Training in progress, step 1068, 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 83945296
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1618bfb7ac8137979d2aa97bc8ca912641014a934ddcfbd4df4476876a1be2bb
|
| 3 |
size 83945296
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 43123028
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6f987b2a4bca372072c860dd5a297312ec3d56284c95a3d34030a7aa5d1f9ba8
|
| 3 |
size 43123028
|
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:c31e0209567b5f8eea6c523f42b84d20306d2185e71cccf7f4c0774a1b41f9ac
|
| 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:de4f7f48a1effb528b177c21ecc6b8eb666d1c537cd6834234a4c7e3f0bfa7a7
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 356,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5015,6 +5015,2506 @@
|
|
| 5015 |
"eval_samples_per_second": 19.127,
|
| 5016 |
"eval_steps_per_second": 9.57,
|
| 5017 |
"step": 712
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5018 |
}
|
| 5019 |
],
|
| 5020 |
"logging_steps": 1,
|
|
@@ -5034,7 +7534,7 @@
|
|
| 5034 |
"attributes": {}
|
| 5035 |
}
|
| 5036 |
},
|
| 5037 |
-
"total_flos":
|
| 5038 |
"train_batch_size": 2,
|
| 5039 |
"trial_name": null,
|
| 5040 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.30042194092827,
|
| 5 |
"eval_steps": 356,
|
| 6 |
+
"global_step": 1068,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5015 |
"eval_samples_per_second": 19.127,
|
| 5016 |
"eval_steps_per_second": 9.57,
|
| 5017 |
"step": 712
|
| 5018 |
+
},
|
| 5019 |
+
{
|
| 5020 |
+
"epoch": 0.20056258790436005,
|
| 5021 |
+
"grad_norm": 0.48163357377052307,
|
| 5022 |
+
"learning_rate": 0.00010077816442425216,
|
| 5023 |
+
"loss": 0.7656,
|
| 5024 |
+
"step": 713
|
| 5025 |
+
},
|
| 5026 |
+
{
|
| 5027 |
+
"epoch": 0.2008438818565401,
|
| 5028 |
+
"grad_norm": 0.4076929986476898,
|
| 5029 |
+
"learning_rate": 0.0001005558344792571,
|
| 5030 |
+
"loss": 0.4765,
|
| 5031 |
+
"step": 714
|
| 5032 |
+
},
|
| 5033 |
+
{
|
| 5034 |
+
"epoch": 0.2011251758087201,
|
| 5035 |
+
"grad_norm": 0.43237489461898804,
|
| 5036 |
+
"learning_rate": 0.00010033350178661633,
|
| 5037 |
+
"loss": 0.9244,
|
| 5038 |
+
"step": 715
|
| 5039 |
+
},
|
| 5040 |
+
{
|
| 5041 |
+
"epoch": 0.20140646976090015,
|
| 5042 |
+
"grad_norm": 0.3714972138404846,
|
| 5043 |
+
"learning_rate": 0.00010011116744538272,
|
| 5044 |
+
"loss": 0.6263,
|
| 5045 |
+
"step": 716
|
| 5046 |
+
},
|
| 5047 |
+
{
|
| 5048 |
+
"epoch": 0.20168776371308017,
|
| 5049 |
+
"grad_norm": 0.5079951286315918,
|
| 5050 |
+
"learning_rate": 9.988883255461728e-05,
|
| 5051 |
+
"loss": 0.7556,
|
| 5052 |
+
"step": 717
|
| 5053 |
+
},
|
| 5054 |
+
{
|
| 5055 |
+
"epoch": 0.2019690576652602,
|
| 5056 |
+
"grad_norm": 0.5108062624931335,
|
| 5057 |
+
"learning_rate": 9.966649821338368e-05,
|
| 5058 |
+
"loss": 0.6458,
|
| 5059 |
+
"step": 718
|
| 5060 |
+
},
|
| 5061 |
+
{
|
| 5062 |
+
"epoch": 0.20225035161744023,
|
| 5063 |
+
"grad_norm": 0.47007912397384644,
|
| 5064 |
+
"learning_rate": 9.944416552074295e-05,
|
| 5065 |
+
"loss": 0.6839,
|
| 5066 |
+
"step": 719
|
| 5067 |
+
},
|
| 5068 |
+
{
|
| 5069 |
+
"epoch": 0.20253164556962025,
|
| 5070 |
+
"grad_norm": 0.4043806791305542,
|
| 5071 |
+
"learning_rate": 9.922183557574785e-05,
|
| 5072 |
+
"loss": 0.784,
|
| 5073 |
+
"step": 720
|
| 5074 |
+
},
|
| 5075 |
+
{
|
| 5076 |
+
"epoch": 0.2028129395218003,
|
| 5077 |
+
"grad_norm": 0.3821185231208801,
|
| 5078 |
+
"learning_rate": 9.899950947743767e-05,
|
| 5079 |
+
"loss": 0.431,
|
| 5080 |
+
"step": 721
|
| 5081 |
+
},
|
| 5082 |
+
{
|
| 5083 |
+
"epoch": 0.2030942334739803,
|
| 5084 |
+
"grad_norm": 0.4780855178833008,
|
| 5085 |
+
"learning_rate": 9.877718832483268e-05,
|
| 5086 |
+
"loss": 0.8596,
|
| 5087 |
+
"step": 722
|
| 5088 |
+
},
|
| 5089 |
+
{
|
| 5090 |
+
"epoch": 0.20337552742616033,
|
| 5091 |
+
"grad_norm": 0.4073875844478607,
|
| 5092 |
+
"learning_rate": 9.85548732169286e-05,
|
| 5093 |
+
"loss": 0.6402,
|
| 5094 |
+
"step": 723
|
| 5095 |
+
},
|
| 5096 |
+
{
|
| 5097 |
+
"epoch": 0.20365682137834037,
|
| 5098 |
+
"grad_norm": 0.36500582098960876,
|
| 5099 |
+
"learning_rate": 9.833256525269137e-05,
|
| 5100 |
+
"loss": 0.5438,
|
| 5101 |
+
"step": 724
|
| 5102 |
+
},
|
| 5103 |
+
{
|
| 5104 |
+
"epoch": 0.2039381153305204,
|
| 5105 |
+
"grad_norm": 0.4302322268486023,
|
| 5106 |
+
"learning_rate": 9.811026553105161e-05,
|
| 5107 |
+
"loss": 0.6853,
|
| 5108 |
+
"step": 725
|
| 5109 |
+
},
|
| 5110 |
+
{
|
| 5111 |
+
"epoch": 0.2042194092827004,
|
| 5112 |
+
"grad_norm": 0.46161016821861267,
|
| 5113 |
+
"learning_rate": 9.788797515089911e-05,
|
| 5114 |
+
"loss": 0.8635,
|
| 5115 |
+
"step": 726
|
| 5116 |
+
},
|
| 5117 |
+
{
|
| 5118 |
+
"epoch": 0.20450070323488045,
|
| 5119 |
+
"grad_norm": 0.4550755023956299,
|
| 5120 |
+
"learning_rate": 9.766569521107761e-05,
|
| 5121 |
+
"loss": 0.7477,
|
| 5122 |
+
"step": 727
|
| 5123 |
+
},
|
| 5124 |
+
{
|
| 5125 |
+
"epoch": 0.20478199718706047,
|
| 5126 |
+
"grad_norm": 0.5084179043769836,
|
| 5127 |
+
"learning_rate": 9.744342681037916e-05,
|
| 5128 |
+
"loss": 0.7792,
|
| 5129 |
+
"step": 728
|
| 5130 |
+
},
|
| 5131 |
+
{
|
| 5132 |
+
"epoch": 0.20506329113924052,
|
| 5133 |
+
"grad_norm": 0.3855794370174408,
|
| 5134 |
+
"learning_rate": 9.72211710475387e-05,
|
| 5135 |
+
"loss": 0.5083,
|
| 5136 |
+
"step": 729
|
| 5137 |
+
},
|
| 5138 |
+
{
|
| 5139 |
+
"epoch": 0.20534458509142053,
|
| 5140 |
+
"grad_norm": 0.43810632824897766,
|
| 5141 |
+
"learning_rate": 9.699892902122886e-05,
|
| 5142 |
+
"loss": 0.5203,
|
| 5143 |
+
"step": 730
|
| 5144 |
+
},
|
| 5145 |
+
{
|
| 5146 |
+
"epoch": 0.20562587904360055,
|
| 5147 |
+
"grad_norm": 0.4352615177631378,
|
| 5148 |
+
"learning_rate": 9.67767018300543e-05,
|
| 5149 |
+
"loss": 0.6761,
|
| 5150 |
+
"step": 731
|
| 5151 |
+
},
|
| 5152 |
+
{
|
| 5153 |
+
"epoch": 0.2059071729957806,
|
| 5154 |
+
"grad_norm": 0.33978763222694397,
|
| 5155 |
+
"learning_rate": 9.655449057254625e-05,
|
| 5156 |
+
"loss": 0.2973,
|
| 5157 |
+
"step": 732
|
| 5158 |
+
},
|
| 5159 |
+
{
|
| 5160 |
+
"epoch": 0.2061884669479606,
|
| 5161 |
+
"grad_norm": 0.4740019142627716,
|
| 5162 |
+
"learning_rate": 9.633229634715733e-05,
|
| 5163 |
+
"loss": 0.9421,
|
| 5164 |
+
"step": 733
|
| 5165 |
+
},
|
| 5166 |
+
{
|
| 5167 |
+
"epoch": 0.20646976090014066,
|
| 5168 |
+
"grad_norm": 0.3466711640357971,
|
| 5169 |
+
"learning_rate": 9.611012025225587e-05,
|
| 5170 |
+
"loss": 0.435,
|
| 5171 |
+
"step": 734
|
| 5172 |
+
},
|
| 5173 |
+
{
|
| 5174 |
+
"epoch": 0.20675105485232068,
|
| 5175 |
+
"grad_norm": 0.47393620014190674,
|
| 5176 |
+
"learning_rate": 9.588796338612054e-05,
|
| 5177 |
+
"loss": 0.6685,
|
| 5178 |
+
"step": 735
|
| 5179 |
+
},
|
| 5180 |
+
{
|
| 5181 |
+
"epoch": 0.2070323488045007,
|
| 5182 |
+
"grad_norm": 0.3727364242076874,
|
| 5183 |
+
"learning_rate": 9.56658268469351e-05,
|
| 5184 |
+
"loss": 0.4699,
|
| 5185 |
+
"step": 736
|
| 5186 |
+
},
|
| 5187 |
+
{
|
| 5188 |
+
"epoch": 0.20731364275668074,
|
| 5189 |
+
"grad_norm": 0.4060613811016083,
|
| 5190 |
+
"learning_rate": 9.544371173278272e-05,
|
| 5191 |
+
"loss": 0.4513,
|
| 5192 |
+
"step": 737
|
| 5193 |
+
},
|
| 5194 |
+
{
|
| 5195 |
+
"epoch": 0.20759493670886076,
|
| 5196 |
+
"grad_norm": 0.4083583950996399,
|
| 5197 |
+
"learning_rate": 9.522161914164063e-05,
|
| 5198 |
+
"loss": 0.4598,
|
| 5199 |
+
"step": 738
|
| 5200 |
+
},
|
| 5201 |
+
{
|
| 5202 |
+
"epoch": 0.2078762306610408,
|
| 5203 |
+
"grad_norm": 0.4243644177913666,
|
| 5204 |
+
"learning_rate": 9.499955017137484e-05,
|
| 5205 |
+
"loss": 0.5599,
|
| 5206 |
+
"step": 739
|
| 5207 |
+
},
|
| 5208 |
+
{
|
| 5209 |
+
"epoch": 0.20815752461322082,
|
| 5210 |
+
"grad_norm": 0.426358699798584,
|
| 5211 |
+
"learning_rate": 9.477750591973452e-05,
|
| 5212 |
+
"loss": 0.6165,
|
| 5213 |
+
"step": 740
|
| 5214 |
+
},
|
| 5215 |
+
{
|
| 5216 |
+
"epoch": 0.20843881856540084,
|
| 5217 |
+
"grad_norm": 0.478776216506958,
|
| 5218 |
+
"learning_rate": 9.455548748434661e-05,
|
| 5219 |
+
"loss": 0.6198,
|
| 5220 |
+
"step": 741
|
| 5221 |
+
},
|
| 5222 |
+
{
|
| 5223 |
+
"epoch": 0.20872011251758088,
|
| 5224 |
+
"grad_norm": 0.3706241250038147,
|
| 5225 |
+
"learning_rate": 9.433349596271052e-05,
|
| 5226 |
+
"loss": 0.5018,
|
| 5227 |
+
"step": 742
|
| 5228 |
+
},
|
| 5229 |
+
{
|
| 5230 |
+
"epoch": 0.2090014064697609,
|
| 5231 |
+
"grad_norm": 0.40418341755867004,
|
| 5232 |
+
"learning_rate": 9.411153245219261e-05,
|
| 5233 |
+
"loss": 0.6884,
|
| 5234 |
+
"step": 743
|
| 5235 |
+
},
|
| 5236 |
+
{
|
| 5237 |
+
"epoch": 0.20928270042194091,
|
| 5238 |
+
"grad_norm": 0.4024648666381836,
|
| 5239 |
+
"learning_rate": 9.388959805002066e-05,
|
| 5240 |
+
"loss": 0.6593,
|
| 5241 |
+
"step": 744
|
| 5242 |
+
},
|
| 5243 |
+
{
|
| 5244 |
+
"epoch": 0.20956399437412096,
|
| 5245 |
+
"grad_norm": 0.44085848331451416,
|
| 5246 |
+
"learning_rate": 9.366769385327874e-05,
|
| 5247 |
+
"loss": 0.5811,
|
| 5248 |
+
"step": 745
|
| 5249 |
+
},
|
| 5250 |
+
{
|
| 5251 |
+
"epoch": 0.20984528832630098,
|
| 5252 |
+
"grad_norm": 0.4179717004299164,
|
| 5253 |
+
"learning_rate": 9.344582095890146e-05,
|
| 5254 |
+
"loss": 0.5795,
|
| 5255 |
+
"step": 746
|
| 5256 |
+
},
|
| 5257 |
+
{
|
| 5258 |
+
"epoch": 0.21012658227848102,
|
| 5259 |
+
"grad_norm": 0.4396376609802246,
|
| 5260 |
+
"learning_rate": 9.322398046366871e-05,
|
| 5261 |
+
"loss": 0.6511,
|
| 5262 |
+
"step": 747
|
| 5263 |
+
},
|
| 5264 |
+
{
|
| 5265 |
+
"epoch": 0.21040787623066104,
|
| 5266 |
+
"grad_norm": 0.4031636118888855,
|
| 5267 |
+
"learning_rate": 9.30021734642003e-05,
|
| 5268 |
+
"loss": 0.4475,
|
| 5269 |
+
"step": 748
|
| 5270 |
+
},
|
| 5271 |
+
{
|
| 5272 |
+
"epoch": 0.21068917018284106,
|
| 5273 |
+
"grad_norm": 0.38250279426574707,
|
| 5274 |
+
"learning_rate": 9.278040105695043e-05,
|
| 5275 |
+
"loss": 0.531,
|
| 5276 |
+
"step": 749
|
| 5277 |
+
},
|
| 5278 |
+
{
|
| 5279 |
+
"epoch": 0.2109704641350211,
|
| 5280 |
+
"grad_norm": 0.47047433257102966,
|
| 5281 |
+
"learning_rate": 9.255866433820221e-05,
|
| 5282 |
+
"loss": 0.6456,
|
| 5283 |
+
"step": 750
|
| 5284 |
+
},
|
| 5285 |
+
{
|
| 5286 |
+
"epoch": 0.21125175808720112,
|
| 5287 |
+
"grad_norm": 0.46003010869026184,
|
| 5288 |
+
"learning_rate": 9.233696440406247e-05,
|
| 5289 |
+
"loss": 0.6483,
|
| 5290 |
+
"step": 751
|
