| 4038 |
+
"epoch": 0.7002054950909506,
|
| 4039 |
+
"grad_norm": 320.0,
|
| 4040 |
+
"learning_rate": 9.8905935159058e-06,
|
| 4041 |
+
"loss": 22.4451,
|
| 4042 |
+
"step": 5750
|
| 4043 |
+
},
|
| 4044 |
+
{
|
| 4045 |
+
"epoch": 0.7014232437780653,
|
| 4046 |
+
"grad_norm": 394.25,
|
| 4047 |
+
"learning_rate": 9.890403243759549e-06,
|
| 4048 |
+
"loss": 22.3636,
|
| 4049 |
+
"step": 5760
|
| 4050 |
+
},
|
| 4051 |
+
{
|
| 4052 |
+
"epoch": 0.70264099246518,
|
| 4053 |
+
"grad_norm": 348.25,
|
| 4054 |
+
"learning_rate": 9.8902129716133e-06,
|
| 4055 |
+
"loss": 22.2534,
|
| 4056 |
+
"step": 5770
|
| 4057 |
+
},
|
| 4058 |
+
{
|
| 4059 |
+
"epoch": 0.7038587411522947,
|
| 4060 |
+
"grad_norm": 396.75,
|
| 4061 |
+
"learning_rate": 9.890022699467048e-06,
|
| 4062 |
+
"loss": 22.1951,
|
| 4063 |
+
"step": 5780
|
| 4064 |
+
},
|
| 4065 |
+
{
|
| 4066 |
+
"epoch": 0.7050764898394094,
|
| 4067 |
+
"grad_norm": 388.0,
|
| 4068 |
+
"learning_rate": 9.889832427320797e-06,
|
| 4069 |
+
"loss": 22.3651,
|
| 4070 |
+
"step": 5790
|
| 4071 |
+
},
|
| 4072 |
+
{
|
| 4073 |
+
"epoch": 0.7062942385265241,
|
| 4074 |
+
"grad_norm": 490.75,
|
| 4075 |
+
"learning_rate": 9.889642155174546e-06,
|
| 4076 |
+
"loss": 22.3609,
|
| 4077 |
+
"step": 5800
|
| 4078 |
+
},
|
| 4079 |
+
{
|
| 4080 |
+
"epoch": 0.7075119872136388,
|
| 4081 |
+
"grad_norm": 390.75,
|
| 4082 |
+
"learning_rate": 9.889451883028295e-06,
|
| 4083 |
+
"loss": 22.1719,
|
| 4084 |
+
"step": 5810
|
| 4085 |
+
},
|
| 4086 |
+
{
|
| 4087 |
+
"epoch": 0.7087297359007535,
|
| 4088 |
+
"grad_norm": 569.5,
|
| 4089 |
+
"learning_rate": 9.889261610882046e-06,
|
| 4090 |
+
"loss": 22.3145,
|
| 4091 |
+
"step": 5820
|
| 4092 |
+
},
|
| 4093 |
+
{
|
| 4094 |
+
"epoch": 0.7099474845878682,
|
| 4095 |
+
"grad_norm": 395.25,
|
| 4096 |
+
"learning_rate": 9.889071338735795e-06,
|
| 4097 |
+
"loss": 22.0284,
|
| 4098 |
+
"step": 5830
|
| 4099 |
+
},
|
| 4100 |
+
{
|
| 4101 |
+
"epoch": 0.7111652332749828,
|
| 4102 |
+
"grad_norm": 383.25,
|
| 4103 |
+
"learning_rate": 9.888881066589543e-06,
|
| 4104 |
+
"loss": 22.1514,
|
| 4105 |
+
"step": 5840
|
| 4106 |
+
},
|
| 4107 |
+
{
|
| 4108 |
+
"epoch": 0.7123829819620976,
|
| 4109 |
+
"grad_norm": 439.75,
|
| 4110 |
+
"learning_rate": 9.888690794443292e-06,
|
| 4111 |
+
"loss": 22.2325,
|
| 4112 |
+
"step": 5850
|
| 4113 |
+
},
|
| 4114 |
+
{
|
| 4115 |
+
"epoch": 0.7136007306492123,
|
| 4116 |
+
"grad_norm": 419.25,
|
| 4117 |
+
"learning_rate": 9.888500522297043e-06,
|
| 4118 |
+
"loss": 22.072,
|
| 4119 |
+
"step": 5860
|
| 4120 |
+
},
|
| 4121 |
+
{
|
| 4122 |
+
"epoch": 0.7148184793363269,
|
| 4123 |
+
"grad_norm": 375.0,
|
| 4124 |
+
"learning_rate": 9.888310250150792e-06,
|
| 4125 |
+
"loss": 22.3628,
|
| 4126 |
+
"step": 5870
|
| 4127 |
+
},
|
| 4128 |
+
{
|
| 4129 |
+
"epoch": 0.7160362280234417,
|
| 4130 |
+
"grad_norm": 429.0,
|
| 4131 |
+
"learning_rate": 9.88811997800454e-06,
|
| 4132 |
+
"loss": 22.0338,
|
| 4133 |
+
"step": 5880
|
| 4134 |
+
},
|
| 4135 |
+
{
|
| 4136 |
+
"epoch": 0.7172539767105564,
|
| 4137 |
+
"grad_norm": 291.5,
|
| 4138 |
+
"learning_rate": 9.88792970585829e-06,
|
| 4139 |
+
"loss": 22.4026,
|
| 4140 |
+
"step": 5890
|
| 4141 |
+
},
|
| 4142 |
+
{
|
| 4143 |
+
"epoch": 0.718471725397671,
|
| 4144 |
+
"grad_norm": 672.5,
|
| 4145 |
+
"learning_rate": 9.887739433712038e-06,
|
| 4146 |
+
"loss": 22.1481,
|
| 4147 |
+
"step": 5900
|
| 4148 |
+
},
|
| 4149 |
+
{
|
| 4150 |
+
"epoch": 0.7196894740847858,
|
| 4151 |
+
"grad_norm": 340.0,
|
| 4152 |
+
"learning_rate": 9.887549161565789e-06,
|
| 4153 |
+
"loss": 22.348,
|
| 4154 |
+
"step": 5910
|
| 4155 |
+
},
|
| 4156 |
+
{
|
| 4157 |
+
"epoch": 0.7209072227719004,
|
| 4158 |
+
"grad_norm": 395.75,
|
| 4159 |
+
"learning_rate": 9.887358889419538e-06,
|
| 4160 |
+
"loss": 22.2665,
|
| 4161 |
+
"step": 5920
|
| 4162 |
+
},
|
| 4163 |
+
{
|
| 4164 |
+
"epoch": 0.7221249714590151,
|
| 4165 |
+
"grad_norm": 516.5,
|
| 4166 |
+
"learning_rate": 9.887168617273287e-06,
|
| 4167 |
+
"loss": 22.2479,
|
| 4168 |
+
"step": 5930
|
| 4169 |
+
},
|
| 4170 |
+
{
|
| 4171 |
+
"epoch": 0.7233427201461299,
|
| 4172 |
+
"grad_norm": 498.0,
|
| 4173 |
+
"learning_rate": 9.886978345127036e-06,
|
| 4174 |
+
"loss": 22.2421,
|
| 4175 |
+
"step": 5940
|
| 4176 |
+
},
|
| 4177 |
+
{
|
| 4178 |
+
"epoch": 0.7245604688332445,
|
| 4179 |
+
"grad_norm": 450.5,
|
| 4180 |
+
"learning_rate": 9.886788072980784e-06,
|
| 4181 |
+
"loss": 22.3301,
|
| 4182 |
+
"step": 5950
|
| 4183 |
+
},
|
| 4184 |
+
{
|
| 4185 |
+
"epoch": 0.7257782175203592,
|
| 4186 |
+
"grad_norm": 701.5,
|
| 4187 |
+
"learning_rate": 9.886597800834535e-06,
|
| 4188 |
+
"loss": 21.9416,
|
| 4189 |
+
"step": 5960
|
| 4190 |
+
},
|
| 4191 |
+
{
|
| 4192 |
+
"epoch": 0.726995966207474,
|
| 4193 |
+
"grad_norm": 293.0,
|
| 4194 |
+
"learning_rate": 9.886407528688284e-06,
|
| 4195 |
+
"loss": 22.1706,
|
| 4196 |
+
"step": 5970
|
| 4197 |
+
},
|
| 4198 |
+
{
|
| 4199 |
+
"epoch": 0.7282137148945886,
|
| 4200 |
+
"grad_norm": 359.75,
|
| 4201 |
+
"learning_rate": 9.886217256542033e-06,
|
| 4202 |
+
"loss": 22.4014,
|
| 4203 |
+
"step": 5980
|
| 4204 |
+
},
|
| 4205 |
+
{
|
| 4206 |
+
"epoch": 0.7294314635817033,
|
| 4207 |
+
"grad_norm": 308.0,
|
| 4208 |
+
"learning_rate": 9.886026984395782e-06,
|
| 4209 |
+
"loss": 22.3196,
|
| 4210 |
+
"step": 5990
|
| 4211 |
+
},
|
| 4212 |
+
{
|
| 4213 |
+
"epoch": 0.7306492122688181,
|
| 4214 |
+
"grad_norm": 248.0,
|
| 4215 |
+
"learning_rate": 9.885836712249532e-06,
|
| 4216 |
+
"loss": 22.0416,
|
| 4217 |
+
"step": 6000
|
| 4218 |
+
},
|
| 4219 |
+
{
|
| 4220 |
+
"epoch": 0.7318669609559327,
|
| 4221 |
+
"grad_norm": 284.5,
|
| 4222 |
+
"learning_rate": 9.885646440103281e-06,
|
| 4223 |
+
"loss": 22.1623,
|
| 4224 |
+
"step": 6010
|
| 4225 |
+
},
|
| 4226 |
+
{
|
| 4227 |
+
"epoch": 0.7330847096430474,
|
| 4228 |
+
"grad_norm": 306.0,
|
| 4229 |
+
"learning_rate": 9.88545616795703e-06,
|
| 4230 |
+
"loss": 22.3356,
|
| 4231 |
+
"step": 6020
|
| 4232 |
+
},
|
| 4233 |
+
{
|
| 4234 |
+
"epoch": 0.7343024583301622,
|
| 4235 |
+
"grad_norm": 583.5,
|
| 4236 |
+
"learning_rate": 9.885265895810779e-06,
|
| 4237 |
+
"loss": 22.2748,
|
| 4238 |
+
"step": 6030
|
| 4239 |
+
},
|
| 4240 |
+
{
|
| 4241 |
+
"epoch": 0.7355202070172768,
|
| 4242 |
+
"grad_norm": 475.75,
|
| 4243 |
+
"learning_rate": 9.885075623664528e-06,
|
| 4244 |
+
"loss": 22.3207,
|
| 4245 |
+
"step": 6040
|
| 4246 |
+
},
|
| 4247 |
+
{
|
| 4248 |
+
"epoch": 0.7367379557043915,
|
| 4249 |
+
"grad_norm": 300.5,
|
| 4250 |
+
"learning_rate": 9.884885351518278e-06,
|
| 4251 |
+
"loss": 22.1761,
|
| 4252 |
+
"step": 6050
|
| 4253 |
+
},
|
| 4254 |
+
{
|
| 4255 |
+
