Training in progress, step 1303, 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 100059752
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e4513e890d84684f8b33eb5fa24d27ce4da4333d4b27c1fb555bfb79850aa060
|
| 3 |
size 100059752
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 51244404
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cc01379878ae9128b7373085be8d417ce51991b6873881c38392174190701425
|
| 3 |
size 51244404
|
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 sha256:83dcb968eefc2fe34af427267032a5b806dfd430cc9a31cf2e4ecdc4428bf987
|
| 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 sha256:9eabf0c977468cdbb0fb34466e43f850400f356cb7be4266d2d1ac09878380a5
|
| 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":
|
| 5 |
"eval_steps": 326,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -6885,6 +6885,2281 @@
|
|
| 6885 |
"eval_samples_per_second": 27.795,
|
| 6886 |
"eval_steps_per_second": 13.923,
|
| 6887 |
"step": 978
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6888 |
}
|
| 6889 |
],
|
| 6890 |
"logging_steps": 1,
|
|
@@ -6899,12 +9174,12 @@
|
|
| 6899 |
"should_evaluate": false,
|
| 6900 |
"should_log": false,
|
| 6901 |
"should_save": true,
|
| 6902 |
-
"should_training_stop":
|
| 6903 |
},
|
| 6904 |
"attributes": {}
|
| 6905 |
}
|
| 6906 |
},
|
| 6907 |
-
"total_flos":
|
| 6908 |
"train_batch_size": 2,
|
| 6909 |
"trial_name": null,
|
| 6910 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.000575926281436,
|
| 5 |
"eval_steps": 326,
|
| 6 |
+
"global_step": 1303,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 6885 |
"eval_samples_per_second": 27.795,
|
| 6886 |
"eval_steps_per_second": 13.923,
|
| 6887 |
"step": 978
|
| 6888 |
+
},
|
| 6889 |
+
{
|
| 6890 |
+
"epoch": 0.7517757727010943,
|
| 6891 |
+
"grad_norm": 1.2883607149124146,
|
| 6892 |
+
"learning_rate": 2.9418293126798603e-05,
|
| 6893 |
+
"loss": 1.1005,
|
| 6894 |
+
"step": 979
|
| 6895 |
+
},
|
| 6896 |
+
{
|
| 6897 |
+
"epoch": 0.7525436744096755,
|
| 6898 |
+
"grad_norm": 1.434813141822815,
|
| 6899 |
+
"learning_rate": 2.9246383624288387e-05,
|
| 6900 |
+
"loss": 1.201,
|
| 6901 |
+
"step": 980
|
| 6902 |
+
},
|
| 6903 |
+
{
|
| 6904 |
+
"epoch": 0.7533115761182568,
|
| 6905 |
+
"grad_norm": 2.126904249191284,
|
| 6906 |
+
"learning_rate": 2.9074891808937753e-05,
|
| 6907 |
+
"loss": 1.4774,
|
| 6908 |
+
"step": 981
|
| 6909 |
+
},
|
| 6910 |
+
{
|
| 6911 |
+
"epoch": 0.7540794778268382,
|
| 6912 |
+
"grad_norm": 1.5914756059646606,
|
| 6913 |
+
"learning_rate": 2.8903818693132077e-05,
|
| 6914 |
+
"loss": 1.3087,
|
| 6915 |
+
"step": 982
|
| 6916 |
+
},
|
| 6917 |
+
{
|
| 6918 |
+
"epoch": 0.7548473795354195,
|
| 6919 |
+
"grad_norm": 1.577763557434082,
|
| 6920 |
+
"learning_rate": 2.873316528678507e-05,
|
| 6921 |
+
"loss": 1.3263,
|
| 6922 |
+
"step": 983
|
| 6923 |
+
},
|
| 6924 |
+
{
|
| 6925 |
+
"epoch": 0.7556152812440008,
|
| 6926 |
+
"grad_norm": 1.5671124458312988,
|
| 6927 |
+
"learning_rate": 2.856293259733266e-05,
|
| 6928 |
+
"loss": 1.2893,
|
| 6929 |
+
"step": 984
|
| 6930 |
+
},
|
| 6931 |
+
{
|
| 6932 |
+
"epoch": 0.7563831829525821,
|
| 6933 |
+
"grad_norm": 1.6029667854309082,
|
| 6934 |
+
"learning_rate": 2.8393121629727138e-05,
|
| 6935 |
+
"loss": 1.1126,
|
| 6936 |
+
"step": 985
|
| 6937 |
+
},
|
| 6938 |
+
{
|
| 6939 |
+
"epoch": 0.7571510846611633,
|
| 6940 |
+
"grad_norm": 1.4934439659118652,
|
| 6941 |
+
"learning_rate": 2.8223733386431185e-05,
|
| 6942 |
+
"loss": 1.2581,
|
| 6943 |
+
"step": 986
|
| 6944 |
+
},
|
| 6945 |
+
{
|
| 6946 |
+
"epoch": 0.7579189863697446,
|
| 6947 |
+
"grad_norm": 1.327414870262146,
|
| 6948 |
+
"learning_rate": 2.8054768867411974e-05,
|
| 6949 |
+
"loss": 1.2253,
|
| 6950 |
+
"step": 987
|
| 6951 |
+
},
|
| 6952 |
+
{
|
| 6953 |
+
"epoch": 0.758686888078326,
|
| 6954 |
+
"grad_norm": 1.5715781450271606,
|
| 6955 |
+
"learning_rate": 2.788622907013526e-05,
|
| 6956 |
+
"loss": 1.1577,
|
| 6957 |
+
"step": 988
|
| 6958 |
+
},
|
| 6959 |
+
{
|
| 6960 |
+
"epoch": 0.7594547897869073,
|
| 6961 |
+
"grad_norm": 1.4220985174179077,
|
| 6962 |
+
"learning_rate": 2.7718114989559552e-05,
|
| 6963 |
+
"loss": 1.1358,
|
| 6964 |
+
"step": 989
|
| 6965 |
+
},
|
| 6966 |
+
{
|
| 6967 |
+
"epoch": 0.7602226914954886,
|
| 6968 |
+
"grad_norm": 1.4818079471588135,
|
| 6969 |
+
"learning_rate": 2.7550427618130127e-05,
|
| 6970 |
+
"loss": 1.2783,
|
| 6971 |
+
"step": 990
|
| 6972 |
+
},
|
| 6973 |
+
{
|
| 6974 |
+
"epoch": 0.7609905932040699,
|
| 6975 |
+
"grad_norm": 1.394391655921936,
|
| 6976 |
+
"learning_rate": 2.738316794577315e-05,
|
| 6977 |
+
"loss": 1.2429,
|
| 6978 |
+
"step": 991
|
| 6979 |
+
},
|
| 6980 |
+
{
|
| 6981 |
+
"epoch": 0.7617584949126511,
|
| 6982 |
+
"grad_norm": 1.7591447830200195,
|
| 6983 |
+
"learning_rate": 2.7216336959890076e-05,
|
| 6984 |
+
"loss": 1.5355,
|
| 6985 |
+
"step": 992
|
| 6986 |
+
},
|
| 6987 |
+
{
|
| 6988 |
+
"epoch": 0.7625263966212324,
|
| 6989 |
+
"grad_norm": 1.226349949836731,
|
| 6990 |
+
"learning_rate": 2.704993564535152e-05,
|
| 6991 |
+
"loss": 1.0599,
|
| 6992 |
+
"step": 993
|
| 6993 |
+
},
|
| 6994 |
+
{
|
| 6995 |
+
"epoch": 0.7632942983298138,
|
| 6996 |
+
"grad_norm": 1.3571503162384033,
|
| 6997 |
+
"learning_rate": 2.688396498449164e-05,
|
| 6998 |
+
"loss": 0.9554,
|
| 6999 |
+
"step": 994
|
| 7000 |
+
},
|
| 7001 |
+
{
|
| 7002 |
+
"epoch": 0.7640622000383951,
|
| 7003 |
+
"grad_norm": 1.546478509902954,
|
| 7004 |
+
"learning_rate": 2.671842595710219e-05,
|
| 7005 |
+
"loss": 1.2609,
|
| 7006 |
+
"step": 995
|
| 7007 |
+
},
|
| 7008 |
+
{
|
| 7009 |
+
"epoch": 0.7648301017469764,
|
| 7010 |
+
"grad_norm": 1.304118275642395,
|
| 7011 |
+
"learning_rate": 2.655331954042699e-05,
|
| 7012 |
+
"loss": 1.1655,
|
| 7013 |
+
"step": 996
|
| 7014 |
+
},
|
| 7015 |
+
{
|
| 7016 |
+
"epoch": 0.7655980034555577,
|
| 7017 |
+
"grad_norm": 1.4672322273254395,
|
| 7018 |
+
"learning_rate": 2.638864670915572e-05,
|
| 7019 |
+
"loss": 1.1936,
|
| 7020 |
+
"step": 997
|
| 7021 |
+
},
|
| 7022 |
+
{
|
| 7023 |
+
"epoch": 0.7663659051641389,
|
| 7024 |
+
"grad_norm": 1.1511002779006958,
|
| 7025 |
+
"learning_rate": 2.622440843541869e-05,
|
| 7026 |
+
"loss": 1.1561,
|
| 7027 |
+
"step": 998
|
| 7028 |
+
},
|
| 7029 |
+
{
|
| 7030 |
+
"epoch": 0.7671338068727203,
|
| 7031 |
+
"grad_norm": 1.617389440536499,
|
| 7032 |
+
"learning_rate": 2.6060605688780694e-05,
|
| 7033 |
+
"loss": 1.2096,
|
| 7034 |
+
"step": 999
|
| 7035 |
+
},
|
| 7036 |
+
{
|
| 7037 |
+
"epoch": 0.7679017085813016,
|
| 7038 |
+
"grad_norm": 1.7287838459014893,
|
| 7039 |
+
"learning_rate": 2.5897239436235466e-05,
|
| 7040 |
+
"loss": 1.3722,
|
| 7041 |
+
"step": 1000
|
| 7042 |
+
},
|
| 7043 |
+
{
|
| 7044 |
+
"epoch": 0.7686696102898829,
|
| 7045 |
+
"grad_norm": 1.4887206554412842,
|
| 7046 |
+
"learning_rate": 2.5734310642199943e-05,
|
| 7047 |
+
"loss": 1.2082,
|
| 7048 |
+
"step": 1001
|
| 7049 |
+
},
|
| 7050 |
+
{
|
| 7051 |
+
"epoch": 0.7694375119984642,
|
| 7052 |
+
"grad_norm": 1.1259337663650513,
|
| 7053 |
+
"learning_rate": 2.557182026850855e-05,
|
| 7054 |
+
"loss": 0.937,
|
| 7055 |
+
"step": 1002
|
| 7056 |
+
},
|
| 7057 |
+
{
|
| 7058 |
+
"epoch": 0.7702054137070455,
|
| 7059 |
+
"grad_norm": 1.7888356447219849,
|
| 7060 |
+
"learning_rate": 2.5409769274407637e-05,
|
| 7061 |
+
"loss": 1.2036,
|
| 7062 |
+
"step": 1003
|
| 7063 |
+
},
|
| 7064 |
+
{
|
| 7065 |
+
"epoch": 0.7709733154156267,
|
| 7066 |
+
"grad_norm": 1.3678765296936035,
|
| 7067 |
+
"learning_rate": 2.524815861654952e-05,
|
| 7068 |
+
"loss": 1.183,
|
| 7069 |
+
"step": 1004
|
| 7070 |
+
},
|
| 7071 |
+
{
|
| 7072 |
+
"epoch": 0.7717412171242081,
|
| 7073 |
+
"grad_norm": 1.4759341478347778,
|
| 7074 |
+
"learning_rate": 2.5086989248987248e-05,
|
| 7075 |
+
"loss": 1.4874,
|
| 7076 |
+
"step": 1005
|
| 7077 |
+
},
|
| 7078 |
+
{
|
| 7079 |
+
"epoch": 0.7725091188327894,
|
| 7080 |
+
"grad_norm": 1.6217724084854126,
|
| 7081 |
+
"learning_rate": 2.492626212316862e-05,
|
| 7082 |
+
"loss": 1.3213,
|
| 7083 |
+
"step": 1006
|
| 7084 |
+
},
|
| 7085 |
+
{
|
| 7086 |
+
"epoch": 0.7732770205413707,
|
| 7087 |
+
"grad_norm": 1.735228180885315,
|
| 7088 |
+
"learning_rate": 2.476597818793075e-05,
|
| 7089 |
+
"loss": 1.4892,
|
| 7090 |
+
"step": 1007
|
| 7091 |
+
},
|
| 7092 |
+
{
|
| 7093 |
+
"epoch": 0.774044922249952,
|
| 7094 |
+
"grad_norm": 1.6505582332611084,
|
| 7095 |
+
"learning_rate": 2.460613838949437e-05,
|
| 7096 |
+
"loss": 1.3946,
|
| 7097 |
+
"step": 1008
|
| 7098 |
+
},
|
| 7099 |
+
{
|
| 7100 |
+
"epoch": 0.7748128239585333,
|
| 7101 |
+
"grad_norm": 1.5245342254638672,
|
| 7102 |
+
"learning_rate": 2.444674367145845e-05,
|
| 7103 |
+
"loss": 1.3344,
|
| 7104 |
+
"step": 1009
|
| 7105 |
+
},
|
| 7106 |
+