| 5291 |
+
},
|
| 5292 |
+
{
|
| 5293 |
+
"epoch": 0.21153305203938116,
|
| 5294 |
+
"grad_norm": 0.4955804646015167,
|
| 5295 |
+
"learning_rate": 9.211530235045618e-05,
|
| 5296 |
+
"loss": 0.6255,
|
| 5297 |
+
"step": 752
|
| 5298 |
+
},
|
| 5299 |
+
{
|
| 5300 |
+
"epoch": 0.21181434599156118,
|
| 5301 |
+
"grad_norm": 0.38930419087409973,
|
| 5302 |
+
"learning_rate": 9.18936792731209e-05,
|
| 5303 |
+
"loss": 0.5435,
|
| 5304 |
+
"step": 753
|
| 5305 |
+
},
|
| 5306 |
+
{
|
| 5307 |
+
"epoch": 0.2120956399437412,
|
| 5308 |
+
"grad_norm": 0.41098153591156006,
|
| 5309 |
+
"learning_rate": 9.167209626760173e-05,
|
| 5310 |
+
"loss": 0.5624,
|
| 5311 |
+
"step": 754
|
| 5312 |
+
},
|
| 5313 |
+
{
|
| 5314 |
+
"epoch": 0.21237693389592124,
|
| 5315 |
+
"grad_norm": 0.44852128624916077,
|
| 5316 |
+
"learning_rate": 9.14505544292456e-05,
|
| 5317 |
+
"loss": 0.7354,
|
| 5318 |
+
"step": 755
|
| 5319 |
+
},
|
| 5320 |
+
{
|
| 5321 |
+
"epoch": 0.21265822784810126,
|
| 5322 |
+
"grad_norm": 0.4212872385978699,
|
| 5323 |
+
"learning_rate": 9.122905485319583e-05,
|
| 5324 |
+
"loss": 0.58,
|
| 5325 |
+
"step": 756
|
| 5326 |
+
},
|
| 5327 |
+
{
|
| 5328 |
+
"epoch": 0.2129395218002813,
|
| 5329 |
+
"grad_norm": 0.40229007601737976,
|
| 5330 |
+
"learning_rate": 9.100759863438702e-05,
|
| 5331 |
+
"loss": 0.5653,
|
| 5332 |
+
"step": 757
|
| 5333 |
+
},
|
| 5334 |
+
{
|
| 5335 |
+
"epoch": 0.21322081575246132,
|
| 5336 |
+
"grad_norm": 0.4235231876373291,
|
| 5337 |
+
"learning_rate": 9.078618686753934e-05,
|
| 5338 |
+
"loss": 0.7402,
|
| 5339 |
+
"step": 758
|
| 5340 |
+
},
|
| 5341 |
+
{
|
| 5342 |
+
"epoch": 0.21350210970464134,
|
| 5343 |
+
"grad_norm": 0.4217592775821686,
|
| 5344 |
+
"learning_rate": 9.056482064715315e-05,
|
| 5345 |
+
"loss": 0.6205,
|
| 5346 |
+
"step": 759
|
| 5347 |
+
},
|
| 5348 |
+
{
|
| 5349 |
+
"epoch": 0.21378340365682139,
|
| 5350 |
+
"grad_norm": 0.37880149483680725,
|
| 5351 |
+
"learning_rate": 9.034350106750383e-05,
|
| 5352 |
+
"loss": 0.5709,
|
| 5353 |
+
"step": 760
|
| 5354 |
+
},
|
| 5355 |
+
{
|
| 5356 |
+
"epoch": 0.2140646976090014,
|
| 5357 |
+
"grad_norm": 0.4029460847377777,
|
| 5358 |
+
"learning_rate": 9.012222922263609e-05,
|
| 5359 |
+
"loss": 0.6891,
|
| 5360 |
+
"step": 761
|
| 5361 |
+
},
|
| 5362 |
+
{
|
| 5363 |
+
"epoch": 0.21434599156118145,
|
| 5364 |
+
"grad_norm": 0.43274563550949097,
|
| 5365 |
+
"learning_rate": 8.990100620635865e-05,
|
| 5366 |
+
"loss": 0.5444,
|
| 5367 |
+
"step": 762
|
| 5368 |
+
},
|
| 5369 |
+
{
|
| 5370 |
+
"epoch": 0.21462728551336147,
|
| 5371 |
+
"grad_norm": 0.3894621431827545,
|
| 5372 |
+
"learning_rate": 8.967983311223897e-05,
|
| 5373 |
+
"loss": 0.4634,
|
| 5374 |
+
"step": 763
|
| 5375 |
+
},
|
| 5376 |
+
{
|
| 5377 |
+
"epoch": 0.21490857946554148,
|
| 5378 |
+
"grad_norm": 0.44674593210220337,
|
| 5379 |
+
"learning_rate": 8.945871103359762e-05,
|
| 5380 |
+
"loss": 0.8734,
|
| 5381 |
+
"step": 764
|
| 5382 |
+
},
|
| 5383 |
+
{
|
| 5384 |
+
"epoch": 0.21518987341772153,
|
| 5385 |
+
"grad_norm": 0.40442943572998047,
|
| 5386 |
+
"learning_rate": 8.923764106350304e-05,
|
| 5387 |
+
"loss": 0.43,
|
| 5388 |
+
"step": 765
|
| 5389 |
+
},
|
| 5390 |
+
{
|
| 5391 |
+
"epoch": 0.21547116736990155,
|
| 5392 |
+
"grad_norm": 0.3951616585254669,
|
| 5393 |
+
"learning_rate": 8.901662429476607e-05,
|
| 5394 |
+
"loss": 0.5221,
|
| 5395 |
+
"step": 766
|
| 5396 |
+
},
|
| 5397 |
+
{
|
| 5398 |
+
"epoch": 0.21575246132208156,
|
| 5399 |
+
"grad_norm": 0.4749213755130768,
|
| 5400 |
+
"learning_rate": 8.87956618199346e-05,
|
| 5401 |
+
"loss": 0.679,
|
| 5402 |
+
"step": 767
|
| 5403 |
+
},
|
| 5404 |
+
{
|
| 5405 |
+
"epoch": 0.2160337552742616,
|
| 5406 |
+
"grad_norm": 0.43756482005119324,
|
| 5407 |
+
"learning_rate": 8.8574754731288e-05,
|
| 5408 |
+
"loss": 0.6271,
|
| 5409 |
+
"step": 768
|
| 5410 |
+
},
|
| 5411 |
+
{
|
| 5412 |
+
"epoch": 0.21631504922644162,
|
| 5413 |
+
"grad_norm": 0.4354095160961151,
|
| 5414 |
+
"learning_rate": 8.835390412083205e-05,
|
| 5415 |
+
"loss": 0.6499,
|
| 5416 |
+
"step": 769
|
| 5417 |
+
},
|
| 5418 |
+
{
|
| 5419 |
+
"epoch": 0.21659634317862167,
|
| 5420 |
+
"grad_norm": 0.432358980178833,
|
| 5421 |
+
"learning_rate": 8.813311108029319e-05,
|
| 5422 |
+
"loss": 0.5587,
|
| 5423 |
+
"step": 770
|
| 5424 |
+
},
|
| 5425 |
+
{
|
| 5426 |
+
"epoch": 0.2168776371308017,
|
| 5427 |
+
"grad_norm": 0.34371715784072876,
|
| 5428 |
+
"learning_rate": 8.791237670111329e-05,
|
| 5429 |
+
"loss": 0.4995,
|
| 5430 |
+
"step": 771
|
| 5431 |
+
},
|
| 5432 |
+
{
|
| 5433 |
+
"epoch": 0.2171589310829817,
|
| 5434 |
+
"grad_norm": 0.49070432782173157,
|
| 5435 |
+
"learning_rate": 8.769170207444431e-05,
|
| 5436 |
+
"loss": 0.8673,
|
| 5437 |
+
"step": 772
|
| 5438 |
+
},
|
| 5439 |
+
{
|
| 5440 |
+
"epoch": 0.21744022503516175,
|
| 5441 |
+
"grad_norm": 0.44228169322013855,
|
| 5442 |
+
"learning_rate": 8.747108829114283e-05,
|
| 5443 |
+
"loss": 0.6069,
|
| 5444 |
+
"step": 773
|
| 5445 |
+
},
|
| 5446 |
+
{
|
| 5447 |
+
"epoch": 0.21772151898734177,
|
| 5448 |
+
"grad_norm": 0.43570342659950256,
|
| 5449 |
+
"learning_rate": 8.725053644176456e-05,
|
| 5450 |
+
"loss": 0.5758,
|
| 5451 |
+
"step": 774
|
| 5452 |
+
},
|
| 5453 |
+
{
|
| 5454 |
+
"epoch": 0.2180028129395218,
|
| 5455 |
+
"grad_norm": 0.4716907739639282,
|
| 5456 |
+
"learning_rate": 8.703004761655917e-05,
|
| 5457 |
+
"loss": 0.7674,
|
| 5458 |
+
"step": 775
|
| 5459 |
+
},
|
| 5460 |
+
{
|
| 5461 |
+
"epoch": 0.21828410689170183,
|
| 5462 |
+
"grad_norm": 0.44911226630210876,
|
| 5463 |
+
"learning_rate": 8.680962290546477e-05,
|
| 5464 |
+
"loss": 0.6778,
|
| 5465 |
+
"step": 776
|
| 5466 |
+
},
|
| 5467 |
+
{
|
| 5468 |
+
"epoch": 0.21856540084388185,
|
| 5469 |
+
"grad_norm": 0.4381526708602905,
|
| 5470 |
+
"learning_rate": 8.658926339810242e-05,
|
| 5471 |
+
"loss": 0.7249,
|
| 5472 |
+
"step": 777
|
| 5473 |
+
},
|
| 5474 |
+
{
|
| 5475 |
+
"epoch": 0.2188466947960619,
|
| 5476 |
+
"grad_norm": 0.48972153663635254,
|
| 5477 |
+
"learning_rate": 8.636897018377102e-05,
|
| 5478 |
+
"loss": 0.5147,
|
| 5479 |
+
"step": 778
|
| 5480 |
+
},
|
| 5481 |
+
{
|
| 5482 |
+
"epoch": 0.2191279887482419,
|
| 5483 |
+
"grad_norm": 0.44663965702056885,
|
| 5484 |
+
"learning_rate": 8.614874435144169e-05,
|
| 5485 |
+
"loss": 0.692,
|
| 5486 |
+
"step": 779
|
| 5487 |
+
},
|
| 5488 |
+
{
|
| 5489 |
+
"epoch": 0.21940928270042195,
|
| 5490 |
+
"grad_norm": 0.4131266176700592,
|
| 5491 |
+
"learning_rate": 8.59285869897524e-05,
|
| 5492 |
+
"loss": 0.6216,
|
| 5493 |
+
"step": 780
|
| 5494 |
+
},
|
| 5495 |
+
{
|
| 5496 |
+
"epoch": 0.21969057665260197,
|
| 5497 |
+
"grad_norm": 0.411931574344635,
|
| 5498 |
+
"learning_rate": 8.570849918700278e-05,
|
| 5499 |
+
"loss": 0.5549,
|
| 5500 |
+
"step": 781
|
| 5501 |
+
},
|
| 5502 |
+
{
|
| 5503 |
+
"epoch": 0.219971870604782,
|
| 5504 |
+
"grad_norm": 0.44478118419647217,
|
| 5505 |
+
"learning_rate": 8.548848203114854e-05,
|
| 5506 |
+
"loss": 0.8206,
|
| 5507 |
+
"step": 782
|
| 5508 |
+
},
|
| 5509 |
+
{
|
| 5510 |
+
"epoch": 0.22025316455696203,
|
| 5511 |
+
"grad_norm": 0.4329167902469635,
|
| 5512 |
+
"learning_rate": 8.526853660979608e-05,
|
| 5513 |
+
"loss": 0.7115,
|
| 5514 |
+
"step": 783
|
| 5515 |
+
},
|
| 5516 |
+
{
|
| 5517 |
+
"epoch": 0.22053445850914205,
|
| 5518 |
+
"grad_norm": 0.48116809129714966,
|
| 5519 |
+
"learning_rate": 8.504866401019737e-05,
|
| 5520 |
+
"loss": 0.4649,
|
| 5521 |
+
"step": 784
|
| 5522 |
+
},
|
| 5523 |
+
{
|
| 5524 |
+
"epoch": 0.22081575246132207,
|
| 5525 |
+
"grad_norm": 0.4212972819805145,
|
| 5526 |
+
"learning_rate": 8.482886531924433e-05,
|
| 5527 |
+
"loss": 0.6371,
|
| 5528 |
+
"step": 785
|
| 5529 |
+
},
|
| 5530 |
+
{
|
| 5531 |
+
"epoch": 0.2210970464135021,
|
| 5532 |
+
"grad_norm": 0.37539881467819214,
|
| 5533 |
+
"learning_rate": 8.460914162346343e-05,
|
| 5534 |
+
"loss": 0.5898,
|
| 5535 |
+
"step": 786
|
| 5536 |
+
},
|
| 5537 |
+
{
|
| 5538 |
+
"epoch": 0.22137834036568213,
|
| 5539 |
+
"grad_norm": 0.3806806802749634,
|
| 5540 |
+
"learning_rate": 8.438949400901056e-05,
|
| 5541 |
+
"loss": 0.5613,
|
| 5542 |
+
"step": 787
|
| 5543 |
+
},
|
| 5544 |
+
{
|
| 5545 |
+
"epoch": 0.22165963431786218,
|
| 5546 |
+
"grad_norm": 0.4030994176864624,
|
| 5547 |
+
"learning_rate": 8.416992356166547e-05,
|
| 5548 |
+
"loss": 0.6001,
|
| 5549 |
+
"step": 788
|
| 5550 |
+
},
|
| 5551 |
+
{
|
| 5552 |
+
"epoch": 0.2219409282700422,
|
| 5553 |
+
"grad_norm": 0.42301586270332336,
|
| 5554 |
+
"learning_rate": 8.395043136682638e-05,
|
| 5555 |
+
"loss": 0.5865,
|
| 5556 |
+
"step": 789
|
| 5557 |
+
},
|
| 5558 |
+
{
|
| 5559 |
+
"epoch": 0.2222222222222222,
|
| 5560 |
+
"grad_norm": 0.41840770840644836,
|
| 5561 |
+
"learning_rate": 8.37310185095048e-05,
|
| 5562 |
+
"loss": 0.6001,
|
| 5563 |
+
"step": 790
|
| 5564 |
+
},
|
| 5565 |
+
{
|
| 5566 |
+
"epoch": 0.22250351617440225,
|
| 5567 |
+
"grad_norm": 0.4628492295742035,
|
| 5568 |
+
"learning_rate": 8.351168607432e-05,
|
| 5569 |
+
"loss": 0.8088,
|
| 5570 |
+
"step": 791
|
| 5571 |
+
},
|
| 5572 |
+
{
|
| 5573 |
+
"epoch": 0.22278481012658227,
|
| 5574 |
+
"grad_norm": 0.42985421419143677,
|
| 5575 |
+
"learning_rate": 8.329243514549365e-05,
|
| 5576 |
+
"loss": 0.6074,
|
| 5577 |
+
"step": 792
|
| 5578 |
+
},
|
| 5579 |
+
{
|
| 5580 |
+
"epoch": 0.22306610407876232,
|
| 5581 |
+
"grad_norm": 0.4796198308467865,
|
| 5582 |
+
"learning_rate": 8.307326680684461e-05,
|
| 5583 |
+
"loss": 0.7649,
|
| 5584 |
+
"step": 793
|
| 5585 |
+
},
|
| 5586 |
+
{
|
| 5587 |
+
"epoch": 0.22334739803094233,
|
| 5588 |
+
"grad_norm": 0.40142351388931274,
|
| 5589 |
+
"learning_rate": 8.285418214178343e-05,
|
| 5590 |
+
"loss": 0.484,
|
| 5591 |
+
"step": 794
|
| 5592 |
+
},
|
| 5593 |
+
{
|
| 5594 |
+
"epoch": 0.22362869198312235,
|
| 5595 |
+
"grad_norm": 0.4510243237018585,
|
| 5596 |
+
"learning_rate": 8.263518223330697e-05,
|
| 5597 |
+
"loss": 0.65,
|
| 5598 |
+
"step": 795
|
| 5599 |
+
},
|
| 5600 |
+
{
|
| 5601 |
+
"epoch": 0.2239099859353024,
|
| 5602 |
+
"grad_norm": 0.38650351762771606,
|
| 5603 |
+
"learning_rate": 8.241626816399325e-05,
|
| 5604 |
+
"loss": 0.473,
|
| 5605 |
+
"step": 796
|
| 5606 |
+
},
|
| 5607 |
+
{
|
| 5608 |
+
"epoch": 0.22419127988748241,
|
| 5609 |
+
"grad_norm": 0.43197983503341675,
|
| 5610 |
+
"learning_rate": 8.219744101599588e-05,