"epoch": 0.7379557043915062,
|
| 4256 |
+
"grad_norm": 280.75,
|
| 4257 |
+
"learning_rate": 9.884695079372027e-06,
|
| 4258 |
+
"loss": 22.1228,
|
| 4259 |
+
"step": 6060
|
| 4260 |
+
},
|
| 4261 |
+
{
|
| 4262 |
+
"epoch": 0.7391734530786209,
|
| 4263 |
+
"grad_norm": 518.5,
|
| 4264 |
+
"learning_rate": 9.884504807225776e-06,
|
| 4265 |
+
"loss": 22.4198,
|
| 4266 |
+
"step": 6070
|
| 4267 |
+
},
|
| 4268 |
+
{
|
| 4269 |
+
"epoch": 0.7403912017657356,
|
| 4270 |
+
"grad_norm": 320.75,
|
| 4271 |
+
"learning_rate": 9.884314535079525e-06,
|
| 4272 |
+
"loss": 22.1976,
|
| 4273 |
+
"step": 6080
|
| 4274 |
+
},
|
| 4275 |
+
{
|
| 4276 |
+
"epoch": 0.7416089504528502,
|
| 4277 |
+
"grad_norm": 369.25,
|
| 4278 |
+
"learning_rate": 9.884124262933275e-06,
|
| 4279 |
+
"loss": 22.2466,
|
| 4280 |
+
"step": 6090
|
| 4281 |
+
},
|
| 4282 |
+
{
|
| 4283 |
+
"epoch": 0.742826699139965,
|
| 4284 |
+
"grad_norm": 288.75,
|
| 4285 |
+
"learning_rate": 9.883933990787024e-06,
|
| 4286 |
+
"loss": 22.3804,
|
| 4287 |
+
"step": 6100
|
| 4288 |
+
},
|
| 4289 |
+
{
|
| 4290 |
+
"epoch": 0.7440444478270797,
|
| 4291 |
+
"grad_norm": 437.25,
|
| 4292 |
+
"learning_rate": 9.883743718640773e-06,
|
| 4293 |
+
"loss": 22.2938,
|
| 4294 |
+
"step": 6110
|
| 4295 |
+
},
|
| 4296 |
+
{
|
| 4297 |
+
"epoch": 0.7452621965141943,
|
| 4298 |
+
"grad_norm": 445.75,
|
| 4299 |
+
"learning_rate": 9.883553446494522e-06,
|
| 4300 |
+
"loss": 22.2249,
|
| 4301 |
+
"step": 6120
|
| 4302 |
+
},
|
| 4303 |
+
{
|
| 4304 |
+
"epoch": 0.7464799452013091,
|
| 4305 |
+
"grad_norm": 444.25,
|
| 4306 |
+
"learning_rate": 9.883363174348271e-06,
|
| 4307 |
+
"loss": 22.423,
|
| 4308 |
+
"step": 6130
|
| 4309 |
+
},
|
| 4310 |
+
{
|
| 4311 |
+
"epoch": 0.7476976938884238,
|
| 4312 |
+
"grad_norm": 576.0,
|
| 4313 |
+
"learning_rate": 9.883172902202021e-06,
|
| 4314 |
+
"loss": 22.2609,
|
| 4315 |
+
"step": 6140
|
| 4316 |
+
},
|
| 4317 |
+
{
|
| 4318 |
+
"epoch": 0.7489154425755384,
|
| 4319 |
+
"grad_norm": 330.5,
|
| 4320 |
+
"learning_rate": 9.88298263005577e-06,
|
| 4321 |
+
"loss": 22.248,
|
| 4322 |
+
"step": 6150
|
| 4323 |
+
},
|
| 4324 |
+
{
|
| 4325 |
+
"epoch": 0.7501331912626532,
|
| 4326 |
+
"grad_norm": 716.0,
|
| 4327 |
+
"learning_rate": 9.88279235790952e-06,
|
| 4328 |
+
"loss": 22.1307,
|
| 4329 |
+
"step": 6160
|
| 4330 |
+
},
|
| 4331 |
+
{
|
| 4332 |
+
"epoch": 0.7513509399497679,
|
| 4333 |
+
"grad_norm": 265.5,
|
| 4334 |
+
"learning_rate": 9.882602085763268e-06,
|
| 4335 |
+
"loss": 22.2934,
|
| 4336 |
+
"step": 6170
|
| 4337 |
+
},
|
| 4338 |
+
{
|
| 4339 |
+
"epoch": 0.7525686886368825,
|
| 4340 |
+
"grad_norm": 325.25,
|
| 4341 |
+
"learning_rate": 9.882411813617017e-06,
|
| 4342 |
+
"loss": 21.9465,
|
| 4343 |
+
"step": 6180
|
| 4344 |
+
},
|
| 4345 |
+
{
|
| 4346 |
+
"epoch": 0.7537864373239973,
|
| 4347 |
+
"grad_norm": 545.5,
|
| 4348 |
+
"learning_rate": 9.882221541470767e-06,
|
| 4349 |
+
"loss": 22.468,
|
| 4350 |
+
"step": 6190
|
| 4351 |
+
},
|
| 4352 |
+
{
|
| 4353 |
+
"epoch": 0.755004186011112,
|
| 4354 |
+
"grad_norm": 289.25,
|
| 4355 |
+
"learning_rate": 9.882031269324516e-06,
|
| 4356 |
+
"loss": 22.2477,
|
| 4357 |
+
"step": 6200
|
| 4358 |
+
},
|
| 4359 |
+
{
|
| 4360 |
+
"epoch": 0.7562219346982266,
|
| 4361 |
+
"grad_norm": 456.75,
|
| 4362 |
+
"learning_rate": 9.881840997178265e-06,
|
| 4363 |
+
"loss": 22.4299,
|
| 4364 |
+
"step": 6210
|
| 4365 |
+
},
|
| 4366 |
+
{
|
| 4367 |
+
"epoch": 0.7574396833853414,
|
| 4368 |
+
"grad_norm": 340.25,
|
| 4369 |
+
"learning_rate": 9.881650725032014e-06,
|
| 4370 |
+
"loss": 21.9778,
|
| 4371 |
+
"step": 6220
|
| 4372 |
+
},
|
| 4373 |
+
{
|
| 4374 |
+
"epoch": 0.7586574320724561,
|
| 4375 |
+
"grad_norm": 375.5,
|
| 4376 |
+
"learning_rate": 9.881460452885765e-06,
|
| 4377 |
+
"loss": 22.3145,
|
| 4378 |
+
"step": 6230
|
| 4379 |
+
},
|
| 4380 |
+
{
|
| 4381 |
+
"epoch": 0.7598751807595707,
|
| 4382 |
+
"grad_norm": 342.5,
|
| 4383 |
+
"learning_rate": 9.881270180739514e-06,
|
| 4384 |
+
"loss": 22.521,
|
| 4385 |
+
"step": 6240
|
| 4386 |
+
},
|
| 4387 |
+
{
|
| 4388 |
+
"epoch": 0.7610929294466855,
|
| 4389 |
+
"grad_norm": 405.25,
|
| 4390 |
+
"learning_rate": 9.881079908593262e-06,
|
| 4391 |
+
"loss": 22.0611,
|
| 4392 |
+
"step": 6250
|
| 4393 |
+
},
|
| 4394 |
+
{
|
| 4395 |
+
"epoch": 0.7623106781338002,
|
| 4396 |
+
"grad_norm": 383.5,
|
| 4397 |
+
"learning_rate": 9.880889636447011e-06,
|
| 4398 |
+
"loss": 22.0493,
|
| 4399 |
+
"step": 6260
|
| 4400 |
+
},
|
| 4401 |
+
{
|
| 4402 |
+
"epoch": 0.7635284268209148,
|
| 4403 |
+
"grad_norm": 406.25,
|
| 4404 |
+
"learning_rate": 9.88069936430076e-06,
|
| 4405 |
+
"loss": 22.1901,
|
| 4406 |
+
"step": 6270
|
| 4407 |
+
},
|
| 4408 |
+
{
|
| 4409 |
+
"epoch": 0.7647461755080295,
|
| 4410 |
+
"grad_norm": 363.5,
|
| 4411 |
+
"learning_rate": 9.88050909215451e-06,
|
| 4412 |
+
"loss": 22.2588,
|
| 4413 |
+
"step": 6280
|
| 4414 |
+
},
|
| 4415 |
+
{
|
| 4416 |
+
"epoch": 0.7659639241951443,
|
| 4417 |
+
"grad_norm": 303.25,
|
| 4418 |
+
"learning_rate": 9.88031882000826e-06,
|
| 4419 |
+
"loss": 22.2292,
|
| 4420 |
+
"step": 6290
|
| 4421 |
+
},
|
| 4422 |
+
{
|
| 4423 |
+
"epoch": 0.7671816728822589,
|
| 4424 |
+
"grad_norm": 315.5,
|
| 4425 |
+
"learning_rate": 9.880128547862008e-06,
|
| 4426 |
+
"loss": 22.1524,
|
| 4427 |
+
"step": 6300
|
| 4428 |
+
},
|
| 4429 |
+
{
|
| 4430 |
+
"epoch": 0.7683994215693736,
|
| 4431 |
+
"grad_norm": 343.5,
|
| 4432 |
+
"learning_rate": 9.879938275715757e-06,
|
| 4433 |
+
"loss": 22.0444,
|
| 4434 |
+
"step": 6310
|
| 4435 |
+
},
|
| 4436 |
+
{
|
| 4437 |
+
"epoch": 0.7696171702564883,
|
| 4438 |
+
"grad_norm": 434.25,
|
| 4439 |
+
"learning_rate": 9.879748003569506e-06,
|
| 4440 |
+
"loss": 21.9586,
|
| 4441 |
+
"step": 6320
|
| 4442 |
+
},
|
| 4443 |
+
{
|
| 4444 |
+
"epoch": 0.770834918943603,
|
| 4445 |
+
"grad_norm": 497.75,
|
| 4446 |
+
"learning_rate": 9.879557731423257e-06,
|
| 4447 |
+
"loss": 22.358,
|
| 4448 |
+
"step": 6330
|
| 4449 |
+
},
|
| 4450 |
+
{
|
| 4451 |
+
"epoch": 0.7720526676307177,
|
| 4452 |
+
"grad_norm": 658.5,
|
| 4453 |
+
"learning_rate": 9.879367459277006e-06,
|
| 4454 |
+
"loss": 22.3363,
|
| 4455 |
+
"step": 6340
|
| 4456 |
+
},
|
| 4457 |
+
{
|
| 4458 |
+
"epoch": 0.7732704163178324,
|
| 4459 |
+
"grad_norm": 536.5,
|
| 4460 |
+
"learning_rate": 9.879177187130755e-06,
|
| 4461 |
+
"loss": 22.0662,
|
| 4462 |
+
"step": 6350
|
| 4463 |
+
},
|
| 4464 |
+
{
|
| 4465 |
+
"epoch": 0.7744881650049471,
|
| 4466 |
+
"grad_norm": 501.0,
|
| 4467 |
+
"learning_rate": 9.878986914984503e-06,
|
| 4468 |
+
"loss": 22.1174,
|
| 4469 |
+
"step": 6360
|
| 4470 |
+
},
|
| 4471 |
+
{
|
| 4472 |
+
"epoch": 0.7757059136920618,