{
|
| 7107 |
+
"epoch": 0.7755807256671146,
|
| 7108 |
+
"grad_norm": 1.4345464706420898,
|
| 7109 |
+
"learning_rate": 2.4287794974794208e-05,
|
| 7110 |
+
"loss": 1.304,
|
| 7111 |
+
"step": 1010
|
| 7112 |
+
},
|
| 7113 |
+
{
|
| 7114 |
+
"epoch": 0.7763486273756959,
|
| 7115 |
+
"grad_norm": 1.120750069618225,
|
| 7116 |
+
"learning_rate": 2.4129293237840066e-05,
|
| 7117 |
+
"loss": 1.031,
|
| 7118 |
+
"step": 1011
|
| 7119 |
+
},
|
| 7120 |
+
{
|
| 7121 |
+
"epoch": 0.7771165290842772,
|
| 7122 |
+
"grad_norm": 1.4984389543533325,
|
| 7123 |
+
"learning_rate": 2.397123939629574e-05,
|
| 7124 |
+
"loss": 1.4389,
|
| 7125 |
+
"step": 1012
|
| 7126 |
+
},
|
| 7127 |
+
{
|
| 7128 |
+
"epoch": 0.7778844307928585,
|
| 7129 |
+
"grad_norm": 1.4427404403686523,
|
| 7130 |
+
"learning_rate": 2.3813634383216853e-05,
|
| 7131 |
+
"loss": 1.1667,
|
| 7132 |
+
"step": 1013
|
| 7133 |
+
},
|
| 7134 |
+
{
|
| 7135 |
+
"epoch": 0.7786523325014398,
|
| 7136 |
+
"grad_norm": 1.6551741361618042,
|
| 7137 |
+
"learning_rate": 2.3656479129009422e-05,
|
| 7138 |
+
"loss": 1.3006,
|
| 7139 |
+
"step": 1014
|
| 7140 |
+
},
|
| 7141 |
+
{
|
| 7142 |
+
"epoch": 0.7794202342100212,
|
| 7143 |
+
"grad_norm": 1.3143059015274048,
|
| 7144 |
+
"learning_rate": 2.3499774561424327e-05,
|
| 7145 |
+
"loss": 1.134,
|
| 7146 |
+
"step": 1015
|
| 7147 |
+
},
|
| 7148 |
+
{
|
| 7149 |
+
"epoch": 0.7801881359186024,
|
| 7150 |
+
"grad_norm": 1.7605934143066406,
|
| 7151 |
+
"learning_rate": 2.3343521605551967e-05,
|
| 7152 |
+
"loss": 1.1622,
|
| 7153 |
+
"step": 1016
|
| 7154 |
+
},
|
| 7155 |
+
{
|
| 7156 |
+
"epoch": 0.7809560376271837,
|
| 7157 |
+
"grad_norm": 1.5233832597732544,
|
| 7158 |
+
"learning_rate": 2.3187721183816503e-05,
|
| 7159 |
+
"loss": 1.2939,
|
| 7160 |
+
"step": 1017
|
| 7161 |
+
},
|
| 7162 |
+
{
|
| 7163 |
+
"epoch": 0.781723939335765,
|
| 7164 |
+
"grad_norm": 1.3118071556091309,
|
| 7165 |
+
"learning_rate": 2.303237421597082e-05,
|
| 7166 |
+
"loss": 0.9272,
|
| 7167 |
+
"step": 1018
|
| 7168 |
+
},
|
| 7169 |
+
{
|
| 7170 |
+
"epoch": 0.7824918410443463,
|
| 7171 |
+
"grad_norm": 1.559906005859375,
|
| 7172 |
+
"learning_rate": 2.2877481619090734e-05,
|
| 7173 |
+
"loss": 1.4749,
|
| 7174 |
+
"step": 1019
|
| 7175 |
+
},
|
| 7176 |
+
{
|
| 7177 |
+
"epoch": 0.7832597427529276,
|
| 7178 |
+
"grad_norm": 1.6087397336959839,
|
| 7179 |
+
"learning_rate": 2.2723044307569775e-05,
|
| 7180 |
+
"loss": 1.2059,
|
| 7181 |
+
"step": 1020
|
| 7182 |
+
},
|
| 7183 |
+
{
|
| 7184 |
+
"epoch": 0.784027644461509,
|
| 7185 |
+
"grad_norm": 1.5750290155410767,
|
| 7186 |
+
"learning_rate": 2.2569063193113716e-05,
|
| 7187 |
+
"loss": 1.2115,
|
| 7188 |
+
"step": 1021
|
| 7189 |
+
},
|
| 7190 |
+
{
|
| 7191 |
+
"epoch": 0.7847955461700902,
|
| 7192 |
+
"grad_norm": 1.4662597179412842,
|
| 7193 |
+
"learning_rate": 2.2415539184735323e-05,
|
| 7194 |
+
"loss": 0.9787,
|
| 7195 |
+
"step": 1022
|
| 7196 |
+
},
|
| 7197 |
+
{
|
| 7198 |
+
"epoch": 0.7855634478786715,
|
| 7199 |
+
"grad_norm": 1.5807639360427856,
|
| 7200 |
+
"learning_rate": 2.2262473188748667e-05,
|
| 7201 |
+
"loss": 1.5,
|
| 7202 |
+
"step": 1023
|
| 7203 |
+
},
|
| 7204 |
+
{
|
| 7205 |
+
"epoch": 0.7863313495872528,
|
| 7206 |
+
"grad_norm": 1.4712680578231812,
|
| 7207 |
+
"learning_rate": 2.210986610876421e-05,
|
| 7208 |
+
"loss": 1.1427,
|
| 7209 |
+
"step": 1024
|
| 7210 |
+
},
|
| 7211 |
+
{
|
| 7212 |
+
"epoch": 0.7870992512958341,
|
| 7213 |
+
"grad_norm": 1.7284588813781738,
|
| 7214 |
+
"learning_rate": 2.1957718845683106e-05,
|
| 7215 |
+
"loss": 0.9901,
|
| 7216 |
+
"step": 1025
|
| 7217 |
+
},
|
| 7218 |
+
{
|
| 7219 |
+
"epoch": 0.7878671530044155,
|
| 7220 |
+
"grad_norm": 1.7499277591705322,
|
| 7221 |
+
"learning_rate": 2.180603229769208e-05,
|
| 7222 |
+
"loss": 1.7659,
|
| 7223 |
+
"step": 1026
|
| 7224 |
+
},
|
| 7225 |
+
{
|
| 7226 |
+
"epoch": 0.7886350547129968,
|
| 7227 |
+
"grad_norm": 2.04580020904541,
|
| 7228 |
+
"learning_rate": 2.1654807360258068e-05,
|
| 7229 |
+
"loss": 1.4249,
|
| 7230 |
+
"step": 1027
|
| 7231 |
+
},
|
| 7232 |
+
{
|
| 7233 |
+
"epoch": 0.789402956421578,
|
| 7234 |
+
"grad_norm": 1.8796188831329346,
|
| 7235 |
+
"learning_rate": 2.15040449261229e-05,
|
| 7236 |
+
"loss": 1.4763,
|
| 7237 |
+
"step": 1028
|
| 7238 |
+
},
|
| 7239 |
+
{
|
| 7240 |
+
"epoch": 0.7901708581301593,
|
| 7241 |
+
"grad_norm": 1.7241255044937134,
|
| 7242 |
+
"learning_rate": 2.1353745885298103e-05,
|
| 7243 |
+
"loss": 1.4569,
|
| 7244 |
+
"step": 1029
|
| 7245 |
+
},
|
| 7246 |
+
{
|
| 7247 |
+
"epoch": 0.7909387598387406,
|
| 7248 |
+
"grad_norm": 1.2481935024261475,
|
| 7249 |
+
"learning_rate": 2.120391112505955e-05,
|
| 7250 |
+
"loss": 1.1742,
|
| 7251 |
+
"step": 1030
|
| 7252 |
+
},
|
| 7253 |
+
{
|
| 7254 |
+
"epoch": 0.7917066615473219,
|
| 7255 |
+
"grad_norm": 1.440290093421936,
|
| 7256 |
+
"learning_rate": 2.1054541529942374e-05,
|
| 7257 |
+
"loss": 1.243,
|
| 7258 |
+
"step": 1031
|
| 7259 |
+
},
|
| 7260 |
+
{
|
| 7261 |
+
"epoch": 0.7924745632559033,
|
| 7262 |
+
"grad_norm": 1.8699524402618408,
|
| 7263 |
+
"learning_rate": 2.090563798173557e-05,
|
| 7264 |
+
"loss": 1.3004,
|
| 7265 |
+
"step": 1032
|
| 7266 |
+
},
|
| 7267 |
+
{
|
| 7268 |
+
"epoch": 0.7932424649644846,
|
| 7269 |
+
"grad_norm": 1.7997124195098877,
|
| 7270 |
+
"learning_rate": 2.0757201359476884e-05,
|
| 7271 |
+
"loss": 1.5915,
|
| 7272 |
+
"step": 1033
|
| 7273 |
+
},
|
| 7274 |
+
{
|
| 7275 |
+
"epoch": 0.7940103666730658,
|
| 7276 |
+
"grad_norm": 1.4160553216934204,
|
| 7277 |
+
"learning_rate": 2.060923253944761e-05,
|
| 7278 |
+
"loss": 1.3386,
|
| 7279 |
+
"step": 1034
|
| 7280 |
+
},
|
| 7281 |
+
{
|
| 7282 |
+
"epoch": 0.7947782683816471,
|
| 7283 |
+
"grad_norm": 1.4589799642562866,
|
| 7284 |
+
"learning_rate": 2.0461732395167475e-05,
|
| 7285 |
+
"loss": 1.3019,
|
| 7286 |
+
"step": 1035
|
| 7287 |
+
},
|
| 7288 |
+
{
|
| 7289 |
+
"epoch": 0.7955461700902284,
|
| 7290 |
+
"grad_norm": 1.8283833265304565,
|
| 7291 |
+
"learning_rate": 2.03147017973893e-05,
|
| 7292 |
+
"loss": 1.3423,
|
| 7293 |
+
"step": 1036
|
| 7294 |
+
},
|
| 7295 |
+
{
|
| 7296 |
+
"epoch": 0.7963140717988098,
|
| 7297 |
+
"grad_norm": 1.4732909202575684,
|
| 7298 |
+
"learning_rate": 2.0168141614094126e-05,
|
| 7299 |
+
"loss": 1.2183,
|
| 7300 |
+
"step": 1037
|
| 7301 |
+
},
|
| 7302 |
+
{
|
| 7303 |
+
"epoch": 0.7970819735073911,
|
| 7304 |
+
"grad_norm": 1.5215457677841187,
|
| 7305 |
+
"learning_rate": 2.0022052710485874e-05,
|
| 7306 |
+
"loss": 1.2761,
|
| 7307 |
+
"step": 1038
|
| 7308 |
+
},
|
| 7309 |
+
{
|
| 7310 |
+
"epoch": 0.7978498752159724,
|
| 7311 |
+
"grad_norm": 1.812894344329834,
|
| 7312 |
+
"learning_rate": 1.9876435948986228e-05,
|
| 7313 |
+
"loss": 1.4416,
|
| 7314 |
+
"step": 1039
|
| 7315 |
+
},
|
| 7316 |
+
{
|
| 7317 |
+
"epoch": 0.7986177769245536,
|
| 7318 |
+
"grad_norm": 1.212895154953003,
|
| 7319 |
+
"learning_rate": 1.973129218922981e-05,
|
| 7320 |
+
"loss": 1.0733,
|
| 7321 |
+
"step": 1040
|
| 7322 |
+
},
|
| 7323 |
+
{
|
| 7324 |
+
"epoch": 0.7993856786331349,
|
| 7325 |
+
"grad_norm": 1.5380189418792725,
|
| 7326 |
+
"learning_rate": 1.95866222880588e-05,
|
| 7327 |
+
"loss": 1.4282,
|
| 7328 |
+
"step": 1041
|
| 7329 |
+
},
|
| 7330 |
+
{
|
| 7331 |
+
"epoch": 0.8001535803417162,
|
| 7332 |
+
"grad_norm": 1.671745777130127,
|
| 7333 |
+
"learning_rate": 1.9442427099518022e-05,
|
| 7334 |
+
"loss": 1.4183,
|
| 7335 |
+
"step": 1042
|
| 7336 |
+
},
|
| 7337 |
+
{
|
| 7338 |
+
"epoch": 0.8009214820502976,
|
| 7339 |
+
"grad_norm": 1.6439645290374756,
|
| 7340 |
+
"learning_rate": 1.9298707474849843e-05,
|
| 7341 |
+
"loss": 1.2435,
|
| 7342 |
+
"step": 1043
|
| 7343 |
+
},
|
| 7344 |
+
{
|
| 7345 |
+
"epoch": 0.8016893837588789,
|
| 7346 |
+
"grad_norm": 1.3178609609603882,
|
| 7347 |
+
"learning_rate": 1.9155464262489298e-05,
|
| 7348 |
+
"loss": 0.843,
|
| 7349 |
+
"step": 1044
|
| 7350 |
+
},
|
| 7351 |
+
{
|
| 7352 |
+
"epoch": 0.8024572854674602,
|
| 7353 |
+
"grad_norm": 1.7350953817367554,
|
| 7354 |
+
"learning_rate": 1.9012698308058852e-05,
|
| 7355 |
+
"loss": 1.4906,
|
| 7356 |
+
"step": 1045
|
| 7357 |
+
},
|
| 7358 |
+
{
|
| 7359 |
+
"epoch": 0.8032251871760415,
|
| 7360 |
+
"grad_norm": 1.2103064060211182,
|
| 7361 |
+
"learning_rate": 1.8870410454363573e-05,
|
| 7362 |
+
"loss": 0.9144,
|
| 7363 |
+
"step": 1046
|
| 7364 |
+
},
|
| 7365 |
+
{
|
| 7366 |
+
"epoch": 0.8039930888846227,
|
| 7367 |
+
"grad_norm": 1.1046971082687378,
|
| 7368 |
+
"learning_rate": 1.872860154138608e-05,
|
| 7369 |
+
"loss": 0.9619,
|
| 7370 |
+
"step": 1047
|
| 7371 |
+
},
|
| 7372 |
+
{
|
| 7373 |
+
"epoch": 0.804760990593204,
|
| 7374 |
+
"grad_norm": 1.832051157951355,
|
| 7375 |
+
"learning_rate": 1.858727240628171e-05,