|
| 5611 |
+
"loss": 0.5828,
|
| 5612 |
+
"step": 797
|
| 5613 |
+
},
|
| 5614 |
+
{
|
| 5615 |
+
"epoch": 0.22447257383966246,
|
| 5616 |
+
"grad_norm": 0.47523319721221924,
|
| 5617 |
+
"learning_rate": 8.197870187103873e-05,
|
| 5618 |
+
"loss": 0.5105,
|
| 5619 |
+
"step": 798
|
| 5620 |
+
},
|
| 5621 |
+
{
|
| 5622 |
+
"epoch": 0.22475386779184248,
|
| 5623 |
+
"grad_norm": 0.4083631634712219,
|
| 5624 |
+
"learning_rate": 8.176005181041082e-05,
|
| 5625 |
+
"loss": 0.7744,
|
| 5626 |
+
"step": 799
|
| 5627 |
+
},
|
| 5628 |
+
{
|
| 5629 |
+
"epoch": 0.2250351617440225,
|
| 5630 |
+
"grad_norm": 0.42829808592796326,
|
| 5631 |
+
"learning_rate": 8.154149191496067e-05,
|
| 5632 |
+
"loss": 0.8228,
|
| 5633 |
+
"step": 800
|
| 5634 |
+
},
|
| 5635 |
+
{
|
| 5636 |
+
"epoch": 0.22531645569620254,
|
| 5637 |
+
"grad_norm": 0.41811439394950867,
|
| 5638 |
+
"learning_rate": 8.132302326509105e-05,
|
| 5639 |
+
"loss": 0.6907,
|
| 5640 |
+
"step": 801
|
| 5641 |
+
},
|
| 5642 |
+
{
|
| 5643 |
+
"epoch": 0.22559774964838256,
|
| 5644 |
+
"grad_norm": 0.3861621022224426,
|
| 5645 |
+
"learning_rate": 8.110464694075383e-05,
|
| 5646 |
+
"loss": 0.3678,
|
| 5647 |
+
"step": 802
|
| 5648 |
+
},
|
| 5649 |
+
{
|
| 5650 |
+
"epoch": 0.2258790436005626,
|
| 5651 |
+
"grad_norm": 0.44829320907592773,
|
| 5652 |
+
"learning_rate": 8.088636402144441e-05,
|
| 5653 |
+
"loss": 0.7495,
|
| 5654 |
+
"step": 803
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 0.22616033755274262,
|
| 5658 |
+
"grad_norm": 0.451870322227478,
|
| 5659 |
+
"learning_rate": 8.066817558619639e-05,
|
| 5660 |
+
"loss": 0.7644,
|
| 5661 |
+
"step": 804
|
| 5662 |
+
},
|
| 5663 |
+
{
|
| 5664 |
+
"epoch": 0.22644163150492264,
|
| 5665 |
+
"grad_norm": 0.46685147285461426,
|
| 5666 |
+
"learning_rate": 8.045008271357643e-05,
|
| 5667 |
+
"loss": 0.8189,
|
| 5668 |
+
"step": 805
|
| 5669 |
+
},
|
| 5670 |
+
{
|
| 5671 |
+
"epoch": 0.22672292545710268,
|
| 5672 |
+
"grad_norm": 0.4607045352458954,
|
| 5673 |
+
"learning_rate": 8.023208648167877e-05,
|
| 5674 |
+
"loss": 0.6859,
|
| 5675 |
+
"step": 806
|
| 5676 |
+
},
|
| 5677 |
+
{
|
| 5678 |
+
"epoch": 0.2270042194092827,
|
| 5679 |
+
"grad_norm": 0.4235568940639496,
|
| 5680 |
+
"learning_rate": 8.001418796811981e-05,
|
| 5681 |
+
"loss": 0.5295,
|
| 5682 |
+
"step": 807
|
| 5683 |
+
},
|
| 5684 |
+
{
|
| 5685 |
+
"epoch": 0.22728551336146272,
|
| 5686 |
+
"grad_norm": 0.45782768726348877,
|
| 5687 |
+
"learning_rate": 7.979638825003307e-05,
|
| 5688 |
+
"loss": 0.6722,
|
| 5689 |
+
"step": 808
|
| 5690 |
+
},
|
| 5691 |
+
{
|
| 5692 |
+
"epoch": 0.22756680731364276,
|
| 5693 |
+
"grad_norm": 0.5132092833518982,
|
| 5694 |
+
"learning_rate": 7.957868840406361e-05,
|
| 5695 |
+
"loss": 0.7794,
|
| 5696 |
+
"step": 809
|
| 5697 |
+
},
|
| 5698 |
+
{
|
| 5699 |
+
"epoch": 0.22784810126582278,
|
| 5700 |
+
"grad_norm": 0.44531407952308655,
|
| 5701 |
+
"learning_rate": 7.936108950636277e-05,
|
| 5702 |
+
"loss": 0.7547,
|
| 5703 |
+
"step": 810
|
| 5704 |
+
},
|
| 5705 |
+
{
|
| 5706 |
+
"epoch": 0.22812939521800282,
|
| 5707 |
+
"grad_norm": 0.42250415682792664,
|
| 5708 |
+
"learning_rate": 7.914359263258295e-05,
|
| 5709 |
+
"loss": 0.4298,
|
| 5710 |
+
"step": 811
|
| 5711 |
+
},
|
| 5712 |
+
{
|
| 5713 |
+
"epoch": 0.22841068917018284,
|
| 5714 |
+
"grad_norm": 0.3803527057170868,
|
| 5715 |
+
"learning_rate": 7.892619885787218e-05,
|
| 5716 |
+
"loss": 0.3989,
|
| 5717 |
+
"step": 812
|
| 5718 |
+
},
|
| 5719 |
+
{
|
| 5720 |
+
"epoch": 0.22869198312236286,
|
| 5721 |
+
"grad_norm": 0.4728476405143738,
|
| 5722 |
+
"learning_rate": 7.870890925686875e-05,
|
| 5723 |
+
"loss": 0.5643,
|
| 5724 |
+
"step": 813
|
| 5725 |
+
},
|
| 5726 |
+
{
|
| 5727 |
+
"epoch": 0.2289732770745429,
|
| 5728 |
+
"grad_norm": 0.4278518259525299,
|
| 5729 |
+
"learning_rate": 7.84917249036962e-05,
|
| 5730 |
+
"loss": 0.8081,
|
| 5731 |
+
"step": 814
|
| 5732 |
+
},
|
| 5733 |
+
{
|
| 5734 |
+
"epoch": 0.22925457102672292,
|
| 5735 |
+
"grad_norm": 0.40866798162460327,
|
| 5736 |
+
"learning_rate": 7.827464687195763e-05,
|
| 5737 |
+
"loss": 0.5875,
|
| 5738 |
+
"step": 815
|
| 5739 |
+
},
|
| 5740 |
+
{
|
| 5741 |
+
"epoch": 0.22953586497890296,
|
| 5742 |
+
"grad_norm": 0.3823324739933014,
|
| 5743 |
+
"learning_rate": 7.805767623473055e-05,
|
| 5744 |
+
"loss": 0.4294,
|
| 5745 |
+
"step": 816
|
| 5746 |
+
},
|
| 5747 |
+
{
|
| 5748 |
+
"epoch": 0.22981715893108298,
|
| 5749 |
+
"grad_norm": 0.3771815598011017,
|
| 5750 |
+
"learning_rate": 7.784081406456173e-05,
|
| 5751 |
+
"loss": 0.545,
|
| 5752 |
+
"step": 817
|
| 5753 |
+
},
|
| 5754 |
+
{
|
| 5755 |
+
"epoch": 0.230098452883263,
|
| 5756 |
+
"grad_norm": 0.39243924617767334,
|
| 5757 |
+
"learning_rate": 7.762406143346169e-05,
|
| 5758 |
+
"loss": 0.5315,
|
| 5759 |
+
"step": 818
|
| 5760 |
+
},
|
| 5761 |
+
{
|
| 5762 |
+
"epoch": 0.23037974683544304,
|
| 5763 |
+
"grad_norm": 0.48769742250442505,
|
| 5764 |
+
"learning_rate": 7.740741941289935e-05,
|
| 5765 |
+
"loss": 0.7234,
|
| 5766 |
+
"step": 819
|
| 5767 |
+
},
|
| 5768 |
+
{
|
| 5769 |
+
"epoch": 0.23066104078762306,
|
| 5770 |
+
"grad_norm": 0.37934958934783936,
|
| 5771 |
+
"learning_rate": 7.719088907379706e-05,
|
| 5772 |
+
"loss": 0.4085,
|
| 5773 |
+
"step": 820
|
| 5774 |
+
},
|
| 5775 |
+
{
|
| 5776 |
+
"epoch": 0.2309423347398031,
|
| 5777 |
+
"grad_norm": 0.4139319658279419,
|
| 5778 |
+
"learning_rate": 7.697447148652498e-05,
|
| 5779 |
+
"loss": 0.29,
|
| 5780 |
+
"step": 821
|
| 5781 |
+
},
|
| 5782 |
+
{
|
| 5783 |
+
"epoch": 0.23122362869198312,
|
| 5784 |
+
"grad_norm": 0.43514391779899597,
|
| 5785 |
+
"learning_rate": 7.675816772089586e-05,
|
| 5786 |
+
"loss": 0.71,
|
| 5787 |
+
"step": 822
|
| 5788 |
+
},
|
| 5789 |
+
{
|
| 5790 |
+
"epoch": 0.23150492264416314,
|
| 5791 |
+
"grad_norm": 0.40416961908340454,
|
| 5792 |
+
"learning_rate": 7.65419788461599e-05,
|
| 5793 |
+
"loss": 0.62,
|
| 5794 |
+
"step": 823
|
| 5795 |
+
},
|
| 5796 |
+
{
|
| 5797 |
+
"epoch": 0.2317862165963432,
|
| 5798 |
+
"grad_norm": 0.385342538356781,
|
| 5799 |
+
"learning_rate": 7.632590593099935e-05,
|
| 5800 |
+
"loss": 0.4793,
|
| 5801 |
+
"step": 824
|
| 5802 |
+
},
|
| 5803 |
+
{
|
| 5804 |
+
"epoch": 0.2320675105485232,
|
| 5805 |
+
"grad_norm": 0.47173193097114563,
|
| 5806 |
+
"learning_rate": 7.610995004352313e-05,
|
| 5807 |
+
"loss": 0.6565,
|
| 5808 |
+
"step": 825
|
| 5809 |
+
},
|
| 5810 |
+
{
|
| 5811 |
+
"epoch": 0.23234880450070325,
|
| 5812 |
+
"grad_norm": 0.41295987367630005,
|
| 5813 |
+
"learning_rate": 7.589411225126181e-05,
|
| 5814 |
+
"loss": 0.5361,
|
| 5815 |
+
"step": 826
|
| 5816 |
+
},
|
| 5817 |
+
{
|
| 5818 |
+
"epoch": 0.23263009845288327,
|
| 5819 |
+
"grad_norm": 0.4854215681552887,
|
| 5820 |
+
"learning_rate": 7.567839362116209e-05,
|
| 5821 |
+
"loss": 0.8218,
|
| 5822 |
+
"step": 827
|
| 5823 |
+
},
|
| 5824 |
+
{
|
| 5825 |
+
"epoch": 0.23291139240506328,
|
| 5826 |
+
"grad_norm": 0.4299052357673645,
|
| 5827 |
+
"learning_rate": 7.54627952195816e-05,
|
| 5828 |
+
"loss": 0.6896,
|
| 5829 |
+
"step": 828
|
| 5830 |
+
},
|
| 5831 |
+
{
|
| 5832 |
+
"epoch": 0.23319268635724333,
|
| 5833 |
+
"grad_norm": 0.41072535514831543,
|
| 5834 |
+
"learning_rate": 7.524731811228374e-05,
|
| 5835 |
+
"loss": 0.5736,
|
| 5836 |
+
"step": 829
|
| 5837 |
+
},
|
| 5838 |
+
{
|
| 5839 |
+
"epoch": 0.23347398030942335,
|
| 5840 |
+
"grad_norm": 0.4377497136592865,
|
| 5841 |
+
"learning_rate": 7.503196336443226e-05,
|
| 5842 |
+
"loss": 0.6219,
|
| 5843 |
+
"step": 830
|
| 5844 |
+
},
|
| 5845 |
+
{
|
| 5846 |
+
"epoch": 0.23375527426160336,
|
| 5847 |
+
"grad_norm": 0.41371291875839233,
|
| 5848 |
+
"learning_rate": 7.4816732040586e-05,
|
| 5849 |
+
"loss": 0.4856,
|
| 5850 |
+
"step": 831
|
| 5851 |
+
},
|
| 5852 |
+
{
|
| 5853 |
+
"epoch": 0.2340365682137834,
|
| 5854 |
+
"grad_norm": 0.3942505121231079,
|
| 5855 |
+
"learning_rate": 7.460162520469379e-05,
|
| 5856 |
+
"loss": 0.6,
|
| 5857 |
+
"step": 832
|
| 5858 |
+
},
|
| 5859 |
+
{
|
| 5860 |
+
"epoch": 0.23431786216596343,
|
| 5861 |
+
"grad_norm": 0.4336865544319153,
|
| 5862 |
+
"learning_rate": 7.438664392008903e-05,
|
| 5863 |
+
"loss": 0.5894,
|
| 5864 |
+
"step": 833
|
| 5865 |
+
},
|
| 5866 |
+
{
|
| 5867 |
+
"epoch": 0.23459915611814347,
|
| 5868 |
+
"grad_norm": 0.46640902757644653,
|
| 5869 |
+
"learning_rate": 7.417178924948443e-05,
|
| 5870 |
+
"loss": 0.7385,
|
| 5871 |
+
"step": 834
|
| 5872 |
+
},
|
| 5873 |
+
{
|
| 5874 |
+
"epoch": 0.2348804500703235,
|
| 5875 |
+
"grad_norm": 0.4493373930454254,
|
| 5876 |
+
"learning_rate": 7.39570622549669e-05,
|
| 5877 |
+
"loss": 0.7542,
|
| 5878 |
+
"step": 835
|
| 5879 |
+
},
|
| 5880 |
+
{
|
| 5881 |
+
"epoch": 0.2351617440225035,
|
| 5882 |
+
"grad_norm": 0.39611709117889404,
|
| 5883 |
+
"learning_rate": 7.374246399799218e-05,
|
| 5884 |
+
"loss": 0.437,
|
| 5885 |
+
"step": 836
|
| 5886 |
+
},
|
| 5887 |
+
{
|
| 5888 |
+
"epoch": 0.23544303797468355,
|
| 5889 |
+
"grad_norm": 0.4040944576263428,
|
| 5890 |
+
"learning_rate": 7.352799553937966e-05,
|
| 5891 |
+
"loss": 0.5683,
|
| 5892 |
+
"step": 837
|
| 5893 |
+
},
|
| 5894 |
+
{
|
| 5895 |
+
"epoch": 0.23572433192686357,
|
| 5896 |
+
"grad_norm": 0.35686078667640686,
|
| 5897 |
+
"learning_rate": 7.331365793930698e-05,
|
| 5898 |
+
"loss": 0.388,
|
| 5899 |
+
"step": 838
|
| 5900 |
+
},
|
| 5901 |
+
{
|
| 5902 |
+
"epoch": 0.2360056258790436,
|
| 5903 |
+
"grad_norm": 0.4271821677684784,
|
| 5904 |
+
"learning_rate": 7.309945225730501e-05,
|
| 5905 |
+
"loss": 0.5178,
|
| 5906 |
+
"step": 839
|
| 5907 |
+
},
|
| 5908 |
+
{
|
| 5909 |
+
"epoch": 0.23628691983122363,
|
| 5910 |
+
"grad_norm": 0.39256569743156433,
|
| 5911 |
+
"learning_rate": 7.288537955225256e-05,
|
| 5912 |
+
"loss": 0.5825,
|
| 5913 |
+
"step": 840
|
| 5914 |
+
},
|
| 5915 |
+
{
|
| 5916 |
+
"epoch": 0.23656821378340365,
|
| 5917 |
+
"grad_norm": 0.419803261756897,
|
| 5918 |
+
"learning_rate": 7.267144088237094e-05,
|
| 5919 |
+
"loss": 0.6707,
|
| 5920 |
+
"step": 841
|
| 5921 |
+
},
|
| 5922 |
+
{
|
| 5923 |
+
"epoch": 0.2368495077355837,
|
| 5924 |
+
"grad_norm": 0.40984973311424255,
|
| 5925 |
+
"learning_rate": 7.245763730521902e-05,
|
| 5926 |
+
"loss": 0.6235,
|
| 5927 |
+
"step": 842
|
| 5928 |
+
},
|
| 5929 |
+
{
|
| 5930 |
+
"epoch": 0.2371308016877637,
|
| 5931 |
+