|
| 4473 |
+
"grad_norm": 273.25,
|
| 4474 |
+
"learning_rate": 9.878796642838252e-06,
|
| 4475 |
+
"loss": 22.046,
|
| 4476 |
+
"step": 6370
|
| 4477 |
+
},
|
| 4478 |
+
{
|
| 4479 |
+
"epoch": 0.7769236623791765,
|
| 4480 |
+
"grad_norm": 350.5,
|
| 4481 |
+
"learning_rate": 9.878606370692003e-06,
|
| 4482 |
+
"loss": 22.3518,
|
| 4483 |
+
"step": 6380
|
| 4484 |
+
},
|
| 4485 |
+
{
|
| 4486 |
+
"epoch": 0.7781414110662912,
|
| 4487 |
+
"grad_norm": 330.0,
|
| 4488 |
+
"learning_rate": 9.878416098545752e-06,
|
| 4489 |
+
"loss": 22.2315,
|
| 4490 |
+
"step": 6390
|
| 4491 |
+
},
|
| 4492 |
+
{
|
| 4493 |
+
"epoch": 0.7793591597534059,
|
| 4494 |
+
"grad_norm": 694.5,
|
| 4495 |
+
"learning_rate": 9.8782258263995e-06,
|
| 4496 |
+
"loss": 22.3835,
|
| 4497 |
+
"step": 6400
|
| 4498 |
+
},
|
| 4499 |
+
{
|
| 4500 |
+
"epoch": 0.7805769084405206,
|
| 4501 |
+
"grad_norm": 304.75,
|
| 4502 |
+
"learning_rate": 9.87803555425325e-06,
|
| 4503 |
+
"loss": 22.1816,
|
| 4504 |
+
"step": 6410
|
| 4505 |
+
},
|
| 4506 |
+
{
|
| 4507 |
+
"epoch": 0.7817946571276353,
|
| 4508 |
+
"grad_norm": 351.5,
|
| 4509 |
+
"learning_rate": 9.877845282106998e-06,
|
| 4510 |
+
"loss": 22.1395,
|
| 4511 |
+
"step": 6420
|
| 4512 |
+
},
|
| 4513 |
+
{
|
| 4514 |
+
"epoch": 0.78301240581475,
|
| 4515 |
+
"grad_norm": 308.25,
|
| 4516 |
+
"learning_rate": 9.877655009960747e-06,
|
| 4517 |
+
"loss": 22.0537,
|
| 4518 |
+
"step": 6430
|
| 4519 |
+
},
|
| 4520 |
+
{
|
| 4521 |
+
"epoch": 0.7842301545018647,
|
| 4522 |
+
"grad_norm": 466.25,
|
| 4523 |
+
"learning_rate": 9.877464737814498e-06,
|
| 4524 |
+
"loss": 22.3,
|
| 4525 |
+
"step": 6440
|
| 4526 |
+
},
|
| 4527 |
+
{
|
| 4528 |
+
"epoch": 0.7854479031889794,
|
| 4529 |
+
"grad_norm": 574.5,
|
| 4530 |
+
"learning_rate": 9.877274465668247e-06,
|
| 4531 |
+
"loss": 21.9396,
|
| 4532 |
+
"step": 6450
|
| 4533 |
+
},
|
| 4534 |
+
{
|
| 4535 |
+
"epoch": 0.786665651876094,
|
| 4536 |
+
"grad_norm": 329.5,
|
| 4537 |
+
"learning_rate": 9.877084193521995e-06,
|
| 4538 |
+
"loss": 21.962,
|
| 4539 |
+
"step": 6460
|
| 4540 |
+
},
|
| 4541 |
+
{
|
| 4542 |
+
"epoch": 0.7878834005632087,
|
| 4543 |
+
"grad_norm": 315.0,
|
| 4544 |
+
"learning_rate": 9.876893921375744e-06,
|
| 4545 |
+
"loss": 22.1423,
|
| 4546 |
+
"step": 6470
|
| 4547 |
+
},
|
| 4548 |
+
{
|
| 4549 |
+
"epoch": 0.7891011492503235,
|
| 4550 |
+
"grad_norm": 606.0,
|
| 4551 |
+
"learning_rate": 9.876703649229493e-06,
|
| 4552 |
+
"loss": 22.1091,
|
| 4553 |
+
"step": 6480
|
| 4554 |
+
},
|
| 4555 |
+
{
|
| 4556 |
+
"epoch": 0.7903188979374381,
|
| 4557 |
+
"grad_norm": 325.75,
|
| 4558 |
+
"learning_rate": 9.876513377083244e-06,
|
| 4559 |
+
"loss": 22.2421,
|
| 4560 |
+
"step": 6490
|
| 4561 |
+
},
|
| 4562 |
+
{
|
| 4563 |
+
"epoch": 0.7915366466245528,
|
| 4564 |
+
"grad_norm": 313.25,
|
| 4565 |
+
"learning_rate": 9.876323104936993e-06,
|
| 4566 |
+
"loss": 22.3469,
|
| 4567 |
+
"step": 6500
|
| 4568 |
+
},
|
| 4569 |
+
{
|
| 4570 |
+
"epoch": 0.7927543953116676,
|
| 4571 |
+
"grad_norm": 287.5,
|
| 4572 |
+
"learning_rate": 9.876132832790742e-06,
|
| 4573 |
+
"loss": 22.2427,
|
| 4574 |
+
"step": 6510
|
| 4575 |
+
},
|
| 4576 |
+
{
|
| 4577 |
+
"epoch": 0.7939721439987822,
|
| 4578 |
+
"grad_norm": 409.75,
|
| 4579 |
+
"learning_rate": 9.87594256064449e-06,
|
| 4580 |
+
"loss": 22.3911,
|
| 4581 |
+
"step": 6520
|
| 4582 |
+
},
|
| 4583 |
+
{
|
| 4584 |
+
"epoch": 0.7951898926858969,
|
| 4585 |
+
"grad_norm": 281.0,
|
| 4586 |
+
"learning_rate": 9.87575228849824e-06,
|
| 4587 |
+
"loss": 22.2416,
|
| 4588 |
+
"step": 6530
|
| 4589 |
+
},
|
| 4590 |
+
{
|
| 4591 |
+
"epoch": 0.7964076413730117,
|
| 4592 |
+
"grad_norm": 401.0,
|
| 4593 |
+
"learning_rate": 9.87556201635199e-06,
|
| 4594 |
+
"loss": 22.1661,
|
| 4595 |
+
"step": 6540
|
| 4596 |
+
},
|
| 4597 |
+
{
|
| 4598 |
+
"epoch": 0.7976253900601263,
|
| 4599 |
+
"grad_norm": 763.5,
|
| 4600 |
+
"learning_rate": 9.875371744205739e-06,
|
| 4601 |
+
"loss": 22.3706,
|
| 4602 |
+
"step": 6550
|
| 4603 |
+
},
|
| 4604 |
+
{
|
| 4605 |
+
"epoch": 0.798843138747241,
|
| 4606 |
+
"grad_norm": 351.25,
|
| 4607 |
+
"learning_rate": 9.875181472059488e-06,
|
| 4608 |
+
"loss": 22.1028,
|
| 4609 |
+
"step": 6560
|
| 4610 |
+
},
|
| 4611 |
+
{
|
| 4612 |
+
"epoch": 0.8000608874343558,
|
| 4613 |
+
"grad_norm": 368.25,
|
| 4614 |
+
"learning_rate": 9.874991199913236e-06,
|
| 4615 |
+
"loss": 21.9313,
|
| 4616 |
+
"step": 6570
|
| 4617 |
+
},
|
| 4618 |
+
{
|
| 4619 |
+
"epoch": 0.8012786361214704,
|
| 4620 |
+
"grad_norm": 392.25,
|
| 4621 |
+
"learning_rate": 9.874800927766985e-06,
|
| 4622 |
+
"loss": 22.1766,
|
| 4623 |
+
"step": 6580
|
| 4624 |
+
},
|
| 4625 |
+
{
|
| 4626 |
+
"epoch": 0.8024963848085851,
|
| 4627 |
+
"grad_norm": 541.5,
|
| 4628 |
+
"learning_rate": 9.874610655620736e-06,
|
| 4629 |
+
"loss": 22.1877,
|
| 4630 |
+
"step": 6590
|
| 4631 |
+
},
|
| 4632 |
+
{
|
| 4633 |
+
"epoch": 0.8037141334956999,
|
| 4634 |
+
"grad_norm": 334.5,
|
| 4635 |
+
"learning_rate": 9.874420383474485e-06,
|
| 4636 |
+
"loss": 21.9352,
|
| 4637 |
+
"step": 6600
|
| 4638 |
+
},
|
| 4639 |
+
{
|
| 4640 |
+
"epoch": 0.8049318821828145,
|
| 4641 |
+
"grad_norm": 457.5,
|
| 4642 |
+
"learning_rate": 9.874230111328234e-06,
|
| 4643 |
+
"loss": 22.3417,
|
| 4644 |
+
"step": 6610
|
| 4645 |
+
},
|
| 4646 |
+
{
|
| 4647 |
+
"epoch": 0.8061496308699292,
|
| 4648 |
+
"grad_norm": 300.75,
|
| 4649 |
+
"learning_rate": 9.874039839181983e-06,
|
| 4650 |
+
"loss": 22.295,
|
| 4651 |
+
"step": 6620
|
| 4652 |
+
},
|
| 4653 |
+
{
|
| 4654 |
+
"epoch": 0.807367379557044,
|
| 4655 |
+
"grad_norm": 342.0,
|
| 4656 |
+
"learning_rate": 9.873849567035731e-06,
|
| 4657 |
+
"loss": 22.1422,
|
| 4658 |
+
"step": 6630
|
| 4659 |
+
},
|
| 4660 |
+
{
|
| 4661 |
+
"epoch": 0.8085851282441586,
|
| 4662 |
+
"grad_norm": 329.0,
|
| 4663 |
+
"learning_rate": 9.873659294889482e-06,
|
| 4664 |
+
"loss": 22.3643,
|
| 4665 |
+
"step": 6640
|
| 4666 |
+
},
|
| 4667 |
+
{
|
| 4668 |
+
"epoch": 0.8098028769312733,
|
| 4669 |
+
"grad_norm": 359.5,
|
| 4670 |
+
"learning_rate": 9.87346902274323e-06,
|
| 4671 |
+
"loss": 22.0662,
|
| 4672 |
+
"step": 6650
|
| 4673 |
+
},
|
| 4674 |
+
{
|
| 4675 |
+
"epoch": 0.8110206256183881,
|
| 4676 |
+
"grad_norm": 369.25,
|
| 4677 |
+
"learning_rate": 9.87327875059698e-06,
|
| 4678 |
+
"loss": 22.2069,
|
| 4679 |
+
"step": 6660
|
| 4680 |
+
},
|
| 4681 |
+
{
|
| 4682 |
+
"epoch": 0.8122383743055027,
|
| 4683 |
+
"grad_norm": 415.25,
|
| 4684 |
+
"learning_rate": 9.873088478450729e-06,
|
| 4685 |
+
"loss": 22.0581,
|
| 4686 |
+
"step": 6670
|
| 4687 |
+
},
|
| 4688 |
+
{
|
| 4689 |
+
"epoch": 0.8134561229926174,
|
| 4690 |
+
"grad_norm": 360.25,