|
| 7376 |
+
"loss": 1.4815,
|
| 7377 |
+
"step": 1048
|
| 7378 |
+
},
|
| 7379 |
+
{
|
| 7380 |
+
"epoch": 0.8055288923017854,
|
| 7381 |
+
"grad_norm": 1.1102668046951294,
|
| 7382 |
+
"learning_rate": 1.8446423883373286e-05,
|
| 7383 |
+
"loss": 0.9905,
|
| 7384 |
+
"step": 1049
|
| 7385 |
+
},
|
| 7386 |
+
{
|
| 7387 |
+
"epoch": 0.8062967940103667,
|
| 7388 |
+
"grad_norm": 1.155604600906372,
|
| 7389 |
+
"learning_rate": 1.8306056804146575e-05,
|
| 7390 |
+
"loss": 1.0172,
|
| 7391 |
+
"step": 1050
|
| 7392 |
+
},
|
| 7393 |
+
{
|
| 7394 |
+
"epoch": 0.807064695718948,
|
| 7395 |
+
"grad_norm": 1.4365675449371338,
|
| 7396 |
+
"learning_rate": 1.816617199724512e-05,
|
| 7397 |
+
"loss": 1.3202,
|
| 7398 |
+
"step": 1051
|
| 7399 |
+
},
|
| 7400 |
+
{
|
| 7401 |
+
"epoch": 0.8078325974275293,
|
| 7402 |
+
"grad_norm": 1.2660962343215942,
|
| 7403 |
+
"learning_rate": 1.8026770288465323e-05,
|
| 7404 |
+
"loss": 1.2372,
|
| 7405 |
+
"step": 1052
|
| 7406 |
+
},
|
| 7407 |
+
{
|
| 7408 |
+
"epoch": 0.8086004991361105,
|
| 7409 |
+
"grad_norm": 1.484471082687378,
|
| 7410 |
+
"learning_rate": 1.7887852500751822e-05,
|
| 7411 |
+
"loss": 1.2489,
|
| 7412 |
+
"step": 1053
|
| 7413 |
+
},
|
| 7414 |
+
{
|
| 7415 |
+
"epoch": 0.8093684008446919,
|
| 7416 |
+
"grad_norm": 1.62258780002594,
|
| 7417 |
+
"learning_rate": 1.7749419454192373e-05,
|
| 7418 |
+
"loss": 1.3631,
|
| 7419 |
+
"step": 1054
|
| 7420 |
+
},
|
| 7421 |
+
{
|
| 7422 |
+
"epoch": 0.8101363025532732,
|
| 7423 |
+
"grad_norm": 1.2742317914962769,
|
| 7424 |
+
"learning_rate": 1.7611471966013127e-05,
|
| 7425 |
+
"loss": 1.0303,
|
| 7426 |
+
"step": 1055
|
| 7427 |
+
},
|
| 7428 |
+
{
|
| 7429 |
+
"epoch": 0.8109042042618545,
|
| 7430 |
+
"grad_norm": 1.770555853843689,
|
| 7431 |
+
"learning_rate": 1.7474010850573775e-05,
|
| 7432 |
+
"loss": 1.4317,
|
| 7433 |
+
"step": 1056
|
| 7434 |
+
},
|
| 7435 |
+
{
|
| 7436 |
+
"epoch": 0.8116721059704358,
|
| 7437 |
+
"grad_norm": 1.3370784521102905,
|
| 7438 |
+
"learning_rate": 1.7337036919362827e-05,
|
| 7439 |
+
"loss": 1.1577,
|
| 7440 |
+
"step": 1057
|
| 7441 |
+
},
|
| 7442 |
+
{
|
| 7443 |
+
"epoch": 0.8124400076790171,
|
| 7444 |
+
"grad_norm": 1.4367464780807495,
|
| 7445 |
+
"learning_rate": 1.7200550980992647e-05,
|
| 7446 |
+
"loss": 1.3085,
|
| 7447 |
+
"step": 1058
|
| 7448 |
+
},
|
| 7449 |
+
{
|
| 7450 |
+
"epoch": 0.8132079093875983,
|
| 7451 |
+
"grad_norm": 1.363452434539795,
|
| 7452 |
+
"learning_rate": 1.706455384119485e-05,
|
| 7453 |
+
"loss": 1.1423,
|
| 7454 |
+
"step": 1059
|
| 7455 |
+
},
|
| 7456 |
+
{
|
| 7457 |
+
"epoch": 0.8139758110961797,
|
| 7458 |
+
"grad_norm": 1.3414223194122314,
|
| 7459 |
+
"learning_rate": 1.6929046302815443e-05,
|
| 7460 |
+
"loss": 1.2153,
|
| 7461 |
+
"step": 1060
|
| 7462 |
+
},
|
| 7463 |
+
{
|
| 7464 |
+
"epoch": 0.814743712804761,
|
| 7465 |
+
"grad_norm": 1.577257513999939,
|
| 7466 |
+
"learning_rate": 1.6794029165810133e-05,
|
| 7467 |
+
"loss": 1.357,
|
| 7468 |
+
"step": 1061
|
| 7469 |
+
},
|
| 7470 |
+
{
|
| 7471 |
+
"epoch": 0.8155116145133423,
|
| 7472 |
+
"grad_norm": 1.594196081161499,
|
| 7473 |
+
"learning_rate": 1.665950322723957e-05,
|
| 7474 |
+
"loss": 1.2597,
|
| 7475 |
+
"step": 1062
|
| 7476 |
+
},
|
| 7477 |
+
{
|
| 7478 |
+
"epoch": 0.8162795162219236,
|
| 7479 |
+
"grad_norm": 1.5210363864898682,
|
| 7480 |
+
"learning_rate": 1.652546928126466e-05,
|
| 7481 |
+
"loss": 1.2459,
|
| 7482 |
+
"step": 1063
|
| 7483 |
+
},
|
| 7484 |
+
{
|
| 7485 |
+
"epoch": 0.817047417930505,
|
| 7486 |
+
"grad_norm": 1.3744914531707764,
|
| 7487 |
+
"learning_rate": 1.6391928119141965e-05,
|
| 7488 |
+
"loss": 1.236,
|
| 7489 |
+
"step": 1064
|
| 7490 |
+
},
|
| 7491 |
+
{
|
| 7492 |
+
"epoch": 0.8178153196390862,
|
| 7493 |
+
"grad_norm": 1.56074857711792,
|
| 7494 |
+
"learning_rate": 1.625888052921878e-05,
|
| 7495 |
+
"loss": 1.2514,
|
| 7496 |
+
"step": 1065
|
| 7497 |
+
},
|
| 7498 |
+
{
|
| 7499 |
+
"epoch": 0.8185832213476675,
|
| 7500 |
+
"grad_norm": 1.3938485383987427,
|
| 7501 |
+
"learning_rate": 1.612632729692881e-05,
|
| 7502 |
+
"loss": 1.3724,
|
| 7503 |
+
"step": 1066
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 0.8193511230562488,
|
| 7507 |
+
"grad_norm": 1.2004398107528687,
|
| 7508 |
+
"learning_rate": 1.599426920478726e-05,
|
| 7509 |
+
"loss": 0.8796,
|
| 7510 |
+
"step": 1067
|
| 7511 |
+
},
|
| 7512 |
+
{
|
| 7513 |
+
"epoch": 0.8201190247648301,
|
| 7514 |
+
"grad_norm": 1.6214113235473633,
|
| 7515 |
+
"learning_rate": 1.586270703238637e-05,
|
| 7516 |
+
"loss": 1.3691,
|
| 7517 |
+
"step": 1068
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.8208869264734114,
|
| 7521 |
+
"grad_norm": 1.7719411849975586,
|
| 7522 |
+
"learning_rate": 1.573164155639073e-05,
|
| 7523 |
+
"loss": 1.5033,
|
| 7524 |
+
"step": 1069
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.8216548281819928,
|
| 7528 |
+
"grad_norm": 1.2365126609802246,
|
| 7529 |
+
"learning_rate": 1.560107355053282e-05,
|
| 7530 |
+
"loss": 1.1611,
|
| 7531 |
+
"step": 1070
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.822422729890574,
|
| 7535 |
+
"grad_norm": 1.804200291633606,
|
| 7536 |
+
"learning_rate": 1.5471003785608184e-05,
|
| 7537 |
+
"loss": 1.2228,
|
| 7538 |
+
"step": 1071
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.8231906315991553,
|
| 7542 |
+
"grad_norm": 3.12967586517334,
|
| 7543 |
+
"learning_rate": 1.534143302947123e-05,
|
| 7544 |
+
"loss": 1.0455,
|
| 7545 |
+
"step": 1072
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 0.8239585333077366,
|
| 7549 |
+
"grad_norm": 1.7391802072525024,
|
| 7550 |
+
"learning_rate": 1.5212362047030427e-05,
|
| 7551 |
+
"loss": 1.4632,
|
| 7552 |
+
"step": 1073
|
| 7553 |
+
},
|
| 7554 |
+
{
|
| 7555 |
+
"epoch": 0.8247264350163179,
|
| 7556 |
+
"grad_norm": 1.4144318103790283,
|
| 7557 |
+
"learning_rate": 1.5083791600243857e-05,
|
| 7558 |
+
"loss": 1.2948,
|
| 7559 |
+
"step": 1074
|
| 7560 |
+
},
|
| 7561 |
+
{
|
| 7562 |
+
"epoch": 0.8254943367248992,
|
| 7563 |
+
"grad_norm": 1.2028521299362183,
|
| 7564 |
+
"learning_rate": 1.4955722448114807e-05,
|
| 7565 |
+
"loss": 0.9203,
|
| 7566 |
+
"step": 1075
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 0.8262622384334806,
|
| 7570 |
+
"grad_norm": 1.4866068363189697,
|
| 7571 |
+
"learning_rate": 1.4828155346687123e-05,
|
| 7572 |
+
"loss": 1.3486,
|
| 7573 |
+
"step": 1076
|
| 7574 |
+
},
|
| 7575 |
+
{
|
| 7576 |
+
"epoch": 0.8270301401420618,
|
| 7577 |
+
"grad_norm": 1.471641182899475,
|
| 7578 |
+
"learning_rate": 1.4701091049040994e-05,
|
| 7579 |
+
"loss": 1.1633,
|
| 7580 |
+
"step": 1077
|
| 7581 |
+
},
|
| 7582 |
+
{
|
| 7583 |
+
"epoch": 0.8277980418506431,
|
| 7584 |
+
"grad_norm": 1.3013348579406738,
|
| 7585 |
+
"learning_rate": 1.4574530305288158e-05,
|
| 7586 |
+
"loss": 0.9819,
|
| 7587 |
+
"step": 1078
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 0.8285659435592244,
|
| 7591 |
+
"grad_norm": 1.5000224113464355,
|
| 7592 |
+
"learning_rate": 1.4448473862567857e-05,
|
| 7593 |
+
"loss": 1.1458,
|
| 7594 |
+
"step": 1079
|
| 7595 |
+
},
|
| 7596 |
+
{
|
| 7597 |
+
"epoch": 0.8293338452678057,
|
| 7598 |
+
"grad_norm": 1.4229687452316284,
|
| 7599 |
+
"learning_rate": 1.4322922465042132e-05,
|
| 7600 |
+
"loss": 1.3797,
|
| 7601 |
+
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.8301017469763871,
|
| 7605 |
+
"grad_norm": 1.3966771364212036,
|
| 7606 |
+
"learning_rate": 1.4197876853891557e-05,
|
| 7607 |
+
"loss": 1.4436,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.8308696486849684,
|
| 7612 |
+
"grad_norm": 1.3111258745193481,
|
| 7613 |
+
"learning_rate": 1.4073337767310834e-05,
|
| 7614 |
+
"loss": 1.175,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.8316375503935496,
|
| 7619 |
+
"grad_norm": 1.4146270751953125,
|
| 7620 |
+
"learning_rate": 1.3949305940504541e-05,
|
| 7621 |
+
"loss": 1.1575,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.8324054521021309,
|
| 7626 |
+
"grad_norm": 1.2625914812088013,
|
| 7627 |
+
"learning_rate": 1.3825782105682527e-05,
|
| 7628 |
+
"loss": 0.984,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.8331733538107122,
|
| 7633 |
+
"grad_norm": 1.2037335634231567,
|
| 7634 |
+
"learning_rate": 1.3702766992055927e-05,
|
| 7635 |
+
"loss": 1.0908,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.8339412555192935,
|
| 7640 |
+
"grad_norm": 1.384108304977417,
|
| 7641 |
+
"learning_rate": 1.3580261325832578e-05,
|
| 7642 |
+
"loss": 1.4079,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.8347091572278749,
|
| 7647 |
+
"grad_norm": 1.0640181303024292,
|
| 7648 |
+
"learning_rate": 1.3458265830212891e-05,
|
| 7649 |
+
"loss": 0.68,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.8354770589364562,