"grad_norm": 0.35661789774894714,
|
| 5932 |
+
"learning_rate": 7.224396987768784e-05,
|
| 5933 |
+
"loss": 0.4328,
|
| 5934 |
+
"step": 843
|
| 5935 |
+
},
|
| 5936 |
+
{
|
| 5937 |
+
"epoch": 0.23741209563994375,
|
| 5938 |
+
"grad_norm": 0.3914950489997864,
|
| 5939 |
+
"learning_rate": 7.203043965599537e-05,
|
| 5940 |
+
"loss": 0.6177,
|
| 5941 |
+
"step": 844
|
| 5942 |
+
},
|
| 5943 |
+
{
|
| 5944 |
+
"epoch": 0.23769338959212377,
|
| 5945 |
+
"grad_norm": 0.40430381894111633,
|
| 5946 |
+
"learning_rate": 7.181704769568134e-05,
|
| 5947 |
+
"loss": 0.5992,
|
| 5948 |
+
"step": 845
|
| 5949 |
+
},
|
| 5950 |
+
{
|
| 5951 |
+
"epoch": 0.2379746835443038,
|
| 5952 |
+
"grad_norm": 0.3938002586364746,
|
| 5953 |
+
"learning_rate": 7.160379505160211e-05,
|
| 5954 |
+
"loss": 0.379,
|
| 5955 |
+
"step": 846
|
| 5956 |
+
},
|
| 5957 |
+
{
|
| 5958 |
+
"epoch": 0.23825597749648383,
|
| 5959 |
+
"grad_norm": 0.44991961121559143,
|
| 5960 |
+
"learning_rate": 7.139068277792523e-05,
|
| 5961 |
+
"loss": 0.7602,
|
| 5962 |
+
"step": 847
|
| 5963 |
+
},
|
| 5964 |
+
{
|
| 5965 |
+
"epoch": 0.23853727144866385,
|
| 5966 |
+
"grad_norm": 0.4025498926639557,
|
| 5967 |
+
"learning_rate": 7.117771192812442e-05,
|
| 5968 |
+
"loss": 0.6575,
|
| 5969 |
+
"step": 848
|
| 5970 |
+
},
|
| 5971 |
+
{
|
| 5972 |
+
"epoch": 0.23881856540084387,
|
| 5973 |
+
"grad_norm": 0.3925105631351471,
|
| 5974 |
+
"learning_rate": 7.096488355497435e-05,
|
| 5975 |
+
"loss": 0.6721,
|
| 5976 |
+
"step": 849
|
| 5977 |
+
},
|
| 5978 |
+
{
|
| 5979 |
+
"epoch": 0.2390998593530239,
|
| 5980 |
+
"grad_norm": 0.4031991958618164,
|
| 5981 |
+
"learning_rate": 7.075219871054528e-05,
|
| 5982 |
+
"loss": 0.5951,
|
| 5983 |
+
"step": 850
|
| 5984 |
+
},
|
| 5985 |
+
{
|
| 5986 |
+
"epoch": 0.23938115330520393,
|
| 5987 |
+
"grad_norm": 0.39644140005111694,
|
| 5988 |
+
"learning_rate": 7.053965844619803e-05,
|
| 5989 |
+
"loss": 0.627,
|
| 5990 |
+
"step": 851
|
| 5991 |
+
},
|
| 5992 |
+
{
|
| 5993 |
+
"epoch": 0.23966244725738398,
|
| 5994 |
+
"grad_norm": 0.4215494394302368,
|
| 5995 |
+
"learning_rate": 7.032726381257878e-05,
|
| 5996 |
+
"loss": 0.6483,
|
| 5997 |
+
"step": 852
|
| 5998 |
+
},
|
| 5999 |
+
{
|
| 6000 |
+
"epoch": 0.239943741209564,
|
| 6001 |
+
"grad_norm": 0.46763792634010315,
|
| 6002 |
+
"learning_rate": 7.011501585961369e-05,
|
| 6003 |
+
"loss": 0.8404,
|
| 6004 |
+
"step": 853
|
| 6005 |
+
},
|
| 6006 |
+
{
|
| 6007 |
+
"epoch": 0.240225035161744,
|
| 6008 |
+
"grad_norm": 0.43812862038612366,
|
| 6009 |
+
"learning_rate": 6.990291563650387e-05,
|
| 6010 |
+
"loss": 0.6048,
|
| 6011 |
+
"step": 854
|
| 6012 |
+
},
|
| 6013 |
+
{
|
| 6014 |
+
"epoch": 0.24050632911392406,
|
| 6015 |
+
"grad_norm": 0.4177474081516266,
|
| 6016 |
+
"learning_rate": 6.96909641917203e-05,
|
| 6017 |
+
"loss": 0.5702,
|
| 6018 |
+
"step": 855
|
| 6019 |
+
},
|
| 6020 |
+
{
|
| 6021 |
+
"epoch": 0.24078762306610407,
|
| 6022 |
+
"grad_norm": 0.43847793340682983,
|
| 6023 |
+
"learning_rate": 6.94791625729983e-05,
|
| 6024 |
+
"loss": 0.711,
|
| 6025 |
+
"step": 856
|
| 6026 |
+
},
|
| 6027 |
+
{
|
| 6028 |
+
"epoch": 0.24106891701828412,
|
| 6029 |
+
"grad_norm": 0.44679099321365356,
|
| 6030 |
+
"learning_rate": 6.926751182733265e-05,
|
| 6031 |
+
"loss": 0.543,
|
| 6032 |
+
"step": 857
|
| 6033 |
+
},
|
| 6034 |
+
{
|
| 6035 |
+
"epoch": 0.24135021097046414,
|
| 6036 |
+
"grad_norm": 0.4884023368358612,
|
| 6037 |
+
"learning_rate": 6.905601300097241e-05,
|
| 6038 |
+
"loss": 0.791,
|
| 6039 |
+
"step": 858
|
| 6040 |
+
},
|
| 6041 |
+
{
|
| 6042 |
+
"epoch": 0.24163150492264415,
|
| 6043 |
+
"grad_norm": 0.4562274217605591,
|
| 6044 |
+
"learning_rate": 6.88446671394155e-05,
|
| 6045 |
+
"loss": 0.4497,
|
| 6046 |
+
"step": 859
|
| 6047 |
+
},
|
| 6048 |
+
{
|
| 6049 |
+
"epoch": 0.2419127988748242,
|
| 6050 |
+
"grad_norm": 0.5555647611618042,
|
| 6051 |
+
"learning_rate": 6.863347528740376e-05,
|
| 6052 |
+
"loss": 0.5964,
|
| 6053 |
+
"step": 860
|
| 6054 |
+
},
|
| 6055 |
+
{
|
| 6056 |
+
"epoch": 0.24219409282700421,
|
| 6057 |
+
"grad_norm": 0.3987271189689636,
|
| 6058 |
+
"learning_rate": 6.84224384889178e-05,
|
| 6059 |
+
"loss": 0.4132,
|
| 6060 |
+
"step": 861
|
| 6061 |
+
},
|
| 6062 |
+
{
|
| 6063 |
+
"epoch": 0.24247538677918426,
|
| 6064 |
+
"grad_norm": 0.42632314562797546,
|
| 6065 |
+
"learning_rate": 6.821155778717164e-05,
|
| 6066 |
+
"loss": 0.4722,
|
| 6067 |
+
"step": 862
|
| 6068 |
+
},
|
| 6069 |
+
{
|
| 6070 |
+
"epoch": 0.24275668073136428,
|
| 6071 |
+
"grad_norm": 0.48646071553230286,
|
| 6072 |
+
"learning_rate": 6.800083422460766e-05,
|
| 6073 |
+
"loss": 0.5645,
|
| 6074 |
+
"step": 863
|
| 6075 |
+
},
|
| 6076 |
+
{
|
| 6077 |
+
"epoch": 0.2430379746835443,
|
| 6078 |
+
"grad_norm": 0.3790850043296814,
|
| 6079 |
+
"learning_rate": 6.779026884289162e-05,
|
| 6080 |
+
"loss": 0.5851,
|
| 6081 |
+
"step": 864
|
| 6082 |
+
},
|
| 6083 |
+
{
|
| 6084 |
+
"epoch": 0.24331926863572434,
|
| 6085 |
+
"grad_norm": 0.4753282964229584,
|
| 6086 |
+
"learning_rate": 6.757986268290712e-05,
|
| 6087 |
+
"loss": 0.5518,
|
| 6088 |
+
"step": 865
|
| 6089 |
+
},
|
| 6090 |
+
{
|
| 6091 |
+
"epoch": 0.24360056258790436,
|
| 6092 |
+
"grad_norm": 0.44699737429618835,
|
| 6093 |
+
"learning_rate": 6.736961678475086e-05,
|
| 6094 |
+
"loss": 0.6563,
|
| 6095 |
+
"step": 866
|
| 6096 |
+
},
|
| 6097 |
+
{
|
| 6098 |
+
"epoch": 0.2438818565400844,
|
| 6099 |
+
"grad_norm": 0.4111562669277191,
|
| 6100 |
+
"learning_rate": 6.715953218772732e-05,
|
| 6101 |
+
"loss": 0.4601,
|
| 6102 |
+
"step": 867
|
| 6103 |
+
},
|
| 6104 |
+
{
|
| 6105 |
+
"epoch": 0.24416315049226442,
|
| 6106 |
+
"grad_norm": 0.4261634647846222,
|
| 6107 |
+
"learning_rate": 6.694960993034345e-05,
|
| 6108 |
+
"loss": 0.6538,
|
| 6109 |
+
"step": 868
|
| 6110 |
+
},
|
| 6111 |
+
{
|
| 6112 |
+
"epoch": 0.24444444444444444,
|
| 6113 |
+
"grad_norm": 0.4394458830356598,
|
| 6114 |
+
"learning_rate": 6.673985105030386e-05,
|
| 6115 |
+
"loss": 0.5248,
|
| 6116 |
+
"step": 869
|
| 6117 |
+
},
|
| 6118 |
+
{
|
| 6119 |
+
"epoch": 0.24472573839662448,
|
| 6120 |
+
"grad_norm": 0.4891457259654999,
|
| 6121 |
+
"learning_rate": 6.653025658450556e-05,
|
| 6122 |
+
"loss": 0.8016,
|
| 6123 |
+
"step": 870
|
| 6124 |
+
},
|
| 6125 |
+
{
|
| 6126 |
+
"epoch": 0.2450070323488045,
|
| 6127 |
+
"grad_norm": 0.4216287136077881,
|
| 6128 |
+
"learning_rate": 6.632082756903269e-05,
|
| 6129 |
+
"loss": 0.648,
|
| 6130 |
+
"step": 871
|
| 6131 |
+
},
|
| 6132 |
+
{
|
| 6133 |
+
"epoch": 0.24528832630098452,
|
| 6134 |
+
"grad_norm": 0.45034247636795044,
|
| 6135 |
+
"learning_rate": 6.611156503915158e-05,
|
| 6136 |
+
"loss": 0.7808,
|
| 6137 |
+
"step": 872
|
| 6138 |
+
},
|
| 6139 |
+
{
|
| 6140 |
+
"epoch": 0.24556962025316456,
|
| 6141 |
+
"grad_norm": 0.4352450966835022,
|
| 6142 |
+
"learning_rate": 6.590247002930567e-05,
|
| 6143 |
+
"loss": 0.6218,
|
| 6144 |
+
"step": 873
|
| 6145 |
+
},
|
| 6146 |
+
{
|
| 6147 |
+
"epoch": 0.24585091420534458,
|
| 6148 |
+
"grad_norm": 0.5135374665260315,
|
| 6149 |
+
"learning_rate": 6.569354357311014e-05,
|
| 6150 |
+
"loss": 0.8808,
|
| 6151 |
+
"step": 874
|
| 6152 |
+
},
|
| 6153 |
+
{
|
| 6154 |
+
"epoch": 0.24613220815752462,
|
| 6155 |
+
"grad_norm": 0.4836218059062958,
|
| 6156 |
+
"learning_rate": 6.548478670334705e-05,
|
| 6157 |
+
"loss": 0.7847,
|
| 6158 |
+
"step": 875
|
| 6159 |
+
},
|
| 6160 |
+
{
|
| 6161 |
+
"epoch": 0.24641350210970464,
|
| 6162 |
+
"grad_norm": 0.43567904829978943,
|
| 6163 |
+
"learning_rate": 6.52762004519602e-05,
|
| 6164 |
+
"loss": 0.5548,
|
| 6165 |
+
"step": 876
|
| 6166 |
+
},
|
| 6167 |
+
{
|
| 6168 |
+
"epoch": 0.24669479606188466,
|
| 6169 |
+
"grad_norm": 0.36044174432754517,
|
| 6170 |
+
"learning_rate": 6.506778585004982e-05,
|
| 6171 |
+
"loss": 0.4302,
|
| 6172 |
+
"step": 877
|
| 6173 |
+
},
|
| 6174 |
+
{
|
| 6175 |
+
"epoch": 0.2469760900140647,
|
| 6176 |
+
"grad_norm": 0.41312167048454285,
|
| 6177 |
+
"learning_rate": 6.485954392786775e-05,
|
| 6178 |
+
"loss": 0.625,
|
| 6179 |
+
"step": 878
|
| 6180 |
+
},
|
| 6181 |
+
{
|
| 6182 |
+
"epoch": 0.24725738396624472,
|
| 6183 |
+
"grad_norm": 0.49642160534858704,
|
| 6184 |
+
"learning_rate": 6.465147571481223e-05,
|
| 6185 |
+
"loss": 0.5316,
|
| 6186 |
+
"step": 879
|
| 6187 |
+
},
|
| 6188 |
+
{
|
| 6189 |
+
"epoch": 0.24753867791842477,
|
| 6190 |
+
"grad_norm": 0.42281633615493774,
|
| 6191 |
+
"learning_rate": 6.444358223942274e-05,
|
| 6192 |
+
"loss": 0.5781,
|
| 6193 |
+
"step": 880
|
| 6194 |
+
},
|
| 6195 |
+
{
|
| 6196 |
+
"epoch": 0.24781997187060478,
|
| 6197 |
+
"grad_norm": 0.35468313097953796,
|
| 6198 |
+
"learning_rate": 6.4235864529375e-05,
|
| 6199 |
+
"loss": 0.3558,
|
| 6200 |
+
"step": 881
|
| 6201 |
+
},
|
| 6202 |
+
{
|
| 6203 |
+
"epoch": 0.2481012658227848,
|
| 6204 |
+
"grad_norm": 0.4053089916706085,
|
| 6205 |
+
"learning_rate": 6.402832361147595e-05,
|
| 6206 |
+
"loss": 0.6468,
|
| 6207 |
+
"step": 882
|
| 6208 |
+
},
|
| 6209 |
+
{
|
| 6210 |
+
"epoch": 0.24838255977496485,
|
| 6211 |
+
"grad_norm": 0.4576174318790436,
|
| 6212 |
+
"learning_rate": 6.382096051165847e-05,
|
| 6213 |
+
"loss": 0.545,
|
| 6214 |
+
"step": 883
|
| 6215 |
+
},
|
| 6216 |
+
{
|
| 6217 |
+
"epoch": 0.24866385372714486,
|
| 6218 |
+
"grad_norm": 0.3779012858867645,
|
| 6219 |
+
"learning_rate": 6.36137762549765e-05,
|
| 6220 |
+
"loss": 0.5249,
|
| 6221 |
+
"step": 884
|
| 6222 |
+
},
|
| 6223 |
+
{
|
| 6224 |
+
"epoch": 0.2489451476793249,
|
| 6225 |
+
"grad_norm": 0.4555618166923523,
|
| 6226 |
+
"learning_rate": 6.340677186559998e-05,
|
| 6227 |
+
"loss": 0.6109,
|
| 6228 |
+
"step": 885
|
| 6229 |
+
},
|
| 6230 |
+
{
|
| 6231 |
+
"epoch": 0.24922644163150492,
|
| 6232 |
+
"grad_norm": 0.44699516892433167,
|
| 6233 |
+
"learning_rate": 6.319994836680957e-05,
|
| 6234 |
+
"loss": 0.6436,
|
| 6235 |
+
"step": 886
|
| 6236 |
+
},
|
| 6237 |
+
{
|
| 6238 |
+
"epoch": 0.24950773558368494,
|
| 6239 |
+
"grad_norm": 0.4490407705307007,
|
| 6240 |
+
"learning_rate": 6.299330678099183e-05,
|
| 6241 |
+
"loss": 0.6826,
|
| 6242 |
+
"step": 887
|
| 6243 |
+
},
|
| 6244 |
+
{
|
| 6245 |
+
"epoch": 0.249789029535865,
|
| 6246 |
+
"grad_norm": 0.4012693166732788,
|
| 6247 |
+
"learning_rate": 6.278684812963407e-05,
|
| 6248 |
+
"loss": 0.435,
|
| 6249 |
+
"step": 888
|
| 6250 |
+
},