|
| 4691 |
+
"learning_rate": 9.872898206304477e-06,
|
| 4692 |
+
"loss": 22.2493,
|
| 4693 |
+
"step": 6680
|
| 4694 |
+
},
|
| 4695 |
+
{
|
| 4696 |
+
"epoch": 0.814673871679732,
|
| 4697 |
+
"grad_norm": 329.0,
|
| 4698 |
+
"learning_rate": 9.872707934158228e-06,
|
| 4699 |
+
"loss": 22.1208,
|
| 4700 |
+
"step": 6690
|
| 4701 |
+
},
|
| 4702 |
+
{
|
| 4703 |
+
"epoch": 0.8158916203668468,
|
| 4704 |
+
"grad_norm": 527.5,
|
| 4705 |
+
"learning_rate": 9.872517662011977e-06,
|
| 4706 |
+
"loss": 22.1305,
|
| 4707 |
+
"step": 6700
|
| 4708 |
+
},
|
| 4709 |
+
{
|
| 4710 |
+
"epoch": 0.8171093690539615,
|
| 4711 |
+
"grad_norm": 753.5,
|
| 4712 |
+
"learning_rate": 9.872327389865726e-06,
|
| 4713 |
+
"loss": 22.2309,
|
| 4714 |
+
"step": 6710
|
| 4715 |
+
},
|
| 4716 |
+
{
|
| 4717 |
+
"epoch": 0.8183271177410761,
|
| 4718 |
+
"grad_norm": 415.0,
|
| 4719 |
+
"learning_rate": 9.872137117719475e-06,
|
| 4720 |
+
"loss": 21.9451,
|
| 4721 |
+
"step": 6720
|
| 4722 |
+
},
|
| 4723 |
+
{
|
| 4724 |
+
"epoch": 0.8195448664281909,
|
| 4725 |
+
"grad_norm": 290.0,
|
| 4726 |
+
"learning_rate": 9.871946845573223e-06,
|
| 4727 |
+
"loss": 22.0432,
|
| 4728 |
+
"step": 6730
|
| 4729 |
+
},
|
| 4730 |
+
{
|
| 4731 |
+
"epoch": 0.8207626151153056,
|
| 4732 |
+
"grad_norm": 319.25,
|
| 4733 |
+
"learning_rate": 9.871756573426974e-06,
|
| 4734 |
+
"loss": 22.2189,
|
| 4735 |
+
"step": 6740
|
| 4736 |
+
},
|
| 4737 |
+
{
|
| 4738 |
+
"epoch": 0.8219803638024202,
|
| 4739 |
+
"grad_norm": 508.75,
|
| 4740 |
+
"learning_rate": 9.871566301280723e-06,
|
| 4741 |
+
"loss": 22.2375,
|
| 4742 |
+
"step": 6750
|
| 4743 |
+
},
|
| 4744 |
+
{
|
| 4745 |
+
"epoch": 0.823198112489535,
|
| 4746 |
+
"grad_norm": 366.5,
|
| 4747 |
+
"learning_rate": 9.871376029134472e-06,
|
| 4748 |
+
"loss": 22.1456,
|
| 4749 |
+
"step": 6760
|
| 4750 |
+
},
|
| 4751 |
+
{
|
| 4752 |
+
"epoch": 0.8244158611766497,
|
| 4753 |
+
"grad_norm": 363.5,
|
| 4754 |
+
"learning_rate": 9.87118575698822e-06,
|
| 4755 |
+
"loss": 22.1803,
|
| 4756 |
+
"step": 6770
|
| 4757 |
+
},
|
| 4758 |
+
{
|
| 4759 |
+
"epoch": 0.8256336098637643,
|
| 4760 |
+
"grad_norm": 510.5,
|
| 4761 |
+
"learning_rate": 9.87099548484197e-06,
|
| 4762 |
+
"loss": 22.2005,
|
| 4763 |
+
"step": 6780
|
| 4764 |
+
},
|
| 4765 |
+
{
|
| 4766 |
+
"epoch": 0.8268513585508791,
|
| 4767 |
+
"grad_norm": 456.75,
|
| 4768 |
+
"learning_rate": 9.87080521269572e-06,
|
| 4769 |
+
"loss": 21.7561,
|
| 4770 |
+
"step": 6790
|
| 4771 |
+
},
|
| 4772 |
+
{
|
| 4773 |
+
"epoch": 0.8280691072379938,
|
| 4774 |
+
"grad_norm": 592.5,
|
| 4775 |
+
"learning_rate": 9.870614940549469e-06,
|
| 4776 |
+
"loss": 22.1845,
|
| 4777 |
+
"step": 6800
|
| 4778 |
+
},
|
| 4779 |
+
{
|
| 4780 |
+
"epoch": 0.8292868559251084,
|
| 4781 |
+
"grad_norm": 403.25,
|
| 4782 |
+
"learning_rate": 9.870424668403218e-06,
|
| 4783 |
+
"loss": 22.1656,
|
| 4784 |
+
"step": 6810
|
| 4785 |
+
},
|
| 4786 |
+
{
|
| 4787 |
+
"epoch": 0.8305046046122232,
|
| 4788 |
+
"grad_norm": 590.0,
|
| 4789 |
+
"learning_rate": 9.870234396256967e-06,
|
| 4790 |
+
"loss": 22.0388,
|
| 4791 |
+
"step": 6820
|
| 4792 |
+
},
|
| 4793 |
+
{
|
| 4794 |
+
"epoch": 0.8317223532993379,
|
| 4795 |
+
"grad_norm": 337.75,
|
| 4796 |
+
"learning_rate": 9.870044124110716e-06,
|
| 4797 |
+
"loss": 22.0442,
|
| 4798 |
+
"step": 6830
|
| 4799 |
+
},
|
| 4800 |
+
{
|
| 4801 |
+
"epoch": 0.8329401019864525,
|
| 4802 |
+
"grad_norm": 322.0,
|
| 4803 |
+
"learning_rate": 9.869853851964466e-06,
|
| 4804 |
+
"loss": 22.0503,
|
| 4805 |
+
"step": 6840
|
| 4806 |
+
},
|
| 4807 |
+
{
|
| 4808 |
+
"epoch": 0.8341578506735673,
|
| 4809 |
+
"grad_norm": 309.5,
|
| 4810 |
+
"learning_rate": 9.869663579818215e-06,
|
| 4811 |
+
"loss": 22.3047,
|
| 4812 |
+
"step": 6850
|
| 4813 |
+
},
|
| 4814 |
+
{
|
| 4815 |
+
"epoch": 0.835375599360682,
|
| 4816 |
+
"grad_norm": 312.0,
|
| 4817 |
+
"learning_rate": 9.869473307671964e-06,
|
| 4818 |
+
"loss": 21.8039,
|
| 4819 |
+
"step": 6860
|
| 4820 |
+
},
|
| 4821 |
+
{
|
| 4822 |
+
"epoch": 0.8365933480477966,
|
| 4823 |
+
"grad_norm": 352.25,
|
| 4824 |
+
"learning_rate": 9.869283035525713e-06,
|
| 4825 |
+
"loss": 22.3301,
|
| 4826 |
+
"step": 6870
|
| 4827 |
+
},
|
| 4828 |
+
{
|
| 4829 |
+
"epoch": 0.8378110967349114,
|
| 4830 |
+
"grad_norm": 334.75,
|
| 4831 |
+
"learning_rate": 9.869092763379462e-06,
|
| 4832 |
+
"loss": 22.0786,
|
| 4833 |
+
"step": 6880
|
| 4834 |
+
},
|
| 4835 |
+
{
|
| 4836 |
+
"epoch": 0.839028845422026,
|
| 4837 |
+
"grad_norm": 315.25,
|
| 4838 |
+
"learning_rate": 9.868902491233212e-06,
|
| 4839 |
+
"loss": 21.8268,
|
| 4840 |
+
"step": 6890
|
| 4841 |
+
},
|
| 4842 |
+
{
|
| 4843 |
+
"epoch": 0.8402465941091407,
|
| 4844 |
+
"grad_norm": 408.0,
|
| 4845 |
+
"learning_rate": 9.868712219086961e-06,
|
| 4846 |
+
"loss": 22.3968,
|
| 4847 |
+
"step": 6900
|
| 4848 |
+
},
|
| 4849 |
+
{
|
| 4850 |
+
"epoch": 0.8414643427962554,
|
| 4851 |
+
"grad_norm": 319.75,
|
| 4852 |
+
"learning_rate": 9.86852194694071e-06,
|
| 4853 |
+
"loss": 21.8756,
|
| 4854 |
+
"step": 6910
|
| 4855 |
+
},
|
| 4856 |
+
{
|
| 4857 |
+
"epoch": 0.8426820914833701,
|
| 4858 |
+
"grad_norm": 334.75,
|
| 4859 |
+
"learning_rate": 9.868331674794459e-06,
|
| 4860 |
+
"loss": 22.3082,
|
| 4861 |
+
"step": 6920
|
| 4862 |
+
},
|
| 4863 |
+
{
|
| 4864 |
+
"epoch": 0.8438998401704848,
|
| 4865 |
+
"grad_norm": 486.0,
|
| 4866 |
+
"learning_rate": 9.868141402648208e-06,
|
| 4867 |
+
"loss": 22.2503,
|
| 4868 |
+
"step": 6930
|
| 4869 |
+
},
|
| 4870 |
+
{
|
| 4871 |
+
"epoch": 0.8451175888575995,
|
| 4872 |
+
"grad_norm": 360.5,
|
| 4873 |
+
"learning_rate": 9.867951130501958e-06,
|
| 4874 |
+
"loss": 22.1495,
|
| 4875 |
+
"step": 6940
|
| 4876 |
+
},
|
| 4877 |
+
{
|
| 4878 |
+
"epoch": 0.8463353375447142,
|
| 4879 |
+
"grad_norm": 550.5,
|
| 4880 |
+
"learning_rate": 9.867760858355707e-06,
|
| 4881 |
+
"loss": 22.4421,
|
| 4882 |
+
"step": 6950
|
| 4883 |
+
},
|
| 4884 |
+
{
|
| 4885 |
+
"epoch": 0.8475530862318289,
|
| 4886 |
+
"grad_norm": 338.5,
|
| 4887 |
+
"learning_rate": 9.867570586209456e-06,
|
| 4888 |
+
"loss": 22.3046,
|
| 4889 |
+
"step": 6960
|
| 4890 |
+
},
|
| 4891 |
+
{
|
| 4892 |
+
"epoch": 0.8487708349189436,
|
| 4893 |
+
"grad_norm": 279.75,
|
| 4894 |
+
"learning_rate": 9.867380314063205e-06,
|
| 4895 |
+
"loss": 21.8373,
|
| 4896 |
+
"step": 6970
|
| 4897 |
+
},
|
| 4898 |
+
{
|
| 4899 |
+
"epoch": 0.8499885836060583,
|
| 4900 |
+
"grad_norm": 467.5,
|
| 4901 |
+
"learning_rate": 9.867190041916954e-06,
|
| 4902 |
+
"loss": 22.2306,
|
| 4903 |
+
"step": 6980
|
| 4904 |
+
},
|
| 4905 |
+
{
|
| 4906 |
+
"epoch": 0.851206332293173,
|
| 4907 |
+
"grad_norm": 357.0,
|
| 4908 |