|
| 7654 |
+
"grad_norm": 1.6578110456466675,
|
| 7655 |
+
"learning_rate": 1.333678122538553e-05,
|
| 7656 |
+
"loss": 1.1659,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.8362449606450374,
|
| 7661 |
+
"grad_norm": 1.3299002647399902,
|
| 7662 |
+
"learning_rate": 1.321580822852313e-05,
|
| 7663 |
+
"loss": 1.1961,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.8370128623536187,
|
| 7668 |
+
"grad_norm": 1.5801900625228882,
|
| 7669 |
+
"learning_rate": 1.3095347553778193e-05,
|
| 7670 |
+
"loss": 1.3358,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.8377807640622,
|
| 7675 |
+
"grad_norm": 1.6045485734939575,
|
| 7676 |
+
"learning_rate": 1.2975399912278608e-05,
|
| 7677 |
+
"loss": 1.3452,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.8385486657707814,
|
| 7682 |
+
"grad_norm": 1.3222181797027588,
|
| 7683 |
+
"learning_rate": 1.2855966012123822e-05,
|
| 7684 |
+
"loss": 1.0153,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.8393165674793627,
|
| 7689 |
+
"grad_norm": 1.5065910816192627,
|
| 7690 |
+
"learning_rate": 1.2737046558380305e-05,
|
| 7691 |
+
"loss": 1.3556,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.840084469187944,
|
| 7696 |
+
"grad_norm": 1.6371604204177856,
|
| 7697 |
+
"learning_rate": 1.2618642253077628e-05,
|
| 7698 |
+
"loss": 1.3203,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.8408523708965252,
|
| 7703 |
+
"grad_norm": 1.801613688468933,
|
| 7704 |
+
"learning_rate": 1.2500753795204157e-05,
|
| 7705 |
+
"loss": 1.272,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.8416202726051065,
|
| 7710 |
+
"grad_norm": 1.6735864877700806,
|
| 7711 |
+
"learning_rate": 1.2383381880703138e-05,
|
| 7712 |
+
"loss": 1.2709,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.8423881743136878,
|
| 7717 |
+
"grad_norm": 1.3746118545532227,
|
| 7718 |
+
"learning_rate": 1.2266527202468248e-05,
|
| 7719 |
+
"loss": 1.164,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.8431560760222692,
|
| 7724 |
+
"grad_norm": 1.5882354974746704,
|
| 7725 |
+
"learning_rate": 1.2150190450339915e-05,
|
| 7726 |
+
"loss": 0.8384,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.8439239777308505,
|
| 7731 |
+
"grad_norm": 1.492803931236267,
|
| 7732 |
+
"learning_rate": 1.2034372311100905e-05,
|
| 7733 |
+
"loss": 1.4302,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.8446918794394318,
|
| 7738 |
+
"grad_norm": 1.868151307106018,
|
| 7739 |
+
"learning_rate": 1.1919073468472475e-05,
|
| 7740 |
+
"loss": 1.24,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.845459781148013,
|
| 7745 |
+
"grad_norm": 1.3878445625305176,
|
| 7746 |
+
"learning_rate": 1.1804294603110222e-05,
|
| 7747 |
+
"loss": 1.1959,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.8462276828565943,
|
| 7752 |
+
"grad_norm": 1.6517834663391113,
|
| 7753 |
+
"learning_rate": 1.1690036392600112e-05,
|
| 7754 |
+
"loss": 1.4946,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.8469955845651757,
|
| 7759 |
+
"grad_norm": 1.5959718227386475,
|
| 7760 |
+
"learning_rate": 1.1576299511454513e-05,
|
| 7761 |
+
"loss": 1.3349,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.847763486273757,
|
| 7766 |
+
"grad_norm": 1.3100310564041138,
|
| 7767 |
+
"learning_rate": 1.1463084631108101e-05,
|
| 7768 |
+
"loss": 1.0446,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.8485313879823383,
|
| 7773 |
+
"grad_norm": 1.5334558486938477,
|
| 7774 |
+
"learning_rate": 1.135039241991408e-05,
|
| 7775 |
+
"loss": 1.0935,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.8492992896909196,
|
| 7780 |
+
"grad_norm": 1.306308627128601,
|
| 7781 |
+
"learning_rate": 1.1238223543140024e-05,
|
| 7782 |
+
"loss": 1.0424,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.8500671913995008,
|
| 7787 |
+
"grad_norm": 1.5370351076126099,
|
| 7788 |
+
"learning_rate": 1.1126578662964115e-05,
|
| 7789 |
+
"loss": 1.1695,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.8508350931080821,
|
| 7794 |
+
"grad_norm": 1.5034894943237305,
|
| 7795 |
+
"learning_rate": 1.1015458438471116e-05,
|
| 7796 |
+
"loss": 1.3281,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.8516029948166635,
|
| 7801 |
+
"grad_norm": 1.815669059753418,
|
| 7802 |
+
"learning_rate": 1.0904863525648633e-05,
|
| 7803 |
+
"loss": 1.3144,
|
| 7804 |
+
"step": 1109
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.8523708965252448,
|
| 7808 |
+
"grad_norm": 1.2194162607192993,
|
| 7809 |
+
"learning_rate": 1.0794794577383016e-05,
|
| 7810 |
+
"loss": 1.1239,
|
| 7811 |
+
"step": 1110
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.8531387982338261,
|
| 7815 |
+
"grad_norm": 1.3530551195144653,
|
| 7816 |
+
"learning_rate": 1.0685252243455712e-05,
|
| 7817 |
+
"loss": 1.1957,
|
| 7818 |
+
"step": 1111
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.8539066999424074,
|
| 7822 |
+
"grad_norm": 1.437690019607544,
|
| 7823 |
+
"learning_rate": 1.0576237170539383e-05,
|
| 7824 |
+
"loss": 1.1088,
|
| 7825 |
+
"step": 1112
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.8546746016509886,
|
| 7829 |
+
"grad_norm": 1.7501755952835083,
|
| 7830 |
+
"learning_rate": 1.0467750002193944e-05,
|
| 7831 |
+
"loss": 1.3246,
|
| 7832 |
+
"step": 1113
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.85544250335957,
|
| 7836 |
+
"grad_norm": 1.587377905845642,
|
| 7837 |
+
"learning_rate": 1.0359791378863005e-05,
|
| 7838 |
+
"loss": 1.2558,
|
| 7839 |
+
"step": 1114
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.8562104050681513,
|
| 7843 |
+
"grad_norm": 1.8887481689453125,
|
| 7844 |
+
"learning_rate": 1.025236193786987e-05,
|
| 7845 |
+
"loss": 1.523,
|
| 7846 |
+
"step": 1115
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.8569783067767326,
|
| 7850 |
+
"grad_norm": 1.4728237390518188,
|
| 7851 |
+
"learning_rate": 1.014546231341391e-05,
|
| 7852 |
+
"loss": 1.4592,
|
| 7853 |
+
"step": 1116
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.8577462084853139,
|
| 7857 |
+
"grad_norm": 1.5659328699111938,
|
| 7858 |
+
"learning_rate": 1.0039093136566735e-05,
|
| 7859 |
+
"loss": 1.2452,
|
| 7860 |
+
"step": 1117
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.8585141101938952,
|
| 7864 |
+
"grad_norm": 1.501523733139038,
|
| 7865 |
+
"learning_rate": 9.933255035268574e-06,
|
| 7866 |
+
"loss": 1.1862,
|
| 7867 |
+
"step": 1118
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.8592820119024764,
|
| 7871 |
+
"grad_norm": 1.4726933240890503,
|
| 7872 |
+
"learning_rate": 9.827948634324447e-06,
|
| 7873 |
+
"loss": 1.1691,
|
| 7874 |
+
"step": 1119
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.8600499136110578,
|
| 7878 |
+
"grad_norm": 1.601131558418274,
|
| 7879 |
+
"learning_rate": 9.72317455540055e-06,
|
| 7880 |
+
"loss": 1.2678,
|
| 7881 |
+
"step": 1120
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.8608178153196391,
|
| 7885 |
+
"grad_norm": 1.5712275505065918,
|
| 7886 |
+
"learning_rate": 9.61893341702056e-06,
|
| 7887 |
+
"loss": 1.1582,
|
| 7888 |
+
"step": 1121
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.8615857170282204,
|
| 7892 |
+
"grad_norm": 1.4145914316177368,
|
| 7893 |
+
"learning_rate": 9.515225834562003e-06,
|
| 7894 |
+
"loss": 0.9806,
|
| 7895 |
+
"step": 1122
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.8623536187368017,
|
| 7899 |
+
"grad_norm": 1.4771418571472168,
|
| 7900 |
+
"learning_rate": 9.412052420252605e-06,
|
| 7901 |
+
"loss": 1.233,
|
| 7902 |
+
"step": 1123
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.863121520445383,
|
| 7906 |
+
"grad_norm": 1.4225813150405884,
|
| 7907 |
+
"learning_rate": 9.309413783166654e-06,
|
| 7908 |
+
"loss": 0.8999,
|
| 7909 |
+
"step": 1124
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.8638894221539642,
|
| 7913 |
+
"grad_norm": 1.6784931421279907,
|
| 7914 |
+
"learning_rate": 9.207310529221525e-06,
|
| 7915 |
+
"loss": 1.2208,
|
| 7916 |
+
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.8646573238625456,
|
| 7920 |
+
"grad_norm": 1.05604887008667,
|
| 7921 |
+
"learning_rate": 9.10574326117386e-06,
|
| 7922 |
+
"loss": 0.9784,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.8654252255711269,
|
| 7927 |
+
"grad_norm": 1.5101516246795654,
|
| 7928 |
+
"learning_rate": 9.004712578616304e-06,
|
| 7929 |
+
"loss": 1.2459,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.8661931272797082,
|
| 7934 |
+
"grad_norm": 1.5375887155532837,
|
| 7935 |
+
"learning_rate": 8.90421907797374e-06,
|
| 7936 |
+
"loss": 1.3941,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.8669610289882895,
|
| 7941 |
+
"grad_norm": 1.382124423980713,
|
| 7942 |
+
"learning_rate": 8.804263352499864e-06,
|
| 7943 |
+