|
| 6251 |
+
{
|
| 6252 |
+
"epoch": 0.25007032348804503,
|
| 6253 |
+
"grad_norm": 0.42422980070114136,
|
| 6254 |
+
"learning_rate": 6.258057343331926e-05,
|
| 6255 |
+
"loss": 0.5136,
|
| 6256 |
+
"step": 889
|
| 6257 |
+
},
|
| 6258 |
+
{
|
| 6259 |
+
"epoch": 0.25035161744022505,
|
| 6260 |
+
"grad_norm": 0.4907642900943756,
|
| 6261 |
+
"learning_rate": 6.237448371172108e-05,
|
| 6262 |
+
"loss": 0.8214,
|
| 6263 |
+
"step": 890
|
| 6264 |
+
},
|
| 6265 |
+
{
|
| 6266 |
+
"epoch": 0.25063291139240507,
|
| 6267 |
+
"grad_norm": 0.3971903622150421,
|
| 6268 |
+
"learning_rate": 6.216857998359879e-05,
|
| 6269 |
+
"loss": 0.4636,
|
| 6270 |
+
"step": 891
|
| 6271 |
+
},
|
| 6272 |
+
{
|
| 6273 |
+
"epoch": 0.2509142053445851,
|
| 6274 |
+
"grad_norm": 0.4904184341430664,
|
| 6275 |
+
"learning_rate": 6.19628632667923e-05,
|
| 6276 |
+
"loss": 0.6946,
|
| 6277 |
+
"step": 892
|
| 6278 |
+
},
|
| 6279 |
+
{
|
| 6280 |
+
"epoch": 0.2511954992967651,
|
| 6281 |
+
"grad_norm": 0.41677847504615784,
|
| 6282 |
+
"learning_rate": 6.175733457821691e-05,
|
| 6283 |
+
"loss": 0.5247,
|
| 6284 |
+
"step": 893
|
| 6285 |
+
},
|
| 6286 |
+
{
|
| 6287 |
+
"epoch": 0.2514767932489452,
|
| 6288 |
+
"grad_norm": 0.42049768567085266,
|
| 6289 |
+
"learning_rate": 6.155199493385864e-05,
|
| 6290 |
+
"loss": 0.7187,
|
| 6291 |
+
"step": 894
|
| 6292 |
+
},
|
| 6293 |
+
{
|
| 6294 |
+
"epoch": 0.2517580872011252,
|
| 6295 |
+
"grad_norm": 0.417804092168808,
|
| 6296 |
+
"learning_rate": 6.134684534876891e-05,
|
| 6297 |
+
"loss": 0.7178,
|
| 6298 |
+
"step": 895
|
| 6299 |
+
},
|
| 6300 |
+
{
|
| 6301 |
+
"epoch": 0.2520393811533052,
|
| 6302 |
+
"grad_norm": 0.4474897086620331,
|
| 6303 |
+
"learning_rate": 6.114188683705959e-05,
|
| 6304 |
+
"loss": 0.5688,
|
| 6305 |
+
"step": 896
|
| 6306 |
+
},
|
| 6307 |
+
{
|
| 6308 |
+
"epoch": 0.2523206751054852,
|
| 6309 |
+
"grad_norm": 0.45201632380485535,
|
| 6310 |
+
"learning_rate": 6.09371204118981e-05,
|
| 6311 |
+
"loss": 0.9288,
|
| 6312 |
+
"step": 897
|
| 6313 |
+
},
|
| 6314 |
+
{
|
| 6315 |
+
"epoch": 0.25260196905766524,
|
| 6316 |
+
"grad_norm": 0.41611170768737793,
|
| 6317 |
+
"learning_rate": 6.0732547085502315e-05,
|
| 6318 |
+
"loss": 0.4325,
|
| 6319 |
+
"step": 898
|
| 6320 |
+
},
|
| 6321 |
+
{
|
| 6322 |
+
"epoch": 0.2528832630098453,
|
| 6323 |
+
"grad_norm": 0.4394207000732422,
|
| 6324 |
+
"learning_rate": 6.052816786913546e-05,
|
| 6325 |
+
"loss": 0.592,
|
| 6326 |
+
"step": 899
|
| 6327 |
+
},
|
| 6328 |
+
{
|
| 6329 |
+
"epoch": 0.25316455696202533,
|
| 6330 |
+
"grad_norm": 0.43768590688705444,
|
| 6331 |
+
"learning_rate": 6.032398377310139e-05,
|
| 6332 |
+
"loss": 0.7116,
|
| 6333 |
+
"step": 900
|
| 6334 |
+
},
|
| 6335 |
+
{
|
| 6336 |
+
"epoch": 0.25344585091420535,
|
| 6337 |
+
"grad_norm": 0.4168648421764374,
|
| 6338 |
+
"learning_rate": 6.011999580673931e-05,
|
| 6339 |
+
"loss": 0.4733,
|
| 6340 |
+
"step": 901
|
| 6341 |
+
},
|
| 6342 |
+
{
|
| 6343 |
+
"epoch": 0.25372714486638537,
|
| 6344 |
+
"grad_norm": 0.41316843032836914,
|
| 6345 |
+
"learning_rate": 5.991620497841889e-05,
|
| 6346 |
+
"loss": 0.6799,
|
| 6347 |
+
"step": 902
|
| 6348 |
+
},
|
| 6349 |
+
{
|
| 6350 |
+
"epoch": 0.2540084388185654,
|
| 6351 |
+
"grad_norm": 0.5134892463684082,
|
| 6352 |
+
"learning_rate": 5.971261229553537e-05,
|
| 6353 |
+
"loss": 0.8112,
|
| 6354 |
+
"step": 903
|
| 6355 |
+
},
|
| 6356 |
+
{
|
| 6357 |
+
"epoch": 0.2542897327707454,
|
| 6358 |
+
"grad_norm": 0.3834094703197479,
|
| 6359 |
+
"learning_rate": 5.950921876450449e-05,
|
| 6360 |
+
"loss": 0.4241,
|
| 6361 |
+
"step": 904
|
| 6362 |
+
},
|
| 6363 |
+
{
|
| 6364 |
+
"epoch": 0.2545710267229255,
|
| 6365 |
+
"grad_norm": 0.37081438302993774,
|
| 6366 |
+
"learning_rate": 5.930602539075741e-05,
|
| 6367 |
+
"loss": 0.5255,
|
| 6368 |
+
"step": 905
|
| 6369 |
+
},
|
| 6370 |
+
{
|
| 6371 |
+
"epoch": 0.2548523206751055,
|
| 6372 |
+
"grad_norm": 0.39598196744918823,
|
| 6373 |
+
"learning_rate": 5.9103033178736024e-05,
|
| 6374 |
+
"loss": 0.4012,
|
| 6375 |
+
"step": 906
|
| 6376 |
+
},
|
| 6377 |
+
{
|
| 6378 |
+
"epoch": 0.2551336146272855,
|
| 6379 |
+
"grad_norm": 0.40418726205825806,
|
| 6380 |
+
"learning_rate": 5.890024313188773e-05,
|
| 6381 |
+
"loss": 0.3938,
|
| 6382 |
+
"step": 907
|
| 6383 |
+
},
|
| 6384 |
+
{
|
| 6385 |
+
"epoch": 0.2554149085794655,
|
| 6386 |
+
"grad_norm": 0.4629944860935211,
|
| 6387 |
+
"learning_rate": 5.869765625266055e-05,
|
| 6388 |
+
"loss": 0.6841,
|
| 6389 |
+
"step": 908
|
| 6390 |
+
},
|
| 6391 |
+
{
|
| 6392 |
+
"epoch": 0.25569620253164554,
|
| 6393 |
+
"grad_norm": 0.37959957122802734,
|
| 6394 |
+
"learning_rate": 5.8495273542498265e-05,
|
| 6395 |
+
"loss": 0.4673,
|
| 6396 |
+
"step": 909
|
| 6397 |
+
},
|
| 6398 |
+
{
|
| 6399 |
+
"epoch": 0.2559774964838256,
|
| 6400 |
+
"grad_norm": 0.417077898979187,
|
| 6401 |
+
"learning_rate": 5.829309600183536e-05,
|
| 6402 |
+
"loss": 0.5647,
|
| 6403 |
+
"step": 910
|
| 6404 |
+
},
|
| 6405 |
+
{
|
| 6406 |
+
"epoch": 0.25625879043600563,
|
| 6407 |
+
"grad_norm": 0.4040203392505646,
|
| 6408 |
+
"learning_rate": 5.809112463009203e-05,
|
| 6409 |
+
"loss": 0.6204,
|
| 6410 |
+
"step": 911
|
| 6411 |
+
},
|
| 6412 |
+
{
|
| 6413 |
+
"epoch": 0.25654008438818565,
|
| 6414 |
+
"grad_norm": 0.5162276029586792,
|
| 6415 |
+
"learning_rate": 5.78893604256695e-05,
|
| 6416 |
+
"loss": 0.9063,
|
| 6417 |
+
"step": 912
|
| 6418 |
+
},
|
| 6419 |
+
{
|
| 6420 |
+
"epoch": 0.25682137834036567,
|
| 6421 |
+
"grad_norm": 0.40955114364624023,
|
| 6422 |
+
"learning_rate": 5.7687804385944685e-05,
|
| 6423 |
+
"loss": 0.5338,
|
| 6424 |
+
"step": 913
|
| 6425 |
+
},
|
| 6426 |
+
{
|
| 6427 |
+
"epoch": 0.2571026722925457,
|
| 6428 |
+
"grad_norm": 0.426906943321228,
|
| 6429 |
+
"learning_rate": 5.7486457507265665e-05,
|
| 6430 |
+
"loss": 0.4966,
|
| 6431 |
+
"step": 914
|
| 6432 |
+
},
|
| 6433 |
+
{
|
| 6434 |
+
"epoch": 0.25738396624472576,
|
| 6435 |
+
"grad_norm": 0.5088917016983032,
|
| 6436 |
+
"learning_rate": 5.728532078494655e-05,
|
| 6437 |
+
"loss": 0.7845,
|
| 6438 |
+
"step": 915
|
| 6439 |
+
},
|
| 6440 |
+
{
|
| 6441 |
+
"epoch": 0.2576652601969058,
|
| 6442 |
+
"grad_norm": 0.39224401116371155,
|
| 6443 |
+
"learning_rate": 5.70843952132625e-05,
|
| 6444 |
+
"loss": 0.5273,
|
| 6445 |
+
"step": 916
|
| 6446 |
+
},
|
| 6447 |
+
{
|
| 6448 |
+
"epoch": 0.2579465541490858,
|
| 6449 |
+
"grad_norm": 0.48903387784957886,
|
| 6450 |
+
"learning_rate": 5.68836817854449e-05,
|
| 6451 |
+
"loss": 1.065,
|
| 6452 |
+
"step": 917
|
| 6453 |
+
},
|
| 6454 |
+
{
|
| 6455 |
+
"epoch": 0.2582278481012658,
|
| 6456 |
+
"grad_norm": 0.40586772561073303,
|
| 6457 |
+
"learning_rate": 5.668318149367659e-05,
|
| 6458 |
+
"loss": 0.4201,
|
| 6459 |
+
"step": 918
|
| 6460 |
+
},
|
| 6461 |
+
{
|
| 6462 |
+
"epoch": 0.25850914205344583,
|
| 6463 |
+
"grad_norm": 0.40659916400909424,
|
| 6464 |
+
"learning_rate": 5.648289532908666e-05,
|
| 6465 |
+
"loss": 0.5954,
|
| 6466 |
+
"step": 919
|
| 6467 |
+
},
|
| 6468 |
+
{
|
| 6469 |
+
"epoch": 0.2587904360056259,
|
| 6470 |
+
"grad_norm": 0.5081503987312317,
|
| 6471 |
+
"learning_rate": 5.6282824281745786e-05,
|
| 6472 |
+
"loss": 0.8198,
|
| 6473 |
+
"step": 920
|
| 6474 |
+
},
|
| 6475 |
+
{
|
| 6476 |
+
"epoch": 0.2590717299578059,
|
| 6477 |
+
"grad_norm": 0.4691346287727356,
|
| 6478 |
+
"learning_rate": 5.60829693406613e-05,
|
| 6479 |
+
"loss": 0.6544,
|
| 6480 |
+
"step": 921
|
| 6481 |
+
},
|
| 6482 |
+
{
|
| 6483 |
+
"epoch": 0.25935302390998594,
|
| 6484 |
+
"grad_norm": 0.44746923446655273,
|
| 6485 |
+
"learning_rate": 5.588333149377217e-05,
|
| 6486 |
+
"loss": 0.8579,
|
| 6487 |
+
"step": 922
|
| 6488 |
+
},
|
| 6489 |
+
{
|
| 6490 |
+
"epoch": 0.25963431786216595,
|
| 6491 |
+
"grad_norm": 0.4173818826675415,
|
| 6492 |
+
"learning_rate": 5.56839117279442e-05,
|
| 6493 |
+
"loss": 0.6585,
|
| 6494 |
+
"step": 923
|
| 6495 |
+
},
|
| 6496 |
+
{
|
| 6497 |
+
"epoch": 0.25991561181434597,
|
| 6498 |
+
"grad_norm": 0.43489521741867065,
|
| 6499 |
+
"learning_rate": 5.5484711028965306e-05,
|
| 6500 |
+
"loss": 0.5801,
|
| 6501 |
+
"step": 924
|
| 6502 |
+
},
|
| 6503 |
+
{
|
| 6504 |
+
"epoch": 0.26019690576652604,
|
| 6505 |
+
"grad_norm": 0.41833072900772095,
|
| 6506 |
+
"learning_rate": 5.528573038154028e-05,
|
| 6507 |
+
"loss": 0.5153,
|
| 6508 |
+
"step": 925
|
| 6509 |
+
},
|
| 6510 |
+
{
|
| 6511 |
+
"epoch": 0.26047819971870606,
|
| 6512 |
+
"grad_norm": 0.43373873829841614,
|
| 6513 |
+
"learning_rate": 5.5086970769286294e-05,
|
| 6514 |
+
"loss": 0.5568,
|
| 6515 |
+
"step": 926
|
| 6516 |
+
},
|
| 6517 |
+
{
|
| 6518 |
+
"epoch": 0.2607594936708861,
|
| 6519 |
+
"grad_norm": 0.36584585905075073,
|
| 6520 |
+
"learning_rate": 5.488843317472789e-05,
|
| 6521 |
+
"loss": 0.5018,
|
| 6522 |
+
"step": 927
|
| 6523 |
+
},
|
| 6524 |
+
{
|
| 6525 |
+
"epoch": 0.2610407876230661,
|
| 6526 |
+
"grad_norm": 0.46128782629966736,
|
| 6527 |
+
"learning_rate": 5.4690118579292015e-05,
|
| 6528 |
+
"loss": 0.8399,
|
| 6529 |
+
"step": 928
|
| 6530 |
+
},
|
| 6531 |
+
{
|
| 6532 |
+
"epoch": 0.2613220815752461,
|
| 6533 |
+
"grad_norm": 0.43327847123146057,
|
| 6534 |
+
"learning_rate": 5.449202796330328e-05,
|
| 6535 |
+
"loss": 0.5526,
|
| 6536 |
+
"step": 929
|
| 6537 |
+
},
|
| 6538 |
+
{
|
| 6539 |
+
"epoch": 0.2616033755274262,
|
| 6540 |
+
"grad_norm": 0.43038710951805115,
|
| 6541 |
+
"learning_rate": 5.42941623059792e-05,
|
| 6542 |
+
"loss": 0.7568,
|
| 6543 |
+
"step": 930
|
| 6544 |
+
},
|
| 6545 |
+
{
|
| 6546 |
+
"epoch": 0.2618846694796062,
|
| 6547 |
+
"grad_norm": 0.4167362153530121,
|
| 6548 |
+
"learning_rate": 5.4096522585425134e-05,
|
| 6549 |
+
"loss": 0.6142,
|
| 6550 |
+
"step": 931
|
| 6551 |
+
},
|
| 6552 |
+
{
|
| 6553 |
+
"epoch": 0.2621659634317862,
|
| 6554 |
+
"grad_norm": 0.43088462948799133,
|
| 6555 |
+
"learning_rate": 5.389910977862966e-05,
|
| 6556 |
+
"loss": 0.694,
|
| 6557 |
+
"step": 932
|
| 6558 |
+
},
|
| 6559 |
+
{
|
| 6560 |
+
"epoch": 0.26244725738396624,
|
| 6561 |
+
"grad_norm": 0.422753244638443,
|
| 6562 |
+
"learning_rate": 5.370192486145969e-05,
|
| 6563 |
+
"loss": 0.4352,
|
| 6564 |
+
"step": 933
|
| 6565 |
+
},
|
| 6566 |
+
{
|
| 6567 |
+
"epoch": 0.26272855133614625,
|
| 6568 |
+
"grad_norm": 0.3619067668914795,
|
| 6569 |
+
"learning_rate": 5.350496880865547e-05,