+
"learning_rate": 9.866999769770704e-06,
|
| 4909 |
+
"loss": 22.0495,
|
| 4910 |
+
"step": 6990
|
| 4911 |
+
},
|
| 4912 |
+
{
|
| 4913 |
+
"epoch": 0.8524240809802877,
|
| 4914 |
+
"grad_norm": 277.0,
|
| 4915 |
+
"learning_rate": 9.866809497624453e-06,
|
| 4916 |
+
"loss": 22.2471,
|
| 4917 |
+
"step": 7000
|
| 4918 |
+
},
|
| 4919 |
+
{
|
| 4920 |
+
"epoch": 0.8536418296674024,
|
| 4921 |
+
"grad_norm": 395.5,
|
| 4922 |
+
"learning_rate": 9.866619225478202e-06,
|
| 4923 |
+
"loss": 22.2751,
|
| 4924 |
+
"step": 7010
|
| 4925 |
+
},
|
| 4926 |
+
{
|
| 4927 |
+
"epoch": 0.8548595783545171,
|
| 4928 |
+
"grad_norm": 369.75,
|
| 4929 |
+
"learning_rate": 9.866428953331951e-06,
|
| 4930 |
+
"loss": 21.8535,
|
| 4931 |
+
"step": 7020
|
| 4932 |
+
},
|
| 4933 |
+
{
|
| 4934 |
+
"epoch": 0.8560773270416318,
|
| 4935 |
+
"grad_norm": 295.5,
|
| 4936 |
+
"learning_rate": 9.8662386811857e-06,
|
| 4937 |
+
"loss": 22.1966,
|
| 4938 |
+
"step": 7030
|
| 4939 |
+
},
|
| 4940 |
+
{
|
| 4941 |
+
"epoch": 0.8572950757287465,
|
| 4942 |
+
"grad_norm": 309.0,
|
| 4943 |
+
"learning_rate": 9.86604840903945e-06,
|
| 4944 |
+
"loss": 22.0675,
|
| 4945 |
+
"step": 7040
|
| 4946 |
+
},
|
| 4947 |
+
{
|
| 4948 |
+
"epoch": 0.8585128244158612,
|
| 4949 |
+
"grad_norm": 299.5,
|
| 4950 |
+
"learning_rate": 9.8658581368932e-06,
|
| 4951 |
+
"loss": 21.9049,
|
| 4952 |
+
"step": 7050
|
| 4953 |
+
},
|
| 4954 |
+
{
|
| 4955 |
+
"epoch": 0.8597305731029758,
|
| 4956 |
+
"grad_norm": 528.5,
|
| 4957 |
+
"learning_rate": 9.865667864746948e-06,
|
| 4958 |
+
"loss": 22.1892,
|
| 4959 |
+
"step": 7060
|
| 4960 |
+
},
|
| 4961 |
+
{
|
| 4962 |
+
"epoch": 0.8609483217900906,
|
| 4963 |
+
"grad_norm": 473.25,
|
| 4964 |
+
"learning_rate": 9.865477592600697e-06,
|
| 4965 |
+
"loss": 21.9495,
|
| 4966 |
+
"step": 7070
|
| 4967 |
+
},
|
| 4968 |
+
{
|
| 4969 |
+
"epoch": 0.8621660704772053,
|
| 4970 |
+
"grad_norm": 695.5,
|
| 4971 |
+
"learning_rate": 9.865287320454446e-06,
|
| 4972 |
+
"loss": 22.0613,
|
| 4973 |
+
"step": 7080
|
| 4974 |
+
},
|
| 4975 |
+
{
|
| 4976 |
+
"epoch": 0.8633838191643199,
|
| 4977 |
+
"grad_norm": 397.5,
|
| 4978 |
+
"learning_rate": 9.865097048308196e-06,
|
| 4979 |
+
"loss": 22.3193,
|
| 4980 |
+
"step": 7090
|
| 4981 |
+
},
|
| 4982 |
+
{
|
| 4983 |
+
"epoch": 0.8646015678514346,
|
| 4984 |
+
"grad_norm": 304.25,
|
| 4985 |
+
"learning_rate": 9.864906776161945e-06,
|
| 4986 |
+
"loss": 21.9795,
|
| 4987 |
+
"step": 7100
|
| 4988 |
+
},
|
| 4989 |
+
{
|
| 4990 |
+
"epoch": 0.8658193165385494,
|
| 4991 |
+
"grad_norm": 400.25,
|
| 4992 |
+
"learning_rate": 9.864716504015694e-06,
|
| 4993 |
+
"loss": 22.2449,
|
| 4994 |
+
"step": 7110
|
| 4995 |
+
},
|
| 4996 |
+
{
|
| 4997 |
+
"epoch": 0.867037065225664,
|
| 4998 |
+
"grad_norm": 295.25,
|
| 4999 |
+
"learning_rate": 9.864526231869443e-06,
|
| 5000 |
+
"loss": 21.85,
|
| 5001 |
+
"step": 7120
|
| 5002 |
+
},
|
| 5003 |
+
{
|
| 5004 |
+
"epoch": 0.8682548139127787,
|
| 5005 |
+
"grad_norm": 297.25,
|
| 5006 |
+
"learning_rate": 9.864335959723192e-06,
|
| 5007 |
+
"loss": 22.0741,
|
| 5008 |
+
"step": 7130
|
| 5009 |
+
},
|
| 5010 |
+
{
|
| 5011 |
+
"epoch": 0.8694725625998935,
|
| 5012 |
+
"grad_norm": 351.75,
|
| 5013 |
+
"learning_rate": 9.864145687576942e-06,
|
| 5014 |
+
"loss": 22.0668,
|
| 5015 |
+
"step": 7140
|
| 5016 |
+
},
|
| 5017 |
+
{
|
| 5018 |
+
"epoch": 0.8706903112870081,
|
| 5019 |
+
"grad_norm": 484.5,
|
| 5020 |
+
"learning_rate": 9.863955415430691e-06,
|
| 5021 |
+
"loss": 22.237,
|
| 5022 |
+
"step": 7150
|
| 5023 |
+
},
|
| 5024 |
+
{
|
| 5025 |
+
"epoch": 0.8719080599741228,
|
| 5026 |
+
"grad_norm": 696.5,
|
| 5027 |
+
"learning_rate": 9.86376514328444e-06,
|
| 5028 |
+
"loss": 22.2217,
|
| 5029 |
+
"step": 7160
|
| 5030 |
+
},
|
| 5031 |
+
{
|
| 5032 |
+
"epoch": 0.8731258086612376,
|
| 5033 |
+
"grad_norm": 391.75,
|
| 5034 |
+
"learning_rate": 9.863574871138189e-06,
|
| 5035 |
+
"loss": 21.9983,
|
| 5036 |
+
"step": 7170
|
| 5037 |
+
},
|
| 5038 |
+
{
|
| 5039 |
+
"epoch": 0.8743435573483522,
|
| 5040 |
+
"grad_norm": 474.5,
|
| 5041 |
+
"learning_rate": 9.863384598991938e-06,
|
| 5042 |
+
"loss": 22.8282,
|
| 5043 |
+
"step": 7180
|
| 5044 |
+
},
|
| 5045 |
+
{
|
| 5046 |
+
"epoch": 0.8755613060354669,
|
| 5047 |
+
"grad_norm": 328.5,
|
| 5048 |
+
"learning_rate": 9.863194326845689e-06,
|
| 5049 |
+
"loss": 21.9469,
|
| 5050 |
+
"step": 7190
|
| 5051 |
+
},
|
| 5052 |
+
{
|
| 5053 |
+
"epoch": 0.8767790547225817,
|
| 5054 |
+
"grad_norm": 325.5,
|
| 5055 |
+
"learning_rate": 9.863004054699437e-06,
|
| 5056 |
+
"loss": 22.0664,
|
| 5057 |
+
"step": 7200
|
| 5058 |
+
},
|
| 5059 |
+
{
|
| 5060 |
+
"epoch": 0.8779968034096963,
|
| 5061 |
+
"grad_norm": 435.5,
|
| 5062 |
+
"learning_rate": 9.862813782553186e-06,
|
| 5063 |
+
"loss": 22.2138,
|
| 5064 |
+
"step": 7210
|
| 5065 |
+
},
|
| 5066 |
+
{
|
| 5067 |
+
"epoch": 0.879214552096811,
|
| 5068 |
+
"grad_norm": 430.0,
|
| 5069 |
+
"learning_rate": 9.862623510406935e-06,
|
| 5070 |
+
"loss": 21.8652,
|
| 5071 |
+
"step": 7220
|
| 5072 |
+
},
|
| 5073 |
+
{
|
| 5074 |
+
"epoch": 0.8804323007839258,
|
| 5075 |
+
"grad_norm": 485.75,
|
| 5076 |
+
"learning_rate": 9.862433238260684e-06,
|
| 5077 |
+
"loss": 21.9634,
|
| 5078 |
+
"step": 7230
|
| 5079 |
+
},
|
| 5080 |
+
{
|
| 5081 |
+
"epoch": 0.8816500494710404,
|
| 5082 |
+
"grad_norm": 445.75,
|
| 5083 |
+
"learning_rate": 9.862242966114435e-06,
|
| 5084 |
+
"loss": 22.1161,
|
| 5085 |
+
"step": 7240
|
| 5086 |
+
},
|
| 5087 |
+
{
|
| 5088 |
+
"epoch": 0.8828677981581551,
|
| 5089 |
+
"grad_norm": 348.25,
|
| 5090 |
+
"learning_rate": 9.862052693968183e-06,
|
| 5091 |
+
"loss": 22.2584,
|
| 5092 |
+
"step": 7250
|
| 5093 |
+
},
|
| 5094 |
+
{
|
| 5095 |
+
"epoch": 0.8840855468452699,
|
| 5096 |
+
"grad_norm": 433.5,
|
| 5097 |
+
"learning_rate": 9.861862421821932e-06,
|
| 5098 |
+
"loss": 21.846,
|
| 5099 |
+
"step": 7260
|
| 5100 |
+
},
|
| 5101 |
+
{
|
| 5102 |
+
"epoch": 0.8853032955323845,
|
| 5103 |
+
"grad_norm": 709.5,
|
| 5104 |
+
"learning_rate": 9.861672149675681e-06,
|
| 5105 |
+
"loss": 22.3034,
|
| 5106 |
+
"step": 7270
|
| 5107 |
+
},
|
| 5108 |
+
{
|
| 5109 |
+
"epoch": 0.8865210442194992,
|
| 5110 |
+
"grad_norm": 569.5,
|
| 5111 |
+
"learning_rate": 9.86148187752943e-06,
|
| 5112 |
+
"loss": 22.0625,
|
| 5113 |
+
"step": 7280
|
| 5114 |
+
},
|
| 5115 |
+
{
|
| 5116 |
+
"epoch": 0.887738792906614,
|
| 5117 |
+
"grad_norm": 542.0,
|
| 5118 |
+
"learning_rate": 9.86129160538318e-06,
|
| 5119 |
+
"loss": 21.9613,
|
| 5120 |
+
"step": 7290
|
| 5121 |
+
},
|
| 5122 |
+
{
|
| 5123 |
+
"epoch": 0.8889565415937286,
|
| 5124 |
+
"grad_norm": 272.5,
|
| 5125 |
+
"learning_rate": 9.86110133323693e-06,