"loss": 1.0871,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.8677289306968708,
|
| 7948 |
+
"grad_norm": 1.4245071411132812,
|
| 7949 |
+
"learning_rate": 8.70484599227367e-06,
|
| 7950 |
+
"loss": 1.0823,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.8684968324054521,
|
| 7955 |
+
"grad_norm": 1.651361107826233,
|
| 7956 |
+
"learning_rate": 8.605967584195995e-06,
|
| 7957 |
+
"loss": 1.3007,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.8692647341140334,
|
| 7962 |
+
"grad_norm": 1.8151158094406128,
|
| 7963 |
+
"learning_rate": 8.507628711985983e-06,
|
| 7964 |
+
"loss": 1.3263,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.8700326358226147,
|
| 7969 |
+
"grad_norm": 1.6930426359176636,
|
| 7970 |
+
"learning_rate": 8.409829956177684e-06,
|
| 7971 |
+
"loss": 1.2812,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.870800537531196,
|
| 7976 |
+
"grad_norm": 1.6178615093231201,
|
| 7977 |
+
"learning_rate": 8.312571894116649e-06,
|
| 7978 |
+
"loss": 1.2922,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.8715684392397773,
|
| 7983 |
+
"grad_norm": 1.7102075815200806,
|
| 7984 |
+
"learning_rate": 8.215855099956472e-06,
|
| 7985 |
+
"loss": 1.1281,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.8723363409483587,
|
| 7990 |
+
"grad_norm": 1.4845244884490967,
|
| 7991 |
+
"learning_rate": 8.119680144655428e-06,
|
| 7992 |
+
"loss": 1.0213,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.8731042426569399,
|
| 7997 |
+
"grad_norm": 1.4863834381103516,
|
| 7998 |
+
"learning_rate": 8.024047595973095e-06,
|
| 7999 |
+
"loss": 1.0862,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.8738721443655212,
|
| 8004 |
+
"grad_norm": 1.2989436388015747,
|
| 8005 |
+
"learning_rate": 7.92895801846707e-06,
|
| 8006 |
+
"loss": 1.052,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.8746400460741025,
|
| 8011 |
+
"grad_norm": 1.3424791097640991,
|
| 8012 |
+
"learning_rate": 7.834411973489419e-06,
|
| 8013 |
+
"loss": 1.3913,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.8754079477826838,
|
| 8018 |
+
"grad_norm": 1.5543662309646606,
|
| 8019 |
+
"learning_rate": 7.740410019183697e-06,
|
| 8020 |
+
"loss": 1.4759,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.8761758494912651,
|
| 8025 |
+
"grad_norm": 1.499265193939209,
|
| 8026 |
+
"learning_rate": 7.646952710481336e-06,
|
| 8027 |
+
"loss": 1.3668,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.8769437511998465,
|
| 8032 |
+
"grad_norm": 1.6394531726837158,
|
| 8033 |
+
"learning_rate": 7.554040599098588e-06,
|
| 8034 |
+
"loss": 1.1713,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.8777116529084277,
|
| 8039 |
+
"grad_norm": 1.4457104206085205,
|
| 8040 |
+
"learning_rate": 7.461674233533123e-06,
|
| 8041 |
+
"loss": 1.3674,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.878479554617009,
|
| 8046 |
+
"grad_norm": 1.468809723854065,
|
| 8047 |
+
"learning_rate": 7.369854159060929e-06,
|
| 8048 |
+
"loss": 1.0444,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.8792474563255903,
|
| 8053 |
+
"grad_norm": 1.610374093055725,
|
| 8054 |
+
"learning_rate": 7.278580917732913e-06,
|
| 8055 |
+
"loss": 1.5037,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.8800153580341716,
|
| 8060 |
+
"grad_norm": 1.3369286060333252,
|
| 8061 |
+
"learning_rate": 7.187855048371917e-06,
|
| 8062 |
+
"loss": 1.129,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.880783259742753,
|
| 8067 |
+
"grad_norm": 1.4274189472198486,
|
| 8068 |
+
"learning_rate": 7.097677086569343e-06,
|
| 8069 |
+
"loss": 1.2629,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.8815511614513343,
|
| 8074 |
+
"grad_norm": 1.4373105764389038,
|
| 8075 |
+
"learning_rate": 7.008047564682119e-06,
|
| 8076 |
+
"loss": 1.2244,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.8823190631599155,
|
| 8081 |
+
"grad_norm": 1.5060944557189941,
|
| 8082 |
+
"learning_rate": 6.91896701182948e-06,
|
| 8083 |
+
"loss": 1.3279,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.8830869648684968,
|
| 8088 |
+
"grad_norm": 1.1090878248214722,
|
| 8089 |
+
"learning_rate": 6.83043595388988e-06,
|
| 8090 |
+
"loss": 0.9791,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.8838548665770781,
|
| 8095 |
+
"grad_norm": 1.6525593996047974,
|
| 8096 |
+
"learning_rate": 6.742454913497942e-06,
|
| 8097 |
+
"loss": 1.2751,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.8846227682856594,
|
| 8102 |
+
"grad_norm": 1.5659713745117188,
|
| 8103 |
+
"learning_rate": 6.6550244100412e-06,
|
| 8104 |
+
"loss": 1.3544,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.8853906699942408,
|
| 8109 |
+
"grad_norm": 1.4881879091262817,
|
| 8110 |
+
"learning_rate": 6.568144959657263e-06,
|
| 8111 |
+
"loss": 1.1873,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.8861585717028221,
|
| 8116 |
+
"grad_norm": 1.8786675930023193,
|
| 8117 |
+
"learning_rate": 6.481817075230567e-06,
|
| 8118 |
+
"loss": 1.3176,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.8869264734114033,
|
| 8123 |
+
"grad_norm": 1.4211797714233398,
|
| 8124 |
+
"learning_rate": 6.396041266389474e-06,
|
| 8125 |
+
"loss": 1.3475,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.8876943751199846,
|
| 8130 |
+
"grad_norm": 1.4349972009658813,
|
| 8131 |
+
"learning_rate": 6.3108180395031965e-06,
|
| 8132 |
+
"loss": 1.1019,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.8884622768285659,
|
| 8137 |
+
"grad_norm": 1.3336567878723145,
|
| 8138 |
+
"learning_rate": 6.22614789767888e-06,
|
| 8139 |
+
"loss": 1.1098,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.8892301785371473,
|
| 8144 |
+
"grad_norm": 1.4133919477462769,
|
| 8145 |
+
"learning_rate": 6.142031340758525e-06,
|
| 8146 |
+
"loss": 1.2118,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.8899980802457286,
|
| 8151 |
+
"grad_norm": 1.3893505334854126,
|
| 8152 |
+
"learning_rate": 6.058468865316102e-06,
|
| 8153 |
+
"loss": 1.1315,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.8907659819543099,
|
| 8158 |
+
"grad_norm": 1.786585807800293,
|
| 8159 |
+
"learning_rate": 5.975460964654689e-06,
|
| 8160 |
+
"loss": 1.6215,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.8915338836628911,
|
| 8165 |
+
"grad_norm": 1.3746907711029053,
|
| 8166 |
+
"learning_rate": 5.8930081288034014e-06,
|
| 8167 |
+
"loss": 1.0703,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.8923017853714724,
|
| 8172 |
+
"grad_norm": 1.420802116394043,
|
| 8173 |
+
"learning_rate": 5.8111108445146116e-06,
|
| 8174 |
+
"loss": 1.0426,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.8930696870800537,
|
| 8179 |
+
"grad_norm": 1.5010699033737183,
|
| 8180 |
+
"learning_rate": 5.72976959526107e-06,
|
| 8181 |
+
"loss": 1.3698,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.8938375887886351,
|
| 8186 |
+
"grad_norm": 1.3752045631408691,
|
| 8187 |
+
"learning_rate": 5.648984861232986e-06,
|
| 8188 |
+
"loss": 1.2014,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.8946054904972164,
|
| 8193 |
+
"grad_norm": 1.5743210315704346,
|
| 8194 |
+
"learning_rate": 5.568757119335244e-06,
|
| 8195 |
+
"loss": 1.4007,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.8953733922057977,
|
| 8200 |
+
"grad_norm": 1.4705897569656372,
|
| 8201 |
+
"learning_rate": 5.489086843184632e-06,
|
| 8202 |
+
"loss": 1.0485,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.8961412939143789,
|
| 8207 |
+
"grad_norm": 1.4664305448532104,
|
| 8208 |
+
"learning_rate": 5.40997450310693e-06,
|
| 8209 |
+
"loss": 1.3765,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.8969091956229602,
|
| 8214 |
+
"grad_norm": 1.6887331008911133,
|
| 8215 |
+
"learning_rate": 5.3314205661342155e-06,
|
| 8216 |
+
"loss": 1.3474,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.8976770973315416,
|
| 8221 |
+
"grad_norm": 1.5927088260650635,
|
| 8222 |
+
"learning_rate": 5.253425496002084e-06,
|
| 8223 |
+
"loss": 1.2915,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.8984449990401229,
|
| 8228 |
+
"grad_norm": 1.3032876253128052,
|
| 8229 |
+
"learning_rate": 5.175989753146948e-06,
|
| 8230 |
+
"loss": 1.2463,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.8992129007487042,
|
| 8235 |
+
"grad_norm": 1.2626043558120728,
|
| 8236 |
+
"learning_rate": 5.099113794703225e-06,
|
| 8237 |
+
"loss": 1.2223,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.8999808024572855,
|
| 8242 |
+
"grad_norm": 1.5485360622406006,
|
| 8243 |
+
"learning_rate": 5.0227980745007345e-06,
|
| 8244 |
+
"loss": 1.4437,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.9007487041658667,
|
| 8249 |
+
"grad_norm": 1.3756585121154785,
|
| 8250 |
+
"learning_rate": 4.947043043062016e-06,
|
| 8251 |
+