|
| 6570 |
+
"loss": 0.4662,
|
| 6571 |
+
"step": 934
|
| 6572 |
+
},
|
| 6573 |
+
{
|
| 6574 |
+
"epoch": 0.2630098452883263,
|
| 6575 |
+
"grad_norm": 0.38884440064430237,
|
| 6576 |
+
"learning_rate": 5.330824259382601e-05,
|
| 6577 |
+
"loss": 0.5199,
|
| 6578 |
+
"step": 935
|
| 6579 |
+
},
|
| 6580 |
+
{
|
| 6581 |
+
"epoch": 0.26329113924050634,
|
| 6582 |
+
"grad_norm": 0.3974436819553375,
|
| 6583 |
+
"learning_rate": 5.3111747189444174e-05,
|
| 6584 |
+
"loss": 0.6657,
|
| 6585 |
+
"step": 936
|
| 6586 |
+
},
|
| 6587 |
+
{
|
| 6588 |
+
"epoch": 0.26357243319268636,
|
| 6589 |
+
"grad_norm": 0.39121660590171814,
|
| 6590 |
+
"learning_rate": 5.291548356684177e-05,
|
| 6591 |
+
"loss": 0.451,
|
| 6592 |
+
"step": 937
|
| 6593 |
+
},
|
| 6594 |
+
{
|
| 6595 |
+
"epoch": 0.2638537271448664,
|
| 6596 |
+
"grad_norm": 0.43525072932243347,
|
| 6597 |
+
"learning_rate": 5.271945269620495e-05,
|
| 6598 |
+
"loss": 0.8528,
|
| 6599 |
+
"step": 938
|
| 6600 |
+
},
|
| 6601 |
+
{
|
| 6602 |
+
"epoch": 0.2641350210970464,
|
| 6603 |
+
"grad_norm": 0.43280383944511414,
|
| 6604 |
+
"learning_rate": 5.252365554656933e-05,
|
| 6605 |
+
"loss": 0.7017,
|
| 6606 |
+
"step": 939
|
| 6607 |
+
},
|
| 6608 |
+
{
|
| 6609 |
+
"epoch": 0.26441631504922647,
|
| 6610 |
+
"grad_norm": 0.41669878363609314,
|
| 6611 |
+
"learning_rate": 5.2328093085815034e-05,
|
| 6612 |
+
"loss": 0.5281,
|
| 6613 |
+
"step": 940
|
| 6614 |
+
},
|
| 6615 |
+
{
|
| 6616 |
+
"epoch": 0.2646976090014065,
|
| 6617 |
+
"grad_norm": 0.4898998737335205,
|
| 6618 |
+
"learning_rate": 5.213276628066213e-05,
|
| 6619 |
+
"loss": 0.6522,
|
| 6620 |
+
"step": 941
|
| 6621 |
+
},
|
| 6622 |
+
{
|
| 6623 |
+
"epoch": 0.2649789029535865,
|
| 6624 |
+
"grad_norm": 0.44008949398994446,
|
| 6625 |
+
"learning_rate": 5.193767609666584e-05,
|
| 6626 |
+
"loss": 0.5851,
|
| 6627 |
+
"step": 942
|
| 6628 |
+
},
|
| 6629 |
+
{
|
| 6630 |
+
"epoch": 0.2652601969057665,
|
| 6631 |
+
"grad_norm": 0.46987125277519226,
|
| 6632 |
+
"learning_rate": 5.174282349821158e-05,
|
| 6633 |
+
"loss": 0.7581,
|
| 6634 |
+
"step": 943
|
| 6635 |
+
},
|
| 6636 |
+
{
|
| 6637 |
+
"epoch": 0.26554149085794654,
|
| 6638 |
+
"grad_norm": 0.3754156529903412,
|
| 6639 |
+
"learning_rate": 5.1548209448510376e-05,
|
| 6640 |
+
"loss": 0.4245,
|
| 6641 |
+
"step": 944
|
| 6642 |
+
},
|
| 6643 |
+
{
|
| 6644 |
+
"epoch": 0.26582278481012656,
|
| 6645 |
+
"grad_norm": 0.36868390440940857,
|
| 6646 |
+
"learning_rate": 5.1353834909594126e-05,
|
| 6647 |
+
"loss": 0.3979,
|
| 6648 |
+
"step": 945
|
| 6649 |
+
},
|
| 6650 |
+
{
|
| 6651 |
+
"epoch": 0.26610407876230663,
|
| 6652 |
+
"grad_norm": 0.44510623812675476,
|
| 6653 |
+
"learning_rate": 5.115970084231059e-05,
|
| 6654 |
+
"loss": 0.7542,
|
| 6655 |
+
"step": 946
|
| 6656 |
+
},
|
| 6657 |
+
{
|
| 6658 |
+
"epoch": 0.26638537271448665,
|
| 6659 |
+
"grad_norm": 0.4884914755821228,
|
| 6660 |
+
"learning_rate": 5.096580820631888e-05,
|
| 6661 |
+
"loss": 0.5864,
|
| 6662 |
+
"step": 947
|
| 6663 |
+
},
|
| 6664 |
+
{
|
| 6665 |
+
"epoch": 0.26666666666666666,
|
| 6666 |
+
"grad_norm": 0.510437548160553,
|
| 6667 |
+
"learning_rate": 5.077215796008472e-05,
|
| 6668 |
+
"loss": 0.7951,
|
| 6669 |
+
"step": 948
|
| 6670 |
+
},
|
| 6671 |
+
{
|
| 6672 |
+
"epoch": 0.2669479606188467,
|
| 6673 |
+
"grad_norm": 0.42433348298072815,
|
| 6674 |
+
"learning_rate": 5.0578751060875496e-05,
|
| 6675 |
+
"loss": 0.6669,
|
| 6676 |
+
"step": 949
|
| 6677 |
+
},
|
| 6678 |
+
{
|
| 6679 |
+
"epoch": 0.2672292545710267,
|
| 6680 |
+
"grad_norm": 0.4372902810573578,
|
| 6681 |
+
"learning_rate": 5.0385588464755785e-05,
|
| 6682 |
+
"loss": 0.6293,
|
| 6683 |
+
"step": 950
|
| 6684 |
+
},
|
| 6685 |
+
{
|
| 6686 |
+
"epoch": 0.26751054852320677,
|
| 6687 |
+
"grad_norm": 0.5091153979301453,
|
| 6688 |
+
"learning_rate": 5.0192671126582466e-05,
|
| 6689 |
+
"loss": 0.6896,
|
| 6690 |
+
"step": 951
|
| 6691 |
+
},
|
| 6692 |
+
{
|
| 6693 |
+
"epoch": 0.2677918424753868,
|
| 6694 |
+
"grad_norm": 0.4602817893028259,
|
| 6695 |
+
"learning_rate": 5.000000000000002e-05,
|
| 6696 |
+
"loss": 0.5255,
|
| 6697 |
+
"step": 952
|
| 6698 |
+
},
|
| 6699 |
+
{
|
| 6700 |
+
"epoch": 0.2680731364275668,
|
| 6701 |
+
"grad_norm": 0.40581005811691284,
|
| 6702 |
+
"learning_rate": 4.98075760374358e-05,
|
| 6703 |
+
"loss": 0.4145,
|
| 6704 |
+
"step": 953
|
| 6705 |
+
},
|
| 6706 |
+
{
|
| 6707 |
+
"epoch": 0.2683544303797468,
|
| 6708 |
+
"grad_norm": 0.42460790276527405,
|
| 6709 |
+
"learning_rate": 4.9615400190095463e-05,
|
| 6710 |
+
"loss": 0.6349,
|
| 6711 |
+
"step": 954
|
| 6712 |
+
},
|
| 6713 |
+
{
|
| 6714 |
+
"epoch": 0.26863572433192684,
|
| 6715 |
+
"grad_norm": 0.44513189792633057,
|
| 6716 |
+
"learning_rate": 4.9423473407958035e-05,
|
| 6717 |
+
"loss": 0.4941,
|
| 6718 |
+
"step": 955
|
| 6719 |
+
},
|
| 6720 |
+
{
|
| 6721 |
+
"epoch": 0.2689170182841069,
|
| 6722 |
+
"grad_norm": 0.41239869594573975,
|
| 6723 |
+
"learning_rate": 4.923179663977145e-05,
|
| 6724 |
+
"loss": 0.7015,
|
| 6725 |
+
"step": 956
|
| 6726 |
+
},
|
| 6727 |
+
{
|
| 6728 |
+
"epoch": 0.26919831223628693,
|
| 6729 |
+
"grad_norm": 0.4425848424434662,
|
| 6730 |
+
"learning_rate": 4.9040370833047746e-05,
|
| 6731 |
+
"loss": 0.7656,
|
| 6732 |
+
"step": 957
|
| 6733 |
+
},
|
| 6734 |
+
{
|
| 6735 |
+
"epoch": 0.26947960618846695,
|
| 6736 |
+
"grad_norm": 0.45858415961265564,
|
| 6737 |
+
"learning_rate": 4.884919693405833e-05,
|
| 6738 |
+
"loss": 0.7071,
|
| 6739 |
+
"step": 958
|
| 6740 |
+
},
|
| 6741 |
+
{
|
| 6742 |
+
"epoch": 0.26976090014064696,
|
| 6743 |
+
"grad_norm": 0.4543355703353882,
|
| 6744 |
+
"learning_rate": 4.865827588782932e-05,
|
| 6745 |
+
"loss": 0.5379,
|
| 6746 |
+
"step": 959
|
| 6747 |
+
},
|
| 6748 |
+
{
|
| 6749 |
+
"epoch": 0.270042194092827,
|
| 6750 |
+
"grad_norm": 0.5047144293785095,
|
| 6751 |
+
"learning_rate": 4.846760863813708e-05,
|
| 6752 |
+
"loss": 0.5043,
|
| 6753 |
+
"step": 960
|
| 6754 |
+
},
|
| 6755 |
+
{
|
| 6756 |
+
"epoch": 0.27032348804500705,
|
| 6757 |
+
"grad_norm": 0.4006601572036743,
|
| 6758 |
+
"learning_rate": 4.827719612750318e-05,
|
| 6759 |
+
"loss": 0.5147,
|
| 6760 |
+
"step": 961
|
| 6761 |
+
},
|
| 6762 |
+
{
|
| 6763 |
+
"epoch": 0.27060478199718707,
|
| 6764 |
+
"grad_norm": 0.4315263032913208,
|
| 6765 |
+
"learning_rate": 4.80870392971901e-05,
|
| 6766 |
+
"loss": 0.5067,
|
| 6767 |
+
"step": 962
|
| 6768 |
+
},
|
| 6769 |
+
{
|
| 6770 |
+
"epoch": 0.2708860759493671,
|
| 6771 |
+
"grad_norm": 0.48700928688049316,
|
| 6772 |
+
"learning_rate": 4.789713908719638e-05,
|
| 6773 |
+
"loss": 0.8079,
|
| 6774 |
+
"step": 963
|
| 6775 |
+
},
|
| 6776 |
+
{
|
| 6777 |
+
"epoch": 0.2711673699015471,
|
| 6778 |
+
"grad_norm": 0.4333600699901581,
|
| 6779 |
+
"learning_rate": 4.7707496436252e-05,
|
| 6780 |
+
"loss": 0.5568,
|
| 6781 |
+
"step": 964
|
| 6782 |
+
},
|
| 6783 |
+
{
|
| 6784 |
+
"epoch": 0.2714486638537271,
|
| 6785 |
+
"grad_norm": 0.41818660497665405,
|
| 6786 |
+
"learning_rate": 4.751811228181369e-05,
|
| 6787 |
+
"loss": 0.3471,
|
| 6788 |
+
"step": 965
|
| 6789 |
+
},
|
| 6790 |
+
{
|
| 6791 |
+
"epoch": 0.2717299578059072,
|
| 6792 |
+
"grad_norm": 0.4029858112335205,
|
| 6793 |
+
"learning_rate": 4.73289875600605e-05,
|
| 6794 |
+
"loss": 0.3995,
|
| 6795 |
+
"step": 966
|
| 6796 |
+
},
|
| 6797 |
+
{
|
| 6798 |
+
"epoch": 0.2720112517580872,
|
| 6799 |
+
"grad_norm": 0.40747877955436707,
|
| 6800 |
+
"learning_rate": 4.7140123205888884e-05,
|
| 6801 |
+
"loss": 0.6,
|
| 6802 |
+
"step": 967
|
| 6803 |
+
},
|
| 6804 |
+
{
|
| 6805 |
+
"epoch": 0.27229254571026723,
|
| 6806 |
+
"grad_norm": 0.39134928584098816,
|
| 6807 |
+
"learning_rate": 4.6951520152908326e-05,
|
| 6808 |
+
"loss": 0.6793,
|
| 6809 |
+
"step": 968
|
| 6810 |
+
},
|
| 6811 |
+
{
|
| 6812 |
+
"epoch": 0.27257383966244725,
|
| 6813 |
+
"grad_norm": 0.4448183476924896,
|
| 6814 |
+
"learning_rate": 4.6763179333436636e-05,
|
| 6815 |
+
"loss": 0.6989,
|
| 6816 |
+
"step": 969
|
| 6817 |
+
},
|
| 6818 |
+
{
|
| 6819 |
+
"epoch": 0.27285513361462727,
|
| 6820 |
+
"grad_norm": 0.44702470302581787,
|
| 6821 |
+
"learning_rate": 4.657510167849525e-05,
|
| 6822 |
+
"loss": 0.695,
|
| 6823 |
+
"step": 970
|
| 6824 |
+
},
|
| 6825 |
+
{
|
| 6826 |
+
"epoch": 0.27313642756680734,
|
| 6827 |
+
"grad_norm": 0.3992598354816437,
|
| 6828 |
+
"learning_rate": 4.638728811780471e-05,
|
| 6829 |
+
"loss": 0.5069,
|
| 6830 |
+
"step": 971
|
| 6831 |
+
},
|
| 6832 |
+
{
|
| 6833 |
+
"epoch": 0.27341772151898736,
|
| 6834 |
+
"grad_norm": 0.4272211492061615,
|
| 6835 |
+
"learning_rate": 4.619973957978018e-05,
|
| 6836 |
+
"loss": 0.574,
|
| 6837 |
+
"step": 972
|
| 6838 |
+
},
|
| 6839 |
+
{
|
| 6840 |
+
"epoch": 0.2736990154711674,
|
| 6841 |
+
"grad_norm": 0.4479410648345947,
|
| 6842 |
+
"learning_rate": 4.601245699152659e-05,
|
| 6843 |
+
"loss": 0.476,
|
| 6844 |
+
"step": 973
|
| 6845 |
+
},
|
| 6846 |
+
{
|
| 6847 |
+
"epoch": 0.2739803094233474,
|
| 6848 |
+
"grad_norm": 0.4400903284549713,
|
| 6849 |
+
"learning_rate": 4.582544127883431e-05,
|
| 6850 |
+
"loss": 0.5728,
|
| 6851 |
+
"step": 974
|
| 6852 |
+
},
|
| 6853 |
+
{
|
| 6854 |
+
"epoch": 0.2742616033755274,
|
| 6855 |
+
"grad_norm": 0.3850638270378113,
|
| 6856 |
+
"learning_rate": 4.56386933661745e-05,
|
| 6857 |
+
"loss": 0.3942,
|
| 6858 |
+
"step": 975
|
| 6859 |
+
},
|
| 6860 |
+
{
|
| 6861 |
+
"epoch": 0.2745428973277075,
|
| 6862 |
+
"grad_norm": 0.3833610713481903,
|
| 6863 |
+
"learning_rate": 4.545221417669438e-05,
|
| 6864 |
+
"loss": 0.541,
|
| 6865 |
+
"step": 976
|
| 6866 |
+
},
|
| 6867 |
+
{
|
| 6868 |
+
"epoch": 0.2748241912798875,
|
| 6869 |
+
"grad_norm": 0.3508284389972687,
|
| 6870 |
+
"learning_rate": 4.5266004632212863e-05,
|
| 6871 |
+
"loss": 0.2878,
|
| 6872 |
+
"step": 977
|
| 6873 |
+
},
|
| 6874 |
+
{
|
| 6875 |
+
"epoch": 0.2751054852320675,
|
| 6876 |
+
"grad_norm": 0.5146034955978394,
|
| 6877 |
+
"learning_rate": 4.5080065653215986e-05,
|
| 6878 |
+
"loss": 0.5732,
|
| 6879 |
+
"step": 978
|
| 6880 |
+
},
|
| 6881 |
+
{
|
| 6882 |
+
"epoch": 0.27538677918424753,
|
| 6883 |
+
"grad_norm": 0.38999906182289124,
|
| 6884 |
+
"learning_rate": 4.489439815885218e-05,
|
| 6885 |
+
"loss": 0.6098,
|
| 6886 |
+
"step": 979
|
| 6887 |
+
},
|
| 6888 |
+
{
|
| 6889 |
+
"epoch": 0.27566807313642755,
|
| 6890 |