|
| 5126 |
+
"loss": 21.9754,
|
| 5127 |
+
"step": 7300
|
| 5128 |
+
},
|
| 5129 |
+
{
|
| 5130 |
+
"epoch": 0.8901742902808433,
|
| 5131 |
+
"grad_norm": 316.75,
|
| 5132 |
+
"learning_rate": 9.860911061090678e-06,
|
| 5133 |
+
"loss": 21.8645,
|
| 5134 |
+
"step": 7310
|
| 5135 |
+
},
|
| 5136 |
+
{
|
| 5137 |
+
"epoch": 0.8913920389679579,
|
| 5138 |
+
"grad_norm": 434.5,
|
| 5139 |
+
"learning_rate": 9.860720788944427e-06,
|
| 5140 |
+
"loss": 22.0292,
|
| 5141 |
+
"step": 7320
|
| 5142 |
+
},
|
| 5143 |
+
{
|
| 5144 |
+
"epoch": 0.8926097876550727,
|
| 5145 |
+
"grad_norm": 460.25,
|
| 5146 |
+
"learning_rate": 9.860530516798176e-06,
|
| 5147 |
+
"loss": 22.0469,
|
| 5148 |
+
"step": 7330
|
| 5149 |
+
},
|
| 5150 |
+
{
|
| 5151 |
+
"epoch": 0.8938275363421874,
|
| 5152 |
+
"grad_norm": 469.5,
|
| 5153 |
+
"learning_rate": 9.860340244651927e-06,
|
| 5154 |
+
"loss": 21.835,
|
| 5155 |
+
"step": 7340
|
| 5156 |
+
},
|
| 5157 |
+
{
|
| 5158 |
+
"epoch": 0.895045285029302,
|
| 5159 |
+
"grad_norm": 542.0,
|
| 5160 |
+
"learning_rate": 9.860149972505676e-06,
|
| 5161 |
+
"loss": 22.1604,
|
| 5162 |
+
"step": 7350
|
| 5163 |
+
},
|
| 5164 |
+
{
|
| 5165 |
+
"epoch": 0.8962630337164168,
|
| 5166 |
+
"grad_norm": 300.0,
|
| 5167 |
+
"learning_rate": 9.859959700359424e-06,
|
| 5168 |
+
"loss": 22.2402,
|
| 5169 |
+
"step": 7360
|
| 5170 |
+
},
|
| 5171 |
+
{
|
| 5172 |
+
"epoch": 0.8974807824035315,
|
| 5173 |
+
"grad_norm": 414.25,
|
| 5174 |
+
"learning_rate": 9.859769428213173e-06,
|
| 5175 |
+
"loss": 22.086,
|
| 5176 |
+
"step": 7370
|
| 5177 |
+
},
|
| 5178 |
+
{
|
| 5179 |
+
"epoch": 0.8986985310906461,
|
| 5180 |
+
"grad_norm": 274.0,
|
| 5181 |
+
"learning_rate": 9.859579156066922e-06,
|
| 5182 |
+
"loss": 22.0595,
|
| 5183 |
+
"step": 7380
|
| 5184 |
+
},
|
| 5185 |
+
{
|
| 5186 |
+
"epoch": 0.8999162797777609,
|
| 5187 |
+
"grad_norm": 293.0,
|
| 5188 |
+
"learning_rate": 9.859388883920673e-06,
|
| 5189 |
+
"loss": 22.0757,
|
| 5190 |
+
"step": 7390
|
| 5191 |
+
},
|
| 5192 |
+
{
|
| 5193 |
+
"epoch": 0.9011340284648756,
|
| 5194 |
+
"grad_norm": 295.5,
|
| 5195 |
+
"learning_rate": 9.859198611774422e-06,
|
| 5196 |
+
"loss": 22.2322,
|
| 5197 |
+
"step": 7400
|
| 5198 |
+
},
|
| 5199 |
+
{
|
| 5200 |
+
"epoch": 0.9023517771519902,
|
| 5201 |
+
"grad_norm": 538.0,
|
| 5202 |
+
"learning_rate": 9.85900833962817e-06,
|
| 5203 |
+
"loss": 21.8385,
|
| 5204 |
+
"step": 7410
|
| 5205 |
+
},
|
| 5206 |
+
{
|
| 5207 |
+
"epoch": 0.903569525839105,
|
| 5208 |
+
"grad_norm": 528.5,
|
| 5209 |
+
"learning_rate": 9.85881806748192e-06,
|
| 5210 |
+
"loss": 22.345,
|
| 5211 |
+
"step": 7420
|
| 5212 |
+
},
|
| 5213 |
+
{
|
| 5214 |
+
"epoch": 0.9047872745262197,
|
| 5215 |
+
"grad_norm": 635.0,
|
| 5216 |
+
"learning_rate": 9.858627795335668e-06,
|
| 5217 |
+
"loss": 22.1144,
|
| 5218 |
+
"step": 7430
|
| 5219 |
+
},
|
| 5220 |
+
{
|
| 5221 |
+
"epoch": 0.9060050232133343,
|
| 5222 |
+
"grad_norm": 1307.0,
|
| 5223 |
+
"learning_rate": 9.858437523189419e-06,
|
| 5224 |
+
"loss": 22.1034,
|
| 5225 |
+
"step": 7440
|
| 5226 |
+
},
|
| 5227 |
+
{
|
| 5228 |
+
"epoch": 0.9072227719004491,
|
| 5229 |
+
"grad_norm": 453.0,
|
| 5230 |
+
"learning_rate": 9.858247251043168e-06,
|
| 5231 |
+
"loss": 21.9531,
|
| 5232 |
+
"step": 7450
|
| 5233 |
+
},
|
| 5234 |
+
{
|
| 5235 |
+
"epoch": 0.9084405205875637,
|
| 5236 |
+
"grad_norm": 399.75,
|
| 5237 |
+
"learning_rate": 9.858056978896917e-06,
|
| 5238 |
+
"loss": 21.9347,
|
| 5239 |
+
"step": 7460
|
| 5240 |
+
},
|
| 5241 |
+
{
|
| 5242 |
+
"epoch": 0.9096582692746784,
|
| 5243 |
+
"grad_norm": 390.75,
|
| 5244 |
+
"learning_rate": 9.857866706750665e-06,
|
| 5245 |
+
"loss": 22.138,
|
| 5246 |
+
"step": 7470
|
| 5247 |
+
},
|
| 5248 |
+
{
|
| 5249 |
+
"epoch": 0.9108760179617932,
|
| 5250 |
+
"grad_norm": 300.5,
|
| 5251 |
+
"learning_rate": 9.857676434604416e-06,
|
| 5252 |
+
"loss": 22.3572,
|
| 5253 |
+
"step": 7480
|
| 5254 |
+
},
|
| 5255 |
+
{
|
| 5256 |
+
"epoch": 0.9120937666489078,
|
| 5257 |
+
"grad_norm": 652.5,
|
| 5258 |
+
"learning_rate": 9.857486162458165e-06,
|
| 5259 |
+
"loss": 22.3054,
|
| 5260 |
+
"step": 7490
|
| 5261 |
+
},
|
| 5262 |
+
{
|
| 5263 |
+
"epoch": 0.9133115153360225,
|
| 5264 |
+
"grad_norm": 378.25,
|
| 5265 |
+
"learning_rate": 9.857295890311914e-06,
|
| 5266 |
+
"loss": 21.9117,
|
| 5267 |
+
"step": 7500
|
| 5268 |
+
},
|
| 5269 |
+
{
|
| 5270 |
+
"epoch": 0.9145292640231373,
|
| 5271 |
+
"grad_norm": 347.5,
|
| 5272 |
+
"learning_rate": 9.857105618165663e-06,
|
| 5273 |
+
"loss": 22.1442,
|
| 5274 |
+
"step": 7510
|
| 5275 |
+
},
|
| 5276 |
+
{
|
| 5277 |
+
"epoch": 0.9157470127102519,
|
| 5278 |
+
"grad_norm": 383.75,
|
| 5279 |
+
"learning_rate": 9.856915346019411e-06,
|
| 5280 |
+
"loss": 22.4111,
|
| 5281 |
+
"step": 7520
|
| 5282 |
+
},
|
| 5283 |
+
{
|
| 5284 |
+
"epoch": 0.9169647613973666,
|
| 5285 |
+
"grad_norm": 403.25,
|
| 5286 |
+
"learning_rate": 9.856725073873162e-06,
|
| 5287 |
+
"loss": 21.9787,
|
| 5288 |
+
"step": 7530
|
| 5289 |
+
},
|
| 5290 |
+
{
|
| 5291 |
+
"epoch": 0.9181825100844813,
|
| 5292 |
+
"grad_norm": 457.5,
|
| 5293 |
+
"learning_rate": 9.856534801726911e-06,
|
| 5294 |
+
"loss": 21.8651,
|
| 5295 |
+
"step": 7540
|
| 5296 |
+
},
|
| 5297 |
+
{
|
| 5298 |
+
"epoch": 0.919400258771596,
|
| 5299 |
+
"grad_norm": 518.0,
|
| 5300 |
+
"learning_rate": 9.85634452958066e-06,
|
| 5301 |
+
"loss": 21.9291,
|
| 5302 |
+
"step": 7550
|
| 5303 |
+
},
|
| 5304 |
+
{
|
| 5305 |
+
"epoch": 0.9206180074587107,
|
| 5306 |
+
"grad_norm": 558.5,
|
| 5307 |
+
"learning_rate": 9.856154257434409e-06,
|
| 5308 |
+
"loss": 22.1247,
|
| 5309 |
+
"step": 7560
|
| 5310 |
+
},
|
| 5311 |
+
{
|
| 5312 |
+
"epoch": 0.9218357561458254,
|
| 5313 |
+
"grad_norm": 329.25,
|
| 5314 |
+
"learning_rate": 9.855963985288158e-06,
|
| 5315 |
+
"loss": 22.1231,
|
| 5316 |
+
"step": 7570
|
| 5317 |
+
},
|
| 5318 |
+
{
|
| 5319 |
+
"epoch": 0.9230535048329401,
|
| 5320 |
+
"grad_norm": 593.0,
|
| 5321 |
+
"learning_rate": 9.855773713141908e-06,
|
| 5322 |
+
"loss": 22.0844,
|
| 5323 |
+
"step": 7580
|
| 5324 |
+
},
|
| 5325 |
+
{
|
| 5326 |
+
"epoch": 0.9242712535200548,
|
| 5327 |
+
"grad_norm": 361.5,
|
| 5328 |
+
"learning_rate": 9.855583440995657e-06,
|
| 5329 |
+
"loss": 22.0896,
|
| 5330 |
+
"step": 7590
|
| 5331 |
+
},
|
| 5332 |
+
{
|
| 5333 |
+
"epoch": 0.9254890022071695,
|
| 5334 |
+
"grad_norm": 487.25,
|
| 5335 |
+
"learning_rate": 9.855393168849406e-06,
|
| 5336 |
+
"loss": 22.3015,
|
| 5337 |
+
"step": 7600
|
| 5338 |
+
},
|
| 5339 |
+
{
|
| 5340 |
+
"epoch": 0.9267067508942842,
|
| 5341 |
+
"grad_norm": 247.125,
|
| 5342 |
+
"learning_rate": 9.855202896703155e-06,
|