"loss": 1.3702,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.901516605874448,
|
| 8256 |
+
"grad_norm": 1.281550645828247,
|
| 8257 |
+
"learning_rate": 4.87184914759955e-06,
|
| 8258 |
+
"loss": 0.8686,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.9022845075830294,
|
| 8263 |
+
"grad_norm": 1.7812248468399048,
|
| 8264 |
+
"learning_rate": 4.7972168320132845e-06,
|
| 8265 |
+
"loss": 1.2599,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.9030524092916107,
|
| 8270 |
+
"grad_norm": 1.186799168586731,
|
| 8271 |
+
"learning_rate": 4.7231465368879124e-06,
|
| 8272 |
+
"loss": 1.0582,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.903820311000192,
|
| 8277 |
+
"grad_norm": 1.4105154275894165,
|
| 8278 |
+
"learning_rate": 4.649638699490266e-06,
|
| 8279 |
+
"loss": 1.1345,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.9045882127087733,
|
| 8284 |
+
"grad_norm": 1.4764769077301025,
|
| 8285 |
+
"learning_rate": 4.576693753766792e-06,
|
| 8286 |
+
"loss": 1.3989,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.9053561144173545,
|
| 8291 |
+
"grad_norm": 2.1660075187683105,
|
| 8292 |
+
"learning_rate": 4.5043121303409595e-06,
|
| 8293 |
+
"loss": 1.1665,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.9061240161259358,
|
| 8298 |
+
"grad_norm": 1.262603521347046,
|
| 8299 |
+
"learning_rate": 4.432494256510711e-06,
|
| 8300 |
+
"loss": 0.9425,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.9068919178345172,
|
| 8305 |
+
"grad_norm": 1.8108893632888794,
|
| 8306 |
+
"learning_rate": 4.361240556245938e-06,
|
| 8307 |
+
"loss": 1.2851,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.9076598195430985,
|
| 8312 |
+
"grad_norm": 1.5858032703399658,
|
| 8313 |
+
"learning_rate": 4.290551450185986e-06,
|
| 8314 |
+
"loss": 1.4577,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.9084277212516798,
|
| 8319 |
+
"grad_norm": 1.2271817922592163,
|
| 8320 |
+
"learning_rate": 4.220427355637224e-06,
|
| 8321 |
+
"loss": 1.1701,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.9091956229602611,
|
| 8326 |
+
"grad_norm": 1.280598759651184,
|
| 8327 |
+
"learning_rate": 4.150868686570464e-06,
|
| 8328 |
+
"loss": 1.2309,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.9099635246688423,
|
| 8333 |
+
"grad_norm": 1.40516996383667,
|
| 8334 |
+
"learning_rate": 4.081875853618588e-06,
|
| 8335 |
+
"loss": 1.0201,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.9107314263774237,
|
| 8340 |
+
"grad_norm": 1.4448282718658447,
|
| 8341 |
+
"learning_rate": 4.013449264074187e-06,
|
| 8342 |
+
"loss": 1.1834,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.911499328086005,
|
| 8347 |
+
"grad_norm": 1.2964657545089722,
|
| 8348 |
+
"learning_rate": 3.945589321886989e-06,
|
| 8349 |
+
"loss": 1.0199,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.9122672297945863,
|
| 8354 |
+
"grad_norm": 1.3470900058746338,
|
| 8355 |
+
"learning_rate": 3.878296427661676e-06,
|
| 8356 |
+
"loss": 1.0585,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.9130351315031676,
|
| 8361 |
+
"grad_norm": 1.3308759927749634,
|
| 8362 |
+
"learning_rate": 3.8115709786553435e-06,
|
| 8363 |
+
"loss": 1.3752,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.9138030332117489,
|
| 8368 |
+
"grad_norm": 1.4458931684494019,
|
| 8369 |
+
"learning_rate": 3.7454133687752524e-06,
|
| 8370 |
+
"loss": 1.1349,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.9145709349203301,
|
| 8375 |
+
"grad_norm": 1.471204400062561,
|
| 8376 |
+
"learning_rate": 3.6798239885764806e-06,
|
| 8377 |
+
"loss": 1.126,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.9153388366289115,
|
| 8382 |
+
"grad_norm": 1.403232216835022,
|
| 8383 |
+
"learning_rate": 3.614803225259622e-06,
|
| 8384 |
+
"loss": 1.1377,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.9161067383374928,
|
| 8389 |
+
"grad_norm": 1.5859801769256592,
|
| 8390 |
+
"learning_rate": 3.550351462668489e-06,
|
| 8391 |
+
"loss": 1.2045,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.9168746400460741,
|
| 8396 |
+
"grad_norm": 1.2398918867111206,
|
| 8397 |
+
"learning_rate": 3.4864690812878688e-06,
|
| 8398 |
+
"loss": 1.111,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.9176425417546554,
|
| 8403 |
+
"grad_norm": 1.455277681350708,
|
| 8404 |
+
"learning_rate": 3.4231564582412167e-06,
|
| 8405 |
+
"loss": 1.1314,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.9184104434632367,
|
| 8410 |
+
"grad_norm": 1.5992294549942017,
|
| 8411 |
+
"learning_rate": 3.3604139672885227e-06,
|
| 8412 |
+
"loss": 1.2585,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.919178345171818,
|
| 8417 |
+
"grad_norm": 1.6772078275680542,
|
| 8418 |
+
"learning_rate": 3.298241978824046e-06,
|
| 8419 |
+
"loss": 1.0393,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.9199462468803993,
|
| 8424 |
+
"grad_norm": 1.202988862991333,
|
| 8425 |
+
"learning_rate": 3.2366408598741072e-06,
|
| 8426 |
+
"loss": 1.0763,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.9207141485889806,
|
| 8431 |
+
"grad_norm": 1.5782155990600586,
|
| 8432 |
+
"learning_rate": 3.175610974095011e-06,
|
| 8433 |
+
"loss": 1.3341,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.9214820502975619,
|
| 8438 |
+
"grad_norm": 1.611743450164795,
|
| 8439 |
+
"learning_rate": 3.115152681770783e-06,
|
| 8440 |
+
"loss": 1.0767,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.9222499520061432,
|
| 8445 |
+
"grad_norm": 1.6371206045150757,
|
| 8446 |
+
"learning_rate": 3.055266339811147e-06,
|
| 8447 |
+
"loss": 0.7947,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.9230178537147246,
|
| 8452 |
+
"grad_norm": 1.335860252380371,
|
| 8453 |
+
"learning_rate": 2.9959523017493386e-06,
|
| 8454 |
+
"loss": 1.2236,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.9237857554233058,
|
| 8459 |
+
"grad_norm": 1.4622019529342651,
|
| 8460 |
+
"learning_rate": 2.9372109177400854e-06,
|
| 8461 |
+
"loss": 1.5577,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.9245536571318871,
|
| 8466 |
+
"grad_norm": 1.4906941652297974,
|
| 8467 |
+
"learning_rate": 2.8790425345574745e-06,
|
| 8468 |
+
"loss": 1.1759,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.9253215588404684,
|
| 8473 |
+
"grad_norm": 1.2302792072296143,
|
| 8474 |
+
"learning_rate": 2.821447495592977e-06,
|
| 8475 |
+
"loss": 0.7965,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.9260894605490497,
|
| 8480 |
+
"grad_norm": 1.4680209159851074,
|
| 8481 |
+
"learning_rate": 2.7644261408533155e-06,
|
| 8482 |
+
"loss": 1.213,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.926857362257631,
|
| 8487 |
+
"grad_norm": 1.737641453742981,
|
| 8488 |
+
"learning_rate": 2.707978806958611e-06,
|
| 8489 |
+
"loss": 1.1734,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.9276252639662124,
|
| 8494 |
+
"grad_norm": 1.386087417602539,
|
| 8495 |
+
"learning_rate": 2.6521058271402386e-06,
|
| 8496 |
+
"loss": 1.3075,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.9283931656747936,
|
| 8501 |
+
"grad_norm": 1.5818560123443604,
|
| 8502 |
+
"learning_rate": 2.596807531238965e-06,
|
| 8503 |
+
"loss": 1.3811,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.9291610673833749,
|
| 8508 |
+
"grad_norm": 1.2463423013687134,
|
| 8509 |
+
"learning_rate": 2.542084245702947e-06,
|
| 8510 |
+
"loss": 1.2488,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.9299289690919562,
|
| 8515 |
+
"grad_norm": 1.1760039329528809,
|
| 8516 |
+
"learning_rate": 2.487936293585813e-06,
|
| 8517 |
+
"loss": 0.9825,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.9306968708005375,
|
| 8522 |
+
"grad_norm": 1.2284053564071655,
|
| 8523 |
+
"learning_rate": 2.4343639945448306e-06,
|
| 8524 |
+
"loss": 1.12,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.9314647725091189,
|
| 8529 |
+
"grad_norm": 1.4763203859329224,
|
| 8530 |
+
"learning_rate": 2.3813676648388517e-06,
|
| 8531 |
+
"loss": 1.1244,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.9322326742177002,
|
| 8536 |
+
"grad_norm": 1.6483160257339478,
|
| 8537 |
+
"learning_rate": 2.3289476173266376e-06,
|
| 8538 |
+
"loss": 1.3215,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.9330005759262814,
|
| 8543 |
+
"grad_norm": 1.625486135482788,
|
| 8544 |
+
"learning_rate": 2.2771041614648825e-06,
|
| 8545 |
+
"loss": 1.578,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.9337684776348627,
|
| 8550 |
+
"grad_norm": 1.7281734943389893,
|
| 8551 |
+
"learning_rate": 2.2258376033064354e-06,
|
| 8552 |
+
"loss": 1.3752,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.934536379343444,
|
| 8557 |
+
"grad_norm": 1.1822775602340698,
|
| 8558 |
+
"learning_rate": 2.1751482454984706e-06,
|
| 8559 |
+