+
"grad_norm": 0.40867045521736145,
|
| 6891 |
+
"learning_rate": 4.4709003066927966e-05,
|
| 6892 |
+
"loss": 0.5088,
|
| 6893 |
+
"step": 980
|
| 6894 |
+
},
|
| 6895 |
+
{
|
| 6896 |
+
"epoch": 0.2759493670886076,
|
| 6897 |
+
"grad_norm": 0.48017439246177673,
|
| 6898 |
+
"learning_rate": 4.45238812939033e-05,
|
| 6899 |
+
"loss": 0.7656,
|
| 6900 |
+
"step": 981
|
| 6901 |
+
},
|
| 6902 |
+
{
|
| 6903 |
+
"epoch": 0.27623066104078764,
|
| 6904 |
+
"grad_norm": 0.47304192185401917,
|
| 6905 |
+
"learning_rate": 4.433903375488697e-05,
|
| 6906 |
+
"loss": 0.6297,
|
| 6907 |
+
"step": 982
|
| 6908 |
+
},
|
| 6909 |
+
{
|
| 6910 |
+
"epoch": 0.27651195499296766,
|
| 6911 |
+
"grad_norm": 0.4008045792579651,
|
| 6912 |
+
"learning_rate": 4.415446136363216e-05,
|
| 6913 |
+
"loss": 0.5125,
|
| 6914 |
+
"step": 983
|
| 6915 |
+
},
|
| 6916 |
+
{
|
| 6917 |
+
"epoch": 0.2767932489451477,
|
| 6918 |
+
"grad_norm": 0.43953150510787964,
|
| 6919 |
+
"learning_rate": 4.397016503253199e-05,
|
| 6920 |
+
"loss": 0.7034,
|
| 6921 |
+
"step": 984
|
| 6922 |
+
},
|
| 6923 |
+
{
|
| 6924 |
+
"epoch": 0.2770745428973277,
|
| 6925 |
+
"grad_norm": 0.483635812997818,
|
| 6926 |
+
"learning_rate": 4.3786145672614864e-05,
|
| 6927 |
+
"loss": 0.8854,
|
| 6928 |
+
"step": 985
|
| 6929 |
+
},
|
| 6930 |
+
{
|
| 6931 |
+
"epoch": 0.2773558368495077,
|
| 6932 |
+
"grad_norm": 0.46350064873695374,
|
| 6933 |
+
"learning_rate": 4.360240419354007e-05,
|
| 6934 |
+
"loss": 0.8174,
|
| 6935 |
+
"step": 986
|
| 6936 |
+
},
|
| 6937 |
+
{
|
| 6938 |
+
"epoch": 0.2776371308016878,
|
| 6939 |
+
"grad_norm": 0.41599181294441223,
|
| 6940 |
+
"learning_rate": 4.341894150359331e-05,
|
| 6941 |
+
"loss": 0.6696,
|
| 6942 |
+
"step": 987
|
| 6943 |
+
},
|
| 6944 |
+
{
|
| 6945 |
+
"epoch": 0.2779184247538678,
|
| 6946 |
+
"grad_norm": 0.3937719464302063,
|
| 6947 |
+
"learning_rate": 4.3235758509682054e-05,
|
| 6948 |
+
"loss": 0.6162,
|
| 6949 |
+
"step": 988
|
| 6950 |
+
},
|
| 6951 |
+
{
|
| 6952 |
+
"epoch": 0.2781997187060478,
|
| 6953 |
+
"grad_norm": 0.4417009949684143,
|
| 6954 |
+
"learning_rate": 4.305285611733116e-05,
|
| 6955 |
+
"loss": 0.718,
|
| 6956 |
+
"step": 989
|
| 6957 |
+
},
|
| 6958 |
+
{
|
| 6959 |
+
"epoch": 0.27848101265822783,
|
| 6960 |
+
"grad_norm": 0.4368630051612854,
|
| 6961 |
+
"learning_rate": 4.2870235230678524e-05,
|
| 6962 |
+
"loss": 0.6728,
|
| 6963 |
+
"step": 990
|
| 6964 |
+
},
|
| 6965 |
+
{
|
| 6966 |
+
"epoch": 0.27876230661040785,
|
| 6967 |
+
"grad_norm": 0.4150483012199402,
|
| 6968 |
+
"learning_rate": 4.268789675247029e-05,
|
| 6969 |
+
"loss": 0.6939,
|
| 6970 |
+
"step": 991
|
| 6971 |
+
},
|
| 6972 |
+
{
|
| 6973 |
+
"epoch": 0.2790436005625879,
|
| 6974 |
+
"grad_norm": 0.37016063928604126,
|
| 6975 |
+
"learning_rate": 4.250584158405673e-05,
|
| 6976 |
+
"loss": 0.4639,
|
| 6977 |
+
"step": 992
|
| 6978 |
+
},
|
| 6979 |
+
{
|
| 6980 |
+
"epoch": 0.27932489451476794,
|
| 6981 |
+
"grad_norm": 0.43990686535835266,
|
| 6982 |
+
"learning_rate": 4.2324070625387615e-05,
|
| 6983 |
+
"loss": 0.7246,
|
| 6984 |
+
"step": 993
|
| 6985 |
+
},
|
| 6986 |
+
{
|
| 6987 |
+
"epoch": 0.27960618846694796,
|
| 6988 |
+
"grad_norm": 0.5088633298873901,
|
| 6989 |
+
"learning_rate": 4.214258477500772e-05,
|
| 6990 |
+
"loss": 0.7894,
|
| 6991 |
+
"step": 994
|
| 6992 |
+
},
|
| 6993 |
+
{
|
| 6994 |
+
"epoch": 0.279887482419128,
|
| 6995 |
+
"grad_norm": 0.47769808769226074,
|
| 6996 |
+
"learning_rate": 4.196138493005245e-05,
|
| 6997 |
+
"loss": 0.5777,
|
| 6998 |
+
"step": 995
|
| 6999 |
+
},
|
| 7000 |
+
{
|
| 7001 |
+
"epoch": 0.280168776371308,
|
| 7002 |
+
"grad_norm": 0.3788312077522278,
|
| 7003 |
+
"learning_rate": 4.178047198624351e-05,
|
| 7004 |
+
"loss": 0.4211,
|
| 7005 |
+
"step": 996
|
| 7006 |
+
},
|
| 7007 |
+
{
|
| 7008 |
+
"epoch": 0.28045007032348807,
|
| 7009 |
+
"grad_norm": 0.4508630335330963,
|
| 7010 |
+
"learning_rate": 4.159984683788424e-05,
|
| 7011 |
+
"loss": 0.5384,
|
| 7012 |
+
"step": 997
|
| 7013 |
+
},
|
| 7014 |
+
{
|
| 7015 |
+
"epoch": 0.2807313642756681,
|
| 7016 |
+
"grad_norm": 0.37215301394462585,
|
| 7017 |
+
"learning_rate": 4.141951037785541e-05,
|
| 7018 |
+
"loss": 0.4564,
|
| 7019 |
+
"step": 998
|
| 7020 |
+
},
|
| 7021 |
+
{
|
| 7022 |
+
"epoch": 0.2810126582278481,
|
| 7023 |
+
"grad_norm": 0.4205975830554962,
|
| 7024 |
+
"learning_rate": 4.123946349761075e-05,
|
| 7025 |
+
"loss": 0.5412,
|
| 7026 |
+
"step": 999
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.2812939521800281,
|
| 7030 |
+
"grad_norm": 0.4999726712703705,
|
| 7031 |
+
"learning_rate": 4.105970708717244e-05,
|
| 7032 |
+
"loss": 0.7411,
|
| 7033 |
+
"step": 1000
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.28157524613220813,
|
| 7037 |
+
"grad_norm": 0.4311080873012543,
|
| 7038 |
+
"learning_rate": 4.088024203512676e-05,
|
| 7039 |
+
"loss": 0.7735,
|
| 7040 |
+
"step": 1001
|
| 7041 |
+
},
|
| 7042 |
+
{
|
| 7043 |
+
"epoch": 0.2818565400843882,
|
| 7044 |
+
"grad_norm": 0.45798638463020325,
|
| 7045 |
+
"learning_rate": 4.070106922861985e-05,
|
| 7046 |
+
"loss": 0.9134,
|
| 7047 |
+
"step": 1002
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 0.2821378340365682,
|
| 7051 |
+
"grad_norm": 0.4613318145275116,
|
| 7052 |
+
"learning_rate": 4.052218955335307e-05,
|
| 7053 |
+
"loss": 0.7549,
|
| 7054 |
+
"step": 1003
|
| 7055 |
+
},
|
| 7056 |
+
{
|
| 7057 |
+
"epoch": 0.28241912798874824,
|
| 7058 |
+
"grad_norm": 0.45625245571136475,
|
| 7059 |
+
"learning_rate": 4.034360389357883e-05,
|
| 7060 |
+
"loss": 0.7143,
|
| 7061 |
+
"step": 1004
|
| 7062 |
+
},
|
| 7063 |
+
{
|
| 7064 |
+
"epoch": 0.28270042194092826,
|
| 7065 |
+
"grad_norm": 0.421064555644989,
|
| 7066 |
+
"learning_rate": 4.016531313209616e-05,
|
| 7067 |
+
"loss": 0.6179,
|
| 7068 |
+
"step": 1005
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.2829817158931083,
|
| 7072 |
+
"grad_norm": 0.46807312965393066,
|
| 7073 |
+
"learning_rate": 3.9987318150246235e-05,
|
| 7074 |
+
"loss": 0.9203,
|
| 7075 |
+
"step": 1006
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.28326300984528835,
|
| 7079 |
+
"grad_norm": 0.45732802152633667,
|
| 7080 |
+
"learning_rate": 3.980961982790813e-05,
|
| 7081 |
+
"loss": 0.6026,
|
| 7082 |
+
"step": 1007
|
| 7083 |
+
},
|
| 7084 |
+
{
|
| 7085 |
+
"epoch": 0.28354430379746837,
|
| 7086 |
+
"grad_norm": 0.4454231560230255,
|
| 7087 |
+
"learning_rate": 3.9632219043494534e-05,
|
| 7088 |
+
"loss": 0.5536,
|
| 7089 |
+
"step": 1008
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 0.2838255977496484,
|
| 7093 |
+
"grad_norm": 0.45921197533607483,
|
| 7094 |
+
"learning_rate": 3.945511667394719e-05,
|
| 7095 |
+
"loss": 0.6036,
|
| 7096 |
+
"step": 1009
|
| 7097 |
+
},
|
| 7098 |
+
{
|
| 7099 |
+
"epoch": 0.2841068917018284,
|
| 7100 |
+
"grad_norm": 0.4745734632015228,
|
| 7101 |
+
"learning_rate": 3.927831359473281e-05,
|
| 7102 |
+
"loss": 0.7428,
|
| 7103 |
+
"step": 1010
|
| 7104 |
+
},
|
| 7105 |
+
{
|
| 7106 |
+
"epoch": 0.2843881856540084,
|
| 7107 |
+
"grad_norm": 0.3922124207019806,
|
| 7108 |
+
"learning_rate": 3.9101810679838604e-05,
|
| 7109 |
+
"loss": 0.4882,
|
| 7110 |
+
"step": 1011
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.2846694796061885,
|
| 7114 |
+
"grad_norm": 0.4244883060455322,
|
| 7115 |
+
"learning_rate": 3.892560880176794e-05,
|
| 7116 |
+
"loss": 0.6115,
|
| 7117 |
+
"step": 1012
|
| 7118 |
+
},
|
| 7119 |
+
{
|
| 7120 |
+
"epoch": 0.2849507735583685,
|
| 7121 |
+
"grad_norm": 0.4335481524467468,
|
| 7122 |
+
"learning_rate": 3.874970883153607e-05,
|
| 7123 |
+
"loss": 0.5464,
|
| 7124 |
+
"step": 1013
|
| 7125 |
+
},
|
| 7126 |
+
{
|
| 7127 |
+
"epoch": 0.2852320675105485,
|
| 7128 |
+
"grad_norm": 0.40534138679504395,
|
| 7129 |
+
"learning_rate": 3.8574111638665936e-05,
|
| 7130 |
+
"loss": 0.5034,
|
| 7131 |
+
"step": 1014
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 0.28551336146272854,
|
| 7135 |
+
"grad_norm": 0.4612746238708496,
|
| 7136 |
+
"learning_rate": 3.8398818091183596e-05,
|
| 7137 |
+
"loss": 0.8287,
|
| 7138 |
+
"step": 1015
|
| 7139 |
+
},
|
| 7140 |
+
{
|
| 7141 |
+
"epoch": 0.28579465541490856,
|
| 7142 |
+
"grad_norm": 0.41410136222839355,
|
| 7143 |
+
"learning_rate": 3.822382905561428e-05,
|
| 7144 |
+
"loss": 0.4765,
|
| 7145 |
+
"step": 1016
|
| 7146 |
+
},
|
| 7147 |
+
{
|
| 7148 |
+
"epoch": 0.28607594936708863,
|
| 7149 |
+
"grad_norm": 0.4347825348377228,
|
| 7150 |
+
"learning_rate": 3.8049145396977845e-05,
|
| 7151 |
+
"loss": 0.5109,
|
| 7152 |
+
"step": 1017
|
| 7153 |
+
},
|
| 7154 |
+
{
|
| 7155 |
+
"epoch": 0.28635724331926865,
|
| 7156 |
+
"grad_norm": 0.4359002411365509,
|
| 7157 |
+
"learning_rate": 3.787476797878459e-05,
|
| 7158 |
+
"loss": 0.6915,
|
| 7159 |
+
"step": 1018
|
| 7160 |
+
},
|
| 7161 |
+
{
|
| 7162 |
+
"epoch": 0.28663853727144867,
|
| 7163 |
+
"grad_norm": 0.4387228190898895,
|
| 7164 |
+
"learning_rate": 3.770069766303092e-05,
|
| 7165 |
+
"loss": 0.6991,
|
| 7166 |
+
"step": 1019
|
| 7167 |
+
},
|
| 7168 |
+
{
|
| 7169 |
+
"epoch": 0.2869198312236287,
|
| 7170 |
+
"grad_norm": 0.4619642496109009,
|
| 7171 |
+
"learning_rate": 3.75269353101953e-05,
|
| 7172 |
+
"loss": 0.7992,
|
| 7173 |
+
"step": 1020
|
| 7174 |
+
},
|
| 7175 |
+
{
|
| 7176 |
+
"epoch": 0.2872011251758087,
|
| 7177 |
+
"grad_norm": 0.40766462683677673,
|
| 7178 |
+
"learning_rate": 3.735348177923367e-05,
|
| 7179 |
+
"loss": 0.7675,
|
| 7180 |
+
"step": 1021
|
| 7181 |
+
},
|
| 7182 |
+
{
|
| 7183 |
+
"epoch": 0.2874824191279888,
|
| 7184 |
+
"grad_norm": 0.47310078144073486,
|
| 7185 |
+
"learning_rate": 3.71803379275755e-05,
|
| 7186 |
+
"loss": 0.6271,
|
| 7187 |
+
"step": 1022
|
| 7188 |
+
},
|
| 7189 |
+
{
|
| 7190 |
+
"epoch": 0.2877637130801688,
|
| 7191 |
+
"grad_norm": 0.5060867667198181,
|
| 7192 |
+
"learning_rate": 3.700750461111945e-05,
|
| 7193 |
+
"loss": 0.6412,
|
| 7194 |
+
"step": 1023
|
| 7195 |
+
},
|
| 7196 |
+
{
|
| 7197 |
+
"epoch": 0.2880450070323488,
|
| 7198 |
+
"grad_norm": 0.4210134446620941,
|
| 7199 |
+
"learning_rate": 3.6834982684229e-05,
|
| 7200 |
+
"loss": 0.6164,
|
| 7201 |
+
"step": 1024
|
| 7202 |
+
},
|
| 7203 |
+
{
|
| 7204 |
+
"epoch": 0.2883263009845288,
|
| 7205 |
+
"grad_norm": 0.4009053111076355,
|
| 7206 |
+
"learning_rate": 3.666277299972839e-05,
|
| 7207 |
+
"loss": 0.5072,
|
| 7208 |