| 5343 |
+
"loss": 22.0374,
|
| 5344 |
+
"step": 7610
|
| 5345 |
+
},
|
| 5346 |
+
{
|
| 5347 |
+
"epoch": 0.9279244995813989,
|
| 5348 |
+
"grad_norm": 434.5,
|
| 5349 |
+
"learning_rate": 9.855012624556905e-06,
|
| 5350 |
+
"loss": 22.0935,
|
| 5351 |
+
"step": 7620
|
| 5352 |
+
},
|
| 5353 |
+
{
|
| 5354 |
+
"epoch": 0.9291422482685135,
|
| 5355 |
+
"grad_norm": 357.25,
|
| 5356 |
+
"learning_rate": 9.854822352410654e-06,
|
| 5357 |
+
"loss": 22.2531,
|
| 5358 |
+
"step": 7630
|
| 5359 |
+
},
|
| 5360 |
+
{
|
| 5361 |
+
"epoch": 0.9303599969556283,
|
| 5362 |
+
"grad_norm": 382.5,
|
| 5363 |
+
"learning_rate": 9.854632080264403e-06,
|
| 5364 |
+
"loss": 21.9988,
|
| 5365 |
+
"step": 7640
|
| 5366 |
+
},
|
| 5367 |
+
{
|
| 5368 |
+
"epoch": 0.931577745642743,
|
| 5369 |
+
"grad_norm": 518.0,
|
| 5370 |
+
"learning_rate": 9.854441808118152e-06,
|
| 5371 |
+
"loss": 21.9012,
|
| 5372 |
+
"step": 7650
|
| 5373 |
+
},
|
| 5374 |
+
{
|
| 5375 |
+
"epoch": 0.9327954943298576,
|
| 5376 |
+
"grad_norm": 402.75,
|
| 5377 |
+
"learning_rate": 9.8542515359719e-06,
|
| 5378 |
+
"loss": 21.89,
|
| 5379 |
+
"step": 7660
|
| 5380 |
+
},
|
| 5381 |
+
{
|
| 5382 |
+
"epoch": 0.9340132430169724,
|
| 5383 |
+
"grad_norm": 585.0,
|
| 5384 |
+
"learning_rate": 9.854061263825651e-06,
|
| 5385 |
+
"loss": 22.0898,
|
| 5386 |
+
"step": 7670
|
| 5387 |
+
},
|
| 5388 |
+
{
|
| 5389 |
+
"epoch": 0.9352309917040871,
|
| 5390 |
+
"grad_norm": 403.5,
|
| 5391 |
+
"learning_rate": 9.8538709916794e-06,
|
| 5392 |
+
"loss": 22.1882,
|
| 5393 |
+
"step": 7680
|
| 5394 |
+
},
|
| 5395 |
+
{
|
| 5396 |
+
"epoch": 0.9364487403912017,
|
| 5397 |
+
"grad_norm": 278.75,
|
| 5398 |
+
"learning_rate": 9.853680719533149e-06,
|
| 5399 |
+
"loss": 22.0084,
|
| 5400 |
+
"step": 7690
|
| 5401 |
+
},
|
| 5402 |
+
{
|
| 5403 |
+
"epoch": 0.9376664890783165,
|
| 5404 |
+
"grad_norm": 293.75,
|
| 5405 |
+
"learning_rate": 9.853490447386898e-06,
|
| 5406 |
+
"loss": 22.1582,
|
| 5407 |
+
"step": 7700
|
| 5408 |
+
},
|
| 5409 |
+
{
|
| 5410 |
+
"epoch": 0.9388842377654312,
|
| 5411 |
+
"grad_norm": 400.5,
|
| 5412 |
+
"learning_rate": 9.853300175240648e-06,
|
| 5413 |
+
"loss": 21.6699,
|
| 5414 |
+
"step": 7710
|
| 5415 |
+
},
|
| 5416 |
+
{
|
| 5417 |
+
"epoch": 0.9401019864525458,
|
| 5418 |
+
"grad_norm": 300.25,
|
| 5419 |
+
"learning_rate": 9.853109903094397e-06,
|
| 5420 |
+
"loss": 22.1939,
|
| 5421 |
+
"step": 7720
|
| 5422 |
+
},
|
| 5423 |
+
{
|
| 5424 |
+
"epoch": 0.9413197351396606,
|
| 5425 |
+
"grad_norm": 329.25,
|
| 5426 |
+
"learning_rate": 9.852919630948146e-06,
|
| 5427 |
+
"loss": 21.8216,
|
| 5428 |
+
"step": 7730
|
| 5429 |
+
},
|
| 5430 |
+
{
|
| 5431 |
+
"epoch": 0.9425374838267753,
|
| 5432 |
+
"grad_norm": 266.25,
|
| 5433 |
+
"learning_rate": 9.852729358801895e-06,
|
| 5434 |
+
"loss": 22.0178,
|
| 5435 |
+
"step": 7740
|
| 5436 |
+
},
|
| 5437 |
+
{
|
| 5438 |
+
"epoch": 0.9437552325138899,
|
| 5439 |
+
"grad_norm": 325.75,
|
| 5440 |
+
"learning_rate": 9.852539086655644e-06,
|
| 5441 |
+
"loss": 22.0285,
|
| 5442 |
+
"step": 7750
|
| 5443 |
+
},
|
| 5444 |
+
{
|
| 5445 |
+
"epoch": 0.9449729812010046,
|
| 5446 |
+
"grad_norm": 530.5,
|
| 5447 |
+
"learning_rate": 9.852348814509395e-06,
|
| 5448 |
+
"loss": 21.9276,
|
| 5449 |
+
"step": 7760
|
| 5450 |
+
},
|
| 5451 |
+
{
|
| 5452 |
+
"epoch": 0.9461907298881194,
|
| 5453 |
+
"grad_norm": 448.25,
|
| 5454 |
+
"learning_rate": 9.852158542363143e-06,
|
| 5455 |
+
"loss": 21.7508,
|
| 5456 |
+
"step": 7770
|
| 5457 |
+
},
|
| 5458 |
+
{
|
| 5459 |
+
"epoch": 0.947408478575234,
|
| 5460 |
+
"grad_norm": 370.5,
|
| 5461 |
+
"learning_rate": 9.851968270216892e-06,
|
| 5462 |
+
"loss": 22.271,
|
| 5463 |
+
"step": 7780
|
| 5464 |
+
},
|
| 5465 |
+
{
|
| 5466 |
+
"epoch": 0.9486262272623487,
|
| 5467 |
+
"grad_norm": 716.5,
|
| 5468 |
+
"learning_rate": 9.851777998070641e-06,
|
| 5469 |
+
"loss": 21.8806,
|
| 5470 |
+
"step": 7790
|
| 5471 |
+
},
|
| 5472 |
+
{
|
| 5473 |
+
"epoch": 0.9498439759494635,
|
| 5474 |
+
"grad_norm": 549.5,
|
| 5475 |
+
"learning_rate": 9.85158772592439e-06,
|
| 5476 |
+
"loss": 21.8866,
|
| 5477 |
+
"step": 7800
|
| 5478 |
+
},
|
| 5479 |
+
{
|
| 5480 |
+
"epoch": 0.9510617246365781,
|
| 5481 |
+
"grad_norm": 265.75,
|
| 5482 |
+
"learning_rate": 9.85139745377814e-06,
|
| 5483 |
+
"loss": 21.8491,
|
| 5484 |
+
"step": 7810
|
| 5485 |
+
},
|
| 5486 |
+
{
|
| 5487 |
+
"epoch": 0.9522794733236928,
|
| 5488 |
+
"grad_norm": 544.0,
|
| 5489 |
+
"learning_rate": 9.85120718163189e-06,
|
| 5490 |
+
"loss": 21.9407,
|
| 5491 |
+
"step": 7820
|
| 5492 |
+
},
|
| 5493 |
+
{
|
| 5494 |
+
"epoch": 0.9534972220108076,
|
| 5495 |
+
"grad_norm": 254.5,
|
| 5496 |
+
"learning_rate": 9.851016909485638e-06,
|
| 5497 |
+
"loss": 22.0854,
|
| 5498 |
+
"step": 7830
|
| 5499 |
+
},
|
| 5500 |
+
{
|
| 5501 |
+
"epoch": 0.9547149706979222,
|
| 5502 |
+
"grad_norm": 321.0,
|
| 5503 |
+
"learning_rate": 9.850826637339387e-06,
|
| 5504 |
+
"loss": 22.0765,
|
| 5505 |
+
"step": 7840
|
| 5506 |
+
},
|
| 5507 |
+
{
|
| 5508 |
+
"epoch": 0.9559327193850369,
|
| 5509 |
+
"grad_norm": 363.75,
|
| 5510 |
+
"learning_rate": 9.850636365193138e-06,
|
| 5511 |
+
"loss": 22.0777,
|
| 5512 |
+
"step": 7850
|
| 5513 |
+
},
|
| 5514 |
+
{
|
| 5515 |
+
"epoch": 0.9571504680721517,
|
| 5516 |
+
"grad_norm": 383.0,
|
| 5517 |
+
"learning_rate": 9.850446093046887e-06,
|
| 5518 |
+
"loss": 22.0465,
|
| 5519 |
+
"step": 7860
|
| 5520 |
+
},
|
| 5521 |
+
{
|
| 5522 |
+
"epoch": 0.9583682167592663,
|
| 5523 |
+
"grad_norm": 314.25,
|
| 5524 |
+
"learning_rate": 9.850255820900636e-06,
|
| 5525 |
+
"loss": 21.9334,
|
| 5526 |
+
"step": 7870
|
| 5527 |
+
},
|
| 5528 |
+
{
|
| 5529 |
+
"epoch": 0.959585965446381,
|
| 5530 |
+
"grad_norm": 440.0,
|
| 5531 |
+
"learning_rate": 9.850065548754384e-06,
|
| 5532 |
+
"loss": 22.0787,
|
| 5533 |
+
"step": 7880
|
| 5534 |
+
},
|
| 5535 |
+
{
|
| 5536 |
+
"epoch": 0.9608037141334957,
|
| 5537 |
+
"grad_norm": 478.5,
|
| 5538 |
+
"learning_rate": 9.849875276608133e-06,
|
| 5539 |
+
"loss": 22.4122,
|
| 5540 |
+
"step": 7890
|
| 5541 |
+
},
|
| 5542 |
+
{
|
| 5543 |
+
"epoch": 0.9620214628206104,
|
| 5544 |
+
"grad_norm": 385.25,
|
| 5545 |
+
"learning_rate": 9.849685004461884e-06,
|
| 5546 |
+
"loss": 22.0855,
|
| 5547 |
+
"step": 7900
|
| 5548 |
+
},
|
| 5549 |
+
{
|
| 5550 |
+
"epoch": 0.9632392115077251,
|
| 5551 |
+
"grad_norm": 389.75,
|
| 5552 |
+
"learning_rate": 9.849494732315633e-06,
|
| 5553 |
+
"loss": 22.1411,
|
| 5554 |
+
"step": 7910
|
| 5555 |
+
},
|
| 5556 |
+
{
|
| 5557 |
+
"epoch": 0.9644569601948398,
|
| 5558 |
+
"grad_norm": 342.5,
|
| 5559 |
+
"learning_rate": 9.849304460169382e-06,
|
| 5560 |
+
"loss": 22.098,