"loss": 0.9411,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.9353042810520253,
|
| 8564 |
+
"grad_norm": 1.5272676944732666,
|
| 8565 |
+
"learning_rate": 2.1250363872807655e-06,
|
| 8566 |
+
"loss": 1.3861,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.9360721827606067,
|
| 8571 |
+
"grad_norm": 1.6605676412582397,
|
| 8572 |
+
"learning_rate": 2.0755023244838136e-06,
|
| 8573 |
+
"loss": 1.5562,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.936840084469188,
|
| 8578 |
+
"grad_norm": 1.4125683307647705,
|
| 8579 |
+
"learning_rate": 2.026546349527181e-06,
|
| 8580 |
+
"loss": 1.0978,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.9376079861777692,
|
| 8585 |
+
"grad_norm": 1.3512035608291626,
|
| 8586 |
+
"learning_rate": 1.978168751417786e-06,
|
| 8587 |
+
"loss": 0.9322,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.9383758878863505,
|
| 8592 |
+
"grad_norm": 1.7064318656921387,
|
| 8593 |
+
"learning_rate": 1.93036981574809e-06,
|
| 8594 |
+
"loss": 1.2816,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.9391437895949318,
|
| 8599 |
+
"grad_norm": 1.2211792469024658,
|
| 8600 |
+
"learning_rate": 1.8831498246945189e-06,
|
| 8601 |
+
"loss": 1.0071,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.9399116913035132,
|
| 8606 |
+
"grad_norm": 1.3540048599243164,
|
| 8607 |
+
"learning_rate": 1.836509057015734e-06,
|
| 8608 |
+
"loss": 1.202,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.9406795930120945,
|
| 8613 |
+
"grad_norm": 1.3373445272445679,
|
| 8614 |
+
"learning_rate": 1.7904477880510307e-06,
|
| 8615 |
+
"loss": 1.142,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.9414474947206758,
|
| 8620 |
+
"grad_norm": 1.38016939163208,
|
| 8621 |
+
"learning_rate": 1.7449662897186414e-06,
|
| 8622 |
+
"loss": 1.3193,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.942215396429257,
|
| 8627 |
+
"grad_norm": 1.3662525415420532,
|
| 8628 |
+
"learning_rate": 1.7000648305142364e-06,
|
| 8629 |
+
"loss": 1.0969,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.9429832981378383,
|
| 8634 |
+
"grad_norm": 1.7097259759902954,
|
| 8635 |
+
"learning_rate": 1.655743675509258e-06,
|
| 8636 |
+
"loss": 1.4585,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.9437511998464196,
|
| 8641 |
+
"grad_norm": 1.177868127822876,
|
| 8642 |
+
"learning_rate": 1.6120030863493674e-06,
|
| 8643 |
+
"loss": 0.9733,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.944519101555001,
|
| 8648 |
+
"grad_norm": 1.481663465499878,
|
| 8649 |
+
"learning_rate": 1.5688433212529107e-06,
|
| 8650 |
+
"loss": 1.0466,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.9452870032635823,
|
| 8655 |
+
"grad_norm": 1.6859853267669678,
|
| 8656 |
+
"learning_rate": 1.5262646350094334e-06,
|
| 8657 |
+
"loss": 1.2808,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.9460549049721636,
|
| 8662 |
+
"grad_norm": 1.7730293273925781,
|
| 8663 |
+
"learning_rate": 1.4842672789780798e-06,
|
| 8664 |
+
"loss": 1.2563,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.9468228066807448,
|
| 8669 |
+
"grad_norm": 1.5149153470993042,
|
| 8670 |
+
"learning_rate": 1.4428515010861955e-06,
|
| 8671 |
+
"loss": 1.249,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.9475907083893261,
|
| 8676 |
+
"grad_norm": 1.8611276149749756,
|
| 8677 |
+
"learning_rate": 1.4020175458278607e-06,
|
| 8678 |
+
"loss": 1.5504,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.9483586100979075,
|
| 8683 |
+
"grad_norm": 1.5275269746780396,
|
| 8684 |
+
"learning_rate": 1.3617656542623813e-06,
|
| 8685 |
+
"loss": 1.3417,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.9491265118064888,
|
| 8690 |
+
"grad_norm": 1.3223843574523926,
|
| 8691 |
+
"learning_rate": 1.322096064012912e-06,
|
| 8692 |
+
"loss": 1.1907,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.9498944135150701,
|
| 8697 |
+
"grad_norm": 1.4529744386672974,
|
| 8698 |
+
"learning_rate": 1.2830090092650904e-06,
|
| 8699 |
+
"loss": 1.16,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.9506623152236514,
|
| 8704 |
+
"grad_norm": 1.4799784421920776,
|
| 8705 |
+
"learning_rate": 1.244504720765549e-06,
|
| 8706 |
+
"loss": 1.3583,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.9514302169322326,
|
| 8711 |
+
"grad_norm": 1.4964150190353394,
|
| 8712 |
+
"learning_rate": 1.20658342582064e-06,
|
| 8713 |
+
"loss": 1.2563,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.9521981186408139,
|
| 8718 |
+
"grad_norm": 1.318679690361023,
|
| 8719 |
+
"learning_rate": 1.1692453482951115e-06,
|
| 8720 |
+
"loss": 1.1699,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.9529660203493953,
|
| 8725 |
+
"grad_norm": 1.2729301452636719,
|
| 8726 |
+
"learning_rate": 1.1324907086106895e-06,
|
| 8727 |
+
"loss": 1.0768,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.9537339220579766,
|
| 8732 |
+
"grad_norm": 1.4914193153381348,
|
| 8733 |
+
"learning_rate": 1.096319723744843e-06,
|
| 8734 |
+
"loss": 0.9835,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.9545018237665579,
|
| 8739 |
+
"grad_norm": 1.4499318599700928,
|
| 8740 |
+
"learning_rate": 1.0607326072295087e-06,
|
| 8741 |
+
"loss": 1.1102,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.9552697254751392,
|
| 8746 |
+
"grad_norm": 1.4721513986587524,
|
| 8747 |
+
"learning_rate": 1.0257295691497914e-06,
|
| 8748 |
+
"loss": 1.234,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.9560376271837204,
|
| 8753 |
+
"grad_norm": 1.553381323814392,
|
| 8754 |
+
"learning_rate": 9.913108161427543e-07,
|
| 8755 |
+
"loss": 1.2512,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.9568055288923017,
|
| 8760 |
+
"grad_norm": 1.384883165359497,
|
| 8761 |
+
"learning_rate": 9.57476551396197e-07,
|
| 8762 |
+
"loss": 1.2325,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.9575734306008831,
|
| 8767 |
+
"grad_norm": 1.6136502027511597,
|
| 8768 |
+
"learning_rate": 9.242269746474575e-07,
|
| 8769 |
+
"loss": 1.4892,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.9583413323094644,
|
| 8774 |
+
"grad_norm": 1.5573700666427612,
|
| 8775 |
+
"learning_rate": 8.915622821821789e-07,
|
| 8776 |
+
"loss": 1.447,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.9591092340180457,
|
| 8781 |
+
"grad_norm": 1.47049880027771,
|
| 8782 |
+
"learning_rate": 8.594826668332445e-07,
|
| 8783 |
+
"loss": 1.1597,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.959877135726627,
|
| 8788 |
+
"grad_norm": 0.9694231748580933,
|
| 8789 |
+
"learning_rate": 8.279883179795666e-07,
|
| 8790 |
+
"loss": 0.8528,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.9606450374352083,
|
| 8795 |
+
"grad_norm": 1.28829026222229,
|
| 8796 |
+
"learning_rate": 7.970794215450106e-07,
|
| 8797 |
+
"loss": 1.1936,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.9614129391437896,
|
| 8802 |
+
"grad_norm": 1.6050745248794556,
|
| 8803 |
+
"learning_rate": 7.667561599972505e-07,
|
| 8804 |
+
"loss": 1.3697,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.9621808408523709,
|
| 8809 |
+
"grad_norm": 1.1561931371688843,
|
| 8810 |
+
"learning_rate": 7.370187123467708e-07,
|
| 8811 |
+
"loss": 0.9132,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.9629487425609522,
|
| 8816 |
+
"grad_norm": 1.2682461738586426,
|
| 8817 |
+
"learning_rate": 7.078672541456999e-07,
|
| 8818 |
+
"loss": 1.1176,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.9637166442695335,
|
| 8823 |
+
"grad_norm": 1.4032613039016724,
|
| 8824 |
+
"learning_rate": 6.793019574868775e-07,
|
| 8825 |
+
"loss": 1.2812,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.9644845459781148,
|
| 8830 |
+
"grad_norm": 1.855252981185913,
|
| 8831 |
+
"learning_rate": 6.513229910027896e-07,
|
| 8832 |
+
"loss": 1.4774,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.9652524476866962,
|
| 8837 |
+
"grad_norm": 1.4534977674484253,
|
| 8838 |
+
"learning_rate": 6.239305198645462e-07,
|
| 8839 |
+
"loss": 1.097,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.9660203493952774,
|
| 8844 |
+
"grad_norm": 1.4636932611465454,
|
| 8845 |
+
"learning_rate": 5.971247057809826e-07,
|
| 8846 |
+
"loss": 1.2459,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.9667882511038587,
|
| 8851 |
+
"grad_norm": 1.318543553352356,
|
| 8852 |
+
"learning_rate": 5.709057069976265e-07,
|
| 8853 |
+
"loss": 1.342,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.96755615281244,
|
| 8858 |
+
"grad_norm": 1.2237138748168945,
|
| 8859 |
+
"learning_rate": 5.452736782958323e-07,
|
| 8860 |
+
"loss": 1.0311,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.9683240545210213,
|
| 8865 |
+
"grad_norm": 1.3968307971954346,
|
| 8866 |
+
"learning_rate": 5.20228770991793e-07,
|
| 8867 |
+
"loss": 1.2152,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.9690919562296026,