+
"step": 1025
|
| 7209 |
+
},
|
| 7210 |
+
{
|
| 7211 |
+
"epoch": 0.28860759493670884,
|
| 7212 |
+
"grad_norm": 0.4145357608795166,
|
| 7213 |
+
"learning_rate": 3.6490876408898414e-05,
|
| 7214 |
+
"loss": 0.6514,
|
| 7215 |
+
"step": 1026
|
| 7216 |
+
},
|
| 7217 |
+
{
|
| 7218 |
+
"epoch": 0.28888888888888886,
|
| 7219 |
+
"grad_norm": 0.45598262548446655,
|
| 7220 |
+
"learning_rate": 3.631929376147207e-05,
|
| 7221 |
+
"loss": 0.6579,
|
| 7222 |
+
"step": 1027
|
| 7223 |
+
},
|
| 7224 |
+
{
|
| 7225 |
+
"epoch": 0.28917018284106893,
|
| 7226 |
+
"grad_norm": 0.40074342489242554,
|
| 7227 |
+
"learning_rate": 3.614802590563051e-05,
|
| 7228 |
+
"loss": 0.5848,
|
| 7229 |
+
"step": 1028
|
| 7230 |
+
},
|
| 7231 |
+
{
|
| 7232 |
+
"epoch": 0.28945147679324895,
|
| 7233 |
+
"grad_norm": 0.46736159920692444,
|
| 7234 |
+
"learning_rate": 3.597707368799881e-05,
|
| 7235 |
+
"loss": 0.6881,
|
| 7236 |
+
"step": 1029
|
| 7237 |
+
},
|
| 7238 |
+
{
|
| 7239 |
+
"epoch": 0.28973277074542897,
|
| 7240 |
+
"grad_norm": 0.42453956604003906,
|
| 7241 |
+
"learning_rate": 3.580643795364166e-05,
|
| 7242 |
+
"loss": 0.5123,
|
| 7243 |
+
"step": 1030
|
| 7244 |
+
},
|
| 7245 |
+
{
|
| 7246 |
+
"epoch": 0.290014064697609,
|
| 7247 |
+
"grad_norm": 0.3824552595615387,
|
| 7248 |
+
"learning_rate": 3.563611954605931e-05,
|
| 7249 |
+
"loss": 0.5062,
|
| 7250 |
+
"step": 1031
|
| 7251 |
+
},
|
| 7252 |
+
{
|
| 7253 |
+
"epoch": 0.290295358649789,
|
| 7254 |
+
"grad_norm": 0.4980485737323761,
|
| 7255 |
+
"learning_rate": 3.5466119307183465e-05,
|
| 7256 |
+
"loss": 0.478,
|
| 7257 |
+
"step": 1032
|
| 7258 |
+
},
|
| 7259 |
+
{
|
| 7260 |
+
"epoch": 0.2905766526019691,
|
| 7261 |
+
"grad_norm": 0.43941548466682434,
|
| 7262 |
+
"learning_rate": 3.529643807737291e-05,
|
| 7263 |
+
"loss": 0.6748,
|
| 7264 |
+
"step": 1033
|
| 7265 |
+
},
|
| 7266 |
+
{
|
| 7267 |
+
"epoch": 0.2908579465541491,
|
| 7268 |
+
"grad_norm": 0.41296613216400146,
|
| 7269 |
+
"learning_rate": 3.512707669540957e-05,
|
| 7270 |
+
"loss": 0.47,
|
| 7271 |
+
"step": 1034
|
| 7272 |
+
},
|
| 7273 |
+
{
|
| 7274 |
+
"epoch": 0.2911392405063291,
|
| 7275 |
+
"grad_norm": 0.4627281725406647,
|
| 7276 |
+
"learning_rate": 3.495803599849425e-05,
|
| 7277 |
+
"loss": 0.4076,
|
| 7278 |
+
"step": 1035
|
| 7279 |
+
},
|
| 7280 |
+
{
|
| 7281 |
+
"epoch": 0.29142053445850913,
|
| 7282 |
+
"grad_norm": 0.4896155893802643,
|
| 7283 |
+
"learning_rate": 3.47893168222425e-05,
|
| 7284 |
+
"loss": 0.6034,
|
| 7285 |
+
"step": 1036
|
| 7286 |
+
},
|
| 7287 |
+
{
|
| 7288 |
+
"epoch": 0.29170182841068915,
|
| 7289 |
+
"grad_norm": 0.4602232277393341,
|
| 7290 |
+
"learning_rate": 3.4620920000680466e-05,
|
| 7291 |
+
"loss": 0.6293,
|
| 7292 |
+
"step": 1037
|
| 7293 |
+
},
|
| 7294 |
+
{
|
| 7295 |
+
"epoch": 0.2919831223628692,
|
| 7296 |
+
"grad_norm": 0.49549683928489685,
|
| 7297 |
+
"learning_rate": 3.445284636624092e-05,
|
| 7298 |
+
"loss": 0.6625,
|
| 7299 |
+
"step": 1038
|
| 7300 |
+
},
|
| 7301 |
+
{
|
| 7302 |
+
"epoch": 0.29226441631504924,
|
| 7303 |
+
"grad_norm": 0.3931140899658203,
|
| 7304 |
+
"learning_rate": 3.4285096749758895e-05,
|
| 7305 |
+
"loss": 0.5008,
|
| 7306 |
+
"step": 1039
|
| 7307 |
+
},
|
| 7308 |
+
{
|
| 7309 |
+
"epoch": 0.29254571026722925,
|
| 7310 |
+
"grad_norm": 0.4757015109062195,
|
| 7311 |
+
"learning_rate": 3.411767198046779e-05,
|
| 7312 |
+
"loss": 0.6755,
|
| 7313 |
+
"step": 1040
|
| 7314 |
+
},
|
| 7315 |
+
{
|
| 7316 |
+
"epoch": 0.29282700421940927,
|
| 7317 |
+
"grad_norm": 0.4022630453109741,
|
| 7318 |
+
"learning_rate": 3.3950572885995246e-05,
|
| 7319 |
+
"loss": 0.4952,
|
| 7320 |
+
"step": 1041
|
| 7321 |
+
},
|
| 7322 |
+
{
|
| 7323 |
+
"epoch": 0.2931082981715893,
|
| 7324 |
+
"grad_norm": 0.41229626536369324,
|
| 7325 |
+
"learning_rate": 3.3783800292358883e-05,
|
| 7326 |
+
"loss": 0.6766,
|
| 7327 |
+
"step": 1042
|
| 7328 |
+
},
|
| 7329 |
+
{
|
| 7330 |
+
"epoch": 0.29338959212376936,
|
| 7331 |
+
"grad_norm": 0.3830214738845825,
|
| 7332 |
+
"learning_rate": 3.361735502396235e-05,
|
| 7333 |
+
"loss": 0.3165,
|
| 7334 |
+
"step": 1043
|
| 7335 |
+
},
|
| 7336 |
+
{
|
| 7337 |
+
"epoch": 0.2936708860759494,
|
| 7338 |
+
"grad_norm": 0.39929255843162537,
|
| 7339 |
+
"learning_rate": 3.345123790359135e-05,
|
| 7340 |
+
"loss": 0.5273,
|
| 7341 |
+
"step": 1044
|
| 7342 |
+
},
|
| 7343 |
+
{
|
| 7344 |
+
"epoch": 0.2939521800281294,
|
| 7345 |
+
"grad_norm": 0.3855002522468567,
|
| 7346 |
+
"learning_rate": 3.328544975240932e-05,
|
| 7347 |
+
"loss": 0.5124,
|
| 7348 |
+
"step": 1045
|
| 7349 |
+
},
|
| 7350 |
+
{
|
| 7351 |
+
"epoch": 0.2942334739803094,
|
| 7352 |
+
"grad_norm": 0.45187920331954956,
|
| 7353 |
+
"learning_rate": 3.3119991389953596e-05,
|
| 7354 |
+
"loss": 0.524,
|
| 7355 |
+
"step": 1046
|
| 7356 |
+
},
|
| 7357 |
+
{
|
| 7358 |
+
"epoch": 0.29451476793248943,
|
| 7359 |
+
"grad_norm": 0.4005303978919983,
|
| 7360 |
+
"learning_rate": 3.2954863634131295e-05,
|
| 7361 |
+
"loss": 0.6308,
|
| 7362 |
+
"step": 1047
|
| 7363 |
+
},
|
| 7364 |
+
{
|
| 7365 |
+
"epoch": 0.2947960618846695,
|
| 7366 |
+
"grad_norm": 0.4425119161605835,
|
| 7367 |
+
"learning_rate": 3.279006730121517e-05,
|
| 7368 |
+
"loss": 0.7658,
|
| 7369 |
+
"step": 1048
|
| 7370 |
+
},
|
| 7371 |
+
{
|
| 7372 |
+
"epoch": 0.2950773558368495,
|
| 7373 |
+
"grad_norm": 0.438798725605011,
|
| 7374 |
+
"learning_rate": 3.262560320583965e-05,
|
| 7375 |
+
"loss": 0.6371,
|
| 7376 |
+
"step": 1049
|
| 7377 |
+
},
|
| 7378 |
+
{
|
| 7379 |
+
"epoch": 0.29535864978902954,
|
| 7380 |
+
"grad_norm": 0.40039610862731934,
|
| 7381 |
+
"learning_rate": 3.2461472160996966e-05,
|
| 7382 |
+
"loss": 0.4249,
|
| 7383 |
+
"step": 1050
|
| 7384 |
+
},
|
| 7385 |
+
{
|
| 7386 |
+
"epoch": 0.29563994374120955,
|
| 7387 |
+
"grad_norm": 0.41733014583587646,
|
| 7388 |
+
"learning_rate": 3.229767497803281e-05,
|
| 7389 |
+
"loss": 0.6441,
|
| 7390 |
+
"step": 1051
|
| 7391 |
+
},
|
| 7392 |
+
{
|
| 7393 |
+
"epoch": 0.29592123769338957,
|
| 7394 |
+
"grad_norm": 0.42075052857398987,
|
| 7395 |
+
"learning_rate": 3.21342124666426e-05,
|
| 7396 |
+
"loss": 0.4485,
|
| 7397 |
+
"step": 1052
|
| 7398 |
+
},
|
| 7399 |
+
{
|
| 7400 |
+
"epoch": 0.29620253164556964,
|
| 7401 |
+
"grad_norm": 0.4350489675998688,
|
| 7402 |
+
"learning_rate": 3.197108543486741e-05,
|
| 7403 |
+
"loss": 0.3589,
|
| 7404 |
+
"step": 1053
|
| 7405 |
+
},
|
| 7406 |
+
{
|
| 7407 |
+
"epoch": 0.29648382559774966,
|
| 7408 |
+
"grad_norm": 0.4629608392715454,
|
| 7409 |
+
"learning_rate": 3.180829468908986e-05,
|
| 7410 |
+
"loss": 0.6207,
|
| 7411 |
+
"step": 1054
|
| 7412 |
+
},
|
| 7413 |
+
{
|
| 7414 |
+
"epoch": 0.2967651195499297,
|
| 7415 |
+
"grad_norm": 0.4030011296272278,
|
| 7416 |
+
"learning_rate": 3.16458410340302e-05,
|
| 7417 |
+
"loss": 0.5335,
|
| 7418 |
+
"step": 1055
|
| 7419 |
+
},
|
| 7420 |
+
{
|
| 7421 |
+
"epoch": 0.2970464135021097,
|
| 7422 |
+
"grad_norm": 0.40957632660865784,
|
| 7423 |
+
"learning_rate": 3.148372527274246e-05,
|
| 7424 |
+
"loss": 0.7041,
|
| 7425 |
+
"step": 1056
|
| 7426 |
+
},
|
| 7427 |
+
{
|
| 7428 |
+
"epoch": 0.2973277074542897,
|
| 7429 |
+
"grad_norm": 0.42649829387664795,
|
| 7430 |
+
"learning_rate": 3.132194820661025e-05,
|
| 7431 |
+
"loss": 0.5509,
|
| 7432 |
+
"step": 1057
|
| 7433 |
+
},
|
| 7434 |
+
{
|
| 7435 |
+
"epoch": 0.2976090014064698,
|
| 7436 |
+
"grad_norm": 0.43894529342651367,
|
| 7437 |
+
"learning_rate": 3.116051063534299e-05,
|
| 7438 |
+
"loss": 0.6796,
|
| 7439 |
+
"step": 1058
|
| 7440 |
+
},
|
| 7441 |
+
{
|
| 7442 |
+
"epoch": 0.2978902953586498,
|
| 7443 |
+
"grad_norm": 0.5489816665649414,
|
| 7444 |
+
"learning_rate": 3.099941335697189e-05,
|
| 7445 |
+
"loss": 0.5954,
|
| 7446 |
+
"step": 1059
|
| 7447 |
+
},
|
| 7448 |
+
{
|
| 7449 |
+
"epoch": 0.2981715893108298,
|
| 7450 |
+
"grad_norm": 0.45282647013664246,
|
| 7451 |
+
"learning_rate": 3.083865716784592e-05,
|
| 7452 |
+
"loss": 0.7011,
|
| 7453 |
+
"step": 1060
|
| 7454 |
+
},
|
| 7455 |
+
{
|
| 7456 |
+
"epoch": 0.29845288326300984,
|
| 7457 |
+
"grad_norm": 0.42794767022132874,
|
| 7458 |
+
"learning_rate": 3.067824286262797e-05,
|
| 7459 |
+
"loss": 0.576,
|
| 7460 |
+
"step": 1061
|
| 7461 |
+
},
|
| 7462 |
+
{
|
| 7463 |
+
"epoch": 0.29873417721518986,
|
| 7464 |
+
"grad_norm": 0.42666977643966675,
|
| 7465 |
+
"learning_rate": 3.051817123429098e-05,
|
| 7466 |
+
"loss": 0.5674,
|
| 7467 |
+
"step": 1062
|
| 7468 |
+
},
|
| 7469 |
+
{
|
| 7470 |
+
"epoch": 0.29901547116736993,
|
| 7471 |
+
"grad_norm": 0.4605925977230072,
|
| 7472 |
+
"learning_rate": 3.035844307411384e-05,
|
| 7473 |
+
"loss": 0.7286,
|
| 7474 |
+
"step": 1063
|
| 7475 |
+
},
|
| 7476 |
+
{
|
| 7477 |
+
"epoch": 0.29929676511954995,
|
| 7478 |
+
"grad_norm": 0.4057950973510742,
|
| 7479 |
+
"learning_rate": 3.0199059171677656e-05,
|
| 7480 |
+
"loss": 0.5115,
|
| 7481 |
+
"step": 1064
|
| 7482 |
+
},
|
| 7483 |
+
{
|
| 7484 |
+
"epoch": 0.29957805907172996,
|
| 7485 |
+
"grad_norm": 0.3734945058822632,
|
| 7486 |
+
"learning_rate": 3.0040020314861772e-05,
|
| 7487 |
+
"loss": 0.4214,
|
| 7488 |
+
"step": 1065
|
| 7489 |
+
},
|
| 7490 |
+
{
|
| 7491 |
+
"epoch": 0.29985935302391,
|
| 7492 |
+
"grad_norm": 0.4027150571346283,
|
| 7493 |
+
"learning_rate": 2.98813272898398e-05,
|
| 7494 |
+
"loss": 0.6427,
|
| 7495 |
+
"step": 1066
|
| 7496 |
+
},
|
| 7497 |
+
{
|
| 7498 |
+
"epoch": 0.30014064697609,
|
| 7499 |
+
"grad_norm": 0.47763392329216003,
|
| 7500 |
+
"learning_rate": 2.9722980881075847e-05,
|
| 7501 |
+
"loss": 0.5057,
|
| 7502 |
+
"step": 1067
|
| 7503 |
+
},
|
| 7504 |
+
{
|
| 7505 |
+
"epoch": 0.30042194092827,
|
| 7506 |
+
"grad_norm": 0.4706348478794098,
|
| 7507 |
+
"learning_rate": 2.9564981871320652e-05,
|
| 7508 |
+
"loss": 0.6105,
|
| 7509 |
+
"step": 1068
|
| 7510 |
+
},
|
| 7511 |
+
{
|
| 7512 |
+
"epoch": 0.30042194092827,
|
| 7513 |
+
"eval_loss": 0.5802850723266602,
|
| 7514 |
+
"eval_runtime": 78.1644,
|
| 7515 |
+
"eval_samples_per_second": 19.152,
|
| 7516 |
+
"eval_steps_per_second": 9.582,
|
| 7517 |
+
"step": 1068
|
| 7518 |
}
|
| 7519 |
],
|
| 7520 |
"logging_steps": 1,
|
|
|
|
| 7534 |
"attributes": {}
|
| 7535 |
}
|
| 7536 |
},
|
| 7537 |
+
"total_flos": 3.5705837810456986e+17,
|
| 7538 |
"train_batch_size": 2,
|
| 7539 |
"trial_name": null,
|
| 7540 |
"trial_params": null
|