|
| 5561 |
+
"step": 7920
|
| 5562 |
+
},
|
| 5563 |
+
{
|
| 5564 |
+
"epoch": 0.9656747088819545,
|
| 5565 |
+
"grad_norm": 266.75,
|
| 5566 |
+
"learning_rate": 9.84911418802313e-06,
|
| 5567 |
+
"loss": 22.1849,
|
| 5568 |
+
"step": 7930
|
| 5569 |
+
},
|
| 5570 |
+
{
|
| 5571 |
+
"epoch": 0.9668924575690692,
|
| 5572 |
+
"grad_norm": 556.5,
|
| 5573 |
+
"learning_rate": 9.84892391587688e-06,
|
| 5574 |
+
"loss": 22.1534,
|
| 5575 |
+
"step": 7940
|
| 5576 |
+
},
|
| 5577 |
+
{
|
| 5578 |
+
"epoch": 0.9681102062561838,
|
| 5579 |
+
"grad_norm": 413.75,
|
| 5580 |
+
"learning_rate": 9.84873364373063e-06,
|
| 5581 |
+
"loss": 21.9893,
|
| 5582 |
+
"step": 7950
|
| 5583 |
+
},
|
| 5584 |
+
{
|
| 5585 |
+
"epoch": 0.9693279549432986,
|
| 5586 |
+
"grad_norm": 403.75,
|
| 5587 |
+
"learning_rate": 9.848543371584379e-06,
|
| 5588 |
+
"loss": 22.0232,
|
| 5589 |
+
"step": 7960
|
| 5590 |
+
},
|
| 5591 |
+
{
|
| 5592 |
+
"epoch": 0.9705457036304133,
|
| 5593 |
+
"grad_norm": 713.0,
|
| 5594 |
+
"learning_rate": 9.848353099438128e-06,
|
| 5595 |
+
"loss": 22.1614,
|
| 5596 |
+
"step": 7970
|
| 5597 |
+
},
|
| 5598 |
+
{
|
| 5599 |
+
"epoch": 0.9717634523175279,
|
| 5600 |
+
"grad_norm": 349.5,
|
| 5601 |
+
"learning_rate": 9.848162827291876e-06,
|
| 5602 |
+
"loss": 21.9629,
|
| 5603 |
+
"step": 7980
|
| 5604 |
+
},
|
| 5605 |
+
{
|
| 5606 |
+
"epoch": 0.9729812010046427,
|
| 5607 |
+
"grad_norm": 450.25,
|
| 5608 |
+
"learning_rate": 9.847972555145625e-06,
|
| 5609 |
+
"loss": 21.956,
|
| 5610 |
+
"step": 7990
|
| 5611 |
+
},
|
| 5612 |
+
{
|
| 5613 |
+
"epoch": 0.9741989496917574,
|
| 5614 |
+
"grad_norm": 343.75,
|
| 5615 |
+
"learning_rate": 9.847782282999376e-06,
|
| 5616 |
+
"loss": 21.8555,
|
| 5617 |
+
"step": 8000
|
| 5618 |
+
},
|
| 5619 |
+
{
|
| 5620 |
+
"epoch": 0.975416698378872,
|
| 5621 |
+
"grad_norm": 528.5,
|
| 5622 |
+
"learning_rate": 9.847592010853125e-06,
|
| 5623 |
+
"loss": 21.9875,
|
| 5624 |
+
"step": 8010
|
| 5625 |
+
},
|
| 5626 |
+
{
|
| 5627 |
+
"epoch": 0.9766344470659868,
|
| 5628 |
+
"grad_norm": 587.5,
|
| 5629 |
+
"learning_rate": 9.847401738706874e-06,
|
| 5630 |
+
"loss": 21.8694,
|
| 5631 |
+
"step": 8020
|
| 5632 |
+
},
|
| 5633 |
+
{
|
| 5634 |
+
"epoch": 0.9778521957531014,
|
| 5635 |
+
"grad_norm": 807.5,
|
| 5636 |
+
"learning_rate": 9.847211466560623e-06,
|
| 5637 |
+
"loss": 21.7027,
|
| 5638 |
+
"step": 8030
|
| 5639 |
+
},
|
| 5640 |
+
{
|
| 5641 |
+
"epoch": 0.9790699444402161,
|
| 5642 |
+
"grad_norm": 629.0,
|
| 5643 |
+
"learning_rate": 9.847021194414371e-06,
|
| 5644 |
+
"loss": 21.8312,
|
| 5645 |
+
"step": 8040
|
| 5646 |
+
},
|
| 5647 |
+
{
|
| 5648 |
+
"epoch": 0.9802876931273309,
|
| 5649 |
+
"grad_norm": 564.5,
|
| 5650 |
+
"learning_rate": 9.846830922268122e-06,
|
| 5651 |
+
"loss": 22.0643,
|
| 5652 |
+
"step": 8050
|
| 5653 |
+
},
|
| 5654 |
+
{
|
| 5655 |
+
"epoch": 0.9815054418144455,
|
| 5656 |
+
"grad_norm": 439.0,
|
| 5657 |
+
"learning_rate": 9.84664065012187e-06,
|
| 5658 |
+
"loss": 21.993,
|
| 5659 |
+
"step": 8060
|
| 5660 |
+
},
|
| 5661 |
+
{
|
| 5662 |
+
"epoch": 0.9827231905015602,
|
| 5663 |
+
"grad_norm": 380.0,
|
| 5664 |
+
"learning_rate": 9.84645037797562e-06,
|
| 5665 |
+
"loss": 21.8951,
|
| 5666 |
+
"step": 8070
|
| 5667 |
+
},
|
| 5668 |
+
{
|
| 5669 |
+
"epoch": 0.983940939188675,
|
| 5670 |
+
"grad_norm": 828.0,
|
| 5671 |
+
"learning_rate": 9.846260105829369e-06,
|
| 5672 |
+
"loss": 21.9939,
|
| 5673 |
+
"step": 8080
|
| 5674 |
+
},
|
| 5675 |
+
{
|
| 5676 |
+
"epoch": 0.9851586878757896,
|
| 5677 |
+
"grad_norm": 306.75,
|
| 5678 |
+
"learning_rate": 9.846069833683117e-06,
|
| 5679 |
+
"loss": 22.3458,
|
| 5680 |
+
"step": 8090
|
| 5681 |
+
},
|
| 5682 |
+
{
|
| 5683 |
+
"epoch": 0.9863764365629043,
|
| 5684 |
+
"grad_norm": 384.0,
|
| 5685 |
+
"learning_rate": 9.845879561536868e-06,
|
| 5686 |
+
"loss": 21.869,
|
| 5687 |
+
"step": 8100
|
| 5688 |
+
},
|
| 5689 |
+
{
|
| 5690 |
+
"epoch": 0.9875941852500191,
|
| 5691 |
+
"grad_norm": 377.75,
|
| 5692 |
+
"learning_rate": 9.845689289390617e-06,
|
| 5693 |
+
"loss": 22.1,
|
| 5694 |
+
"step": 8110
|
| 5695 |
+
},
|
| 5696 |
+
{
|
| 5697 |
+
"epoch": 0.9888119339371337,
|
| 5698 |
+
"grad_norm": 531.0,
|
| 5699 |
+
"learning_rate": 9.845499017244366e-06,
|
| 5700 |
+
"loss": 22.2576,
|
| 5701 |
+
"step": 8120
|
| 5702 |
+
},
|
| 5703 |
+
{
|
| 5704 |
+
"epoch": 0.9900296826242484,
|
| 5705 |
+
"grad_norm": 531.5,
|
| 5706 |
+
"learning_rate": 9.845308745098115e-06,
|
| 5707 |
+
"loss": 22.0621,
|
| 5708 |
+
"step": 8130
|
| 5709 |
+
},
|
| 5710 |
+
{
|
| 5711 |
+
"epoch": 0.9912474313113632,
|
| 5712 |
+
"grad_norm": 466.25,
|
| 5713 |
+
"learning_rate": 9.845118472951864e-06,
|
| 5714 |
+
"loss": 21.9393,
|
| 5715 |
+
"step": 8140
|
| 5716 |
+
},
|
| 5717 |
+
{
|
| 5718 |
+
"epoch": 0.9924651799984778,
|
| 5719 |
+
"grad_norm": 320.5,
|
| 5720 |
+
"learning_rate": 9.844928200805614e-06,
|
| 5721 |
+
"loss": 21.6988,
|
| 5722 |
+
"step": 8150
|
| 5723 |
+
},
|
| 5724 |
+
{
|
| 5725 |
+
"epoch": 0.9936829286855925,
|
| 5726 |
+
"grad_norm": 485.0,
|
| 5727 |
+
"learning_rate": 9.844737928659363e-06,
|
| 5728 |
+
"loss": 21.8836,
|
| 5729 |
+
"step": 8160
|
| 5730 |
+
},
|
| 5731 |
+
{
|
| 5732 |
+
"epoch": 0.9949006773727072,
|
| 5733 |
+
"grad_norm": 595.0,
|
| 5734 |
+
"learning_rate": 9.844547656513112e-06,
|
| 5735 |
+
"loss": 21.7679,
|
| 5736 |
+
"step": 8170
|
| 5737 |
+
},
|
| 5738 |
+
{
|
| 5739 |
+
"epoch": 0.9961184260598219,
|
| 5740 |
+
"grad_norm": 466.0,
|
| 5741 |
+
"learning_rate": 9.84435738436686e-06,
|
| 5742 |
+
"loss": 22.1523,
|
| 5743 |
+
"step": 8180
|
| 5744 |
+
},
|
| 5745 |
+
{
|
| 5746 |
+
"epoch": 0.9973361747469366,
|
| 5747 |
+
"grad_norm": 303.5,
|
| 5748 |
+
"learning_rate": 9.84416711222061e-06,
|
| 5749 |
+
"loss": 21.9741,
|
| 5750 |
+
"step": 8190
|
| 5751 |
+
},
|
| 5752 |
+
{
|
| 5753 |
+
"epoch": 0.9985539234340512,
|
| 5754 |
+
"grad_norm": 300.5,
|
| 5755 |
+
"learning_rate": 9.84397684007436e-06,
|
| 5756 |
+
"loss": 21.9414,
|
| 5757 |
+
"step": 8200
|
| 5758 |
+
},
|
| 5759 |
+
{
|
| 5760 |
+
"epoch": 0.999771672121166,
|
| 5761 |
+
"grad_norm": 333.75,
|
| 5762 |
+
"learning_rate": 9.843786567928109e-06,
|
| 5763 |
+
"loss": 21.7997,
|
| 5764 |
+
"step": 8210
|
| 5765 |
}
|
| 5766 |
],
|
| 5767 |
"logging_steps": 10,
|
|
|
|
| 5776 |
"should_evaluate": false,
|
| 5777 |
"should_log": false,
|
| 5778 |
"should_save": true,
|
| 5779 |
+
"should_training_stop": true
|
| 5780 |
},
|
| 5781 |
"attributes": {}
|
| 5782 |
}
|
| 5783 |
},
|
| 5784 |
+
"total_flos": 2.8664974749173023e+18,
|
| 5785 |
"train_batch_size": 16,
|
| 5786 |
"trial_name": null,
|
| 5787 |
"trial_params": null
|