|
| 8872 |
+
"grad_norm": 1.2888987064361572,
|
| 8873 |
+
"learning_rate": 4.957711329357073e-07,
|
| 8874 |
+
"loss": 0.9803,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.969859857938184,
|
| 8879 |
+
"grad_norm": 1.2455861568450928,
|
| 8880 |
+
"learning_rate": 4.7190090851090274e-07,
|
| 8881 |
+
"loss": 0.814,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.9706277596467652,
|
| 8886 |
+
"grad_norm": 1.263824462890625,
|
| 8887 |
+
"learning_rate": 4.4861823863292516e-07,
|
| 8888 |
+
"loss": 0.9433,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.9713956613553465,
|
| 8893 |
+
"grad_norm": 1.63260817527771,
|
| 8894 |
+
"learning_rate": 4.259232607487951e-07,
|
| 8895 |
+
"loss": 1.4563,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.9721635630639278,
|
| 8900 |
+
"grad_norm": 1.3201172351837158,
|
| 8901 |
+
"learning_rate": 4.038161088361192e-07,
|
| 8902 |
+
"loss": 1.0101,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.9729314647725091,
|
| 8907 |
+
"grad_norm": 1.456137776374817,
|
| 8908 |
+
"learning_rate": 3.8229691340234684e-07,
|
| 8909 |
+
"loss": 1.3377,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.9736993664810905,
|
| 8914 |
+
"grad_norm": 1.9307949542999268,
|
| 8915 |
+
"learning_rate": 3.613658014839594e-07,
|
| 8916 |
+
"loss": 1.5946,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.9744672681896718,
|
| 8921 |
+
"grad_norm": 1.0594274997711182,
|
| 8922 |
+
"learning_rate": 3.4102289664578177e-07,
|
| 8923 |
+
"loss": 1.0772,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.975235169898253,
|
| 8928 |
+
"grad_norm": 1.3033182621002197,
|
| 8929 |
+
"learning_rate": 3.212683189801724e-07,
|
| 8930 |
+
"loss": 1.2004,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.9760030716068343,
|
| 8935 |
+
"grad_norm": 1.574930191040039,
|
| 8936 |
+
"learning_rate": 3.021021851063899e-07,
|
| 8937 |
+
"loss": 1.266,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.9767709733154156,
|
| 8942 |
+
"grad_norm": 1.5406841039657593,
|
| 8943 |
+
"learning_rate": 2.8352460816986057e-07,
|
| 8944 |
+
"loss": 1.0827,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.977538875023997,
|
| 8949 |
+
"grad_norm": 1.1407910585403442,
|
| 8950 |
+
"learning_rate": 2.6553569784152357e-07,
|
| 8951 |
+
"loss": 0.8456,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.9783067767325783,
|
| 8956 |
+
"grad_norm": 1.4787224531173706,
|
| 8957 |
+
"learning_rate": 2.481355603171531e-07,
|
| 8958 |
+
"loss": 1.3375,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.9790746784411596,
|
| 8963 |
+
"grad_norm": 1.514699935913086,
|
| 8964 |
+
"learning_rate": 2.3132429831682622e-07,
|
| 8965 |
+
"loss": 1.4017,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.9798425801497408,
|
| 8970 |
+
"grad_norm": 1.372117519378662,
|
| 8971 |
+
"learning_rate": 2.1510201108416728e-07,
|
| 8972 |
+
"loss": 1.1619,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.9806104818583221,
|
| 8977 |
+
"grad_norm": 1.2351239919662476,
|
| 8978 |
+
"learning_rate": 1.9946879438592636e-07,
|
| 8979 |
+
"loss": 0.945,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.9813783835669034,
|
| 8984 |
+
"grad_norm": 1.4396629333496094,
|
| 8985 |
+
"learning_rate": 1.8442474051125757e-07,
|
| 8986 |
+
"loss": 0.9939,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.9821462852754848,
|
| 8991 |
+
"grad_norm": 1.9772976636886597,
|
| 8992 |
+
"learning_rate": 1.6996993827129715e-07,
|
| 8993 |
+
"loss": 1.4349,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.9829141869840661,
|
| 8998 |
+
"grad_norm": 1.2860651016235352,
|
| 8999 |
+
"learning_rate": 1.561044729985861e-07,
|
| 9000 |
+
"loss": 0.9433,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.9836820886926474,
|
| 9005 |
+
"grad_norm": 1.496124505996704,
|
| 9006 |
+
"learning_rate": 1.428284265465596e-07,
|
| 9007 |
+
"loss": 1.2376,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.9844499904012286,
|
| 9012 |
+
"grad_norm": 1.2294336557388306,
|
| 9013 |
+
"learning_rate": 1.3014187728906945e-07,
|
| 9014 |
+
"loss": 0.9644,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.9852178921098099,
|
| 9019 |
+
"grad_norm": 1.182978630065918,
|
| 9020 |
+
"learning_rate": 1.1804490011995129e-07,
|
| 9021 |
+
"loss": 1.2047,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.9859857938183912,
|
| 9026 |
+
"grad_norm": 1.6320767402648926,
|
| 9027 |
+
"learning_rate": 1.0653756645252477e-07,
|
| 9028 |
+
"loss": 1.4767,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.9867536955269726,
|
| 9033 |
+
"grad_norm": 1.482164978981018,
|
| 9034 |
+
"learning_rate": 9.561994421924958e-08,
|
| 9035 |
+
"loss": 1.21,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.9875215972355539,
|
| 9040 |
+
"grad_norm": 1.3324884176254272,
|
| 9041 |
+
"learning_rate": 8.529209787123682e-08,
|
| 9042 |
+
"loss": 1.0876,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.9882894989441352,
|
| 9047 |
+
"grad_norm": 1.530920386314392,
|
| 9048 |
+
"learning_rate": 7.555408837794931e-08,
|
| 9049 |
+
"loss": 1.1581,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.9890574006527164,
|
| 9054 |
+
"grad_norm": 1.4899532794952393,
|
| 9055 |
+
"learning_rate": 6.640597322677967e-08,
|
| 9056 |
+
"loss": 1.3396,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.9898253023612977,
|
| 9061 |
+
"grad_norm": 1.2358452081680298,
|
| 9062 |
+
"learning_rate": 5.784780642275056e-08,
|
| 9063 |
+
"loss": 1.0957,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.990593204069879,
|
| 9068 |
+
"grad_norm": 1.703139305114746,
|
| 9069 |
+
"learning_rate": 4.9879638488159465e-08,
|
| 9070 |
+
"loss": 1.2339,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.9913611057784604,
|
| 9075 |
+
"grad_norm": 1.3851401805877686,
|
| 9076 |
+
"learning_rate": 4.2501516462334356e-08,
|
| 9077 |
+
"loss": 1.4007,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.9921290074870417,
|
| 9082 |
+
"grad_norm": 1.52871572971344,
|
| 9083 |
+
"learning_rate": 3.5713483901300696e-08,
|
| 9084 |
+
"loss": 1.3904,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.992896909195623,
|
| 9089 |
+
"grad_norm": 1.9627262353897095,
|
| 9090 |
+
"learning_rate": 2.9515580877559346e-08,
|
| 9091 |
+
"loss": 1.2959,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.9936648109042042,
|
| 9096 |
+
"grad_norm": 2.1106338500976562,
|
| 9097 |
+
"learning_rate": 2.3907843979831257e-08,
|
| 9098 |
+
"loss": 1.7045,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.9944327126127855,
|
| 9103 |
+
"grad_norm": 1.3870468139648438,
|
| 9104 |
+
"learning_rate": 1.8890306312846495e-08,
|
| 9105 |
+
"loss": 1.2606,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.9952006143213669,
|
| 9110 |
+
"grad_norm": 1.4199986457824707,
|
| 9111 |
+
"learning_rate": 1.446299749716662e-08,
|
| 9112 |
+
"loss": 1.3145,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.9959685160299482,
|
| 9117 |
+
"grad_norm": 1.398864984512329,
|
| 9118 |
+
"learning_rate": 1.0625943668973736e-08,
|
| 9119 |
+
"loss": 1.3875,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.9967364177385295,
|
| 9124 |
+
"grad_norm": 1.4852126836776733,
|
| 9125 |
+
"learning_rate": 7.379167479948379e-09,
|
| 9126 |
+
"loss": 1.1144,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.9975043194471108,
|
| 9131 |
+
"grad_norm": 1.3623526096343994,
|
| 9132 |
+
"learning_rate": 4.722688097125172e-09,
|
| 9133 |
+
"loss": 1.3668,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.998272221155692,
|
| 9138 |
+
"grad_norm": 1.426559329032898,
|
| 9139 |
+
"learning_rate": 2.656521202770712e-09,
|
| 9140 |
+
"loss": 1.2487,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.9990401228642733,
|
| 9145 |
+
"grad_norm": 1.8108794689178467,
|
| 9146 |
+
"learning_rate": 1.1806789942947484e-09,
|
| 9147 |
+
"loss": 1.5596,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.9998080245728547,
|
| 9152 |
+
"grad_norm": 1.4854652881622314,
|
| 9153 |
+
"learning_rate": 2.9517018420577305e-10,
|
| 9154 |
+
"loss": 1.0427,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 1.000575926281436,
|
| 9159 |
+
"grad_norm": 2.9019205570220947,
|
| 9160 |
+
"learning_rate": 0.0,
|
| 9161 |
+
"loss": 2.1779,
|
| 9162 |
+
"step": 1303
|
| 9163 |
}
|
| 9164 |
],
|
| 9165 |
"logging_steps": 1,
|
|
|
|
| 9174 |
"should_evaluate": false,
|
| 9175 |
"should_log": false,
|
| 9176 |
"should_save": true,
|
| 9177 |
+
"should_training_stop": true
|
| 9178 |
},
|
| 9179 |
"attributes": {}
|
| 9180 |
}
|
| 9181 |
},
|
| 9182 |
+
"total_flos": 2.5981287804174336e+17,
|
| 9183 |
"train_batch_size": 2,
|
| 9184 |
"trial_name": null,
|
| 9185 |
"trial_params": null
|