Training in progress, step 1073, 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 35237104
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:35391f828d6a65dfb30116575782c1078d38ebbf05b07f0df7ad400119e875bb
|
| 3 |
size 35237104
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 70667778
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4fc07d0557c6f4c953db9a61497b8de246246c744bdfb7befff45007f8b159be
|
| 3 |
size 70667778
|
last-checkpoint/rng_state_0.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:14fcaee8795c235151c3385e44919e8fe5f67a07e67372d5baefe16a71da137a
|
| 3 |
size 14512
|
last-checkpoint/rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:de1dc5b07f9e5751c30f2f36b200155f98d26d9e6c4a4a993e98ac5f2d093a2d
|
| 3 |
size 14512
|
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 sha256:966e055b42cf10d79243d6fb4d4f4ad5954bc720f1b053796dbdfd80c9cf8da7
|
| 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": 269,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5688,6 +5688,1868 @@
|
|
| 5688 |
"eval_samples_per_second": 14.12,
|
| 5689 |
"eval_steps_per_second": 3.532,
|
| 5690 |
"step": 807
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5691 |
}
|
| 5692 |
],
|
| 5693 |
"logging_steps": 1,
|
|
@@ -5702,12 +7564,12 @@
|
|
| 5702 |
"should_evaluate": false,
|
| 5703 |
"should_log": false,
|
| 5704 |
"should_save": true,
|
| 5705 |
-
"should_training_stop":
|
| 5706 |
},
|
| 5707 |
"attributes": {}
|
| 5708 |
}
|
| 5709 |
},
|
| 5710 |
-
"total_flos": 1.
|
| 5711 |
"train_batch_size": 2,
|
| 5712 |
"trial_name": null,
|
| 5713 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
"eval_steps": 269,
|
| 6 |
+
"global_step": 1073,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5688 |
"eval_samples_per_second": 14.12,
|
| 5689 |
"eval_steps_per_second": 3.532,
|
| 5690 |
"step": 807
|
| 5691 |
+
},
|
| 5692 |
+
{
|
| 5693 |
+
"epoch": 0.7530288909599254,
|
| 5694 |
+
"grad_norm": 0.1286248415708542,
|
| 5695 |
+
"learning_rate": 1.5663106812083743e-05,
|
| 5696 |
+
"loss": 0.0035,
|
| 5697 |
+
"step": 808
|
| 5698 |
+
},
|
| 5699 |
+
{
|
| 5700 |
+
"epoch": 0.7539608574091333,
|
| 5701 |
+
"grad_norm": 0.29819682240486145,
|
| 5702 |
+
"learning_rate": 1.555165404621567e-05,
|
| 5703 |
+
"loss": 0.0043,
|
| 5704 |
+
"step": 809
|
| 5705 |
+
},
|
| 5706 |
+
{
|
| 5707 |
+
"epoch": 0.7548928238583411,
|
| 5708 |
+
"grad_norm": 1.4682197570800781,
|
| 5709 |
+
"learning_rate": 1.5440526155510763e-05,
|
| 5710 |
+
"loss": 0.0349,
|
| 5711 |
+
"step": 810
|
| 5712 |
+
},
|
| 5713 |
+
{
|
| 5714 |
+
"epoch": 0.7558247903075489,
|
| 5715 |
+
"grad_norm": 2.5306453704833984,
|
| 5716 |
+
"learning_rate": 1.532972418799298e-05,
|
| 5717 |
+
"loss": 0.1089,
|
| 5718 |
+
"step": 811
|
| 5719 |
+
},
|
| 5720 |
+
{
|
| 5721 |
+
"epoch": 0.7567567567567568,
|
| 5722 |
+
"grad_norm": 0.7194310426712036,
|
| 5723 |
+
"learning_rate": 1.5219249188612556e-05,
|
| 5724 |
+
"loss": 0.0092,
|
| 5725 |
+
"step": 812
|
| 5726 |
+
},
|
| 5727 |
+
{
|
| 5728 |
+
"epoch": 0.7576887232059646,
|
| 5729 |
+
"grad_norm": 0.7676751613616943,
|
| 5730 |
+
"learning_rate": 1.5109102199236152e-05,
|
| 5731 |
+
"loss": 0.0195,
|
| 5732 |
+
"step": 813
|
| 5733 |
+
},
|
| 5734 |
+
{
|
| 5735 |
+
"epoch": 0.7586206896551724,
|
| 5736 |
+
"grad_norm": 0.3930986225605011,
|
| 5737 |
+
"learning_rate": 1.4999284258637053e-05,
|
| 5738 |
+
"loss": 0.0055,
|
| 5739 |
+
"step": 814
|
| 5740 |
+
},
|
| 5741 |
+
{
|
| 5742 |
+
"epoch": 0.7595526561043803,
|
| 5743 |
+
"grad_norm": 1.3155858516693115,
|
| 5744 |
+
"learning_rate": 1.488979640248534e-05,
|
| 5745 |
+
"loss": 0.0408,
|
| 5746 |
+
"step": 815
|
| 5747 |
+
},
|
| 5748 |
+
{
|
| 5749 |
+
"epoch": 0.7604846225535881,
|
| 5750 |
+
"grad_norm": 0.8961852788925171,
|
| 5751 |
+
"learning_rate": 1.4780639663338125e-05,
|
| 5752 |
+
"loss": 0.014,
|
| 5753 |
+
"step": 816
|
| 5754 |
+
},
|
| 5755 |
+
{
|
| 5756 |
+
"epoch": 0.7614165890027959,
|
| 5757 |
+
"grad_norm": 0.16851025819778442,
|
| 5758 |
+
"learning_rate": 1.4671815070629868e-05,
|
| 5759 |
+
"loss": 0.0031,
|
| 5760 |
+
"step": 817
|
| 5761 |
+
},
|
| 5762 |
+
{
|
| 5763 |
+
"epoch": 0.7623485554520038,
|
| 5764 |
+
"grad_norm": 0.2377876490354538,
|
| 5765 |
+
"learning_rate": 1.4563323650662586e-05,
|
| 5766 |
+
"loss": 0.0036,
|
| 5767 |
+
"step": 818
|
| 5768 |
+
},
|
| 5769 |
+
{
|
| 5770 |
+
"epoch": 0.7632805219012115,
|
| 5771 |
+
"grad_norm": 0.26792046427726746,
|
| 5772 |
+
"learning_rate": 1.4455166426596223e-05,
|
| 5773 |
+
"loss": 0.0053,
|
| 5774 |
+
"step": 819
|
| 5775 |
+
},
|
| 5776 |
+
{
|
| 5777 |
+
"epoch": 0.7642124883504194,
|
| 5778 |
+
"grad_norm": 0.1533597856760025,
|
| 5779 |
+
"learning_rate": 1.434734441843899e-05,
|
| 5780 |
+
"loss": 0.0032,
|
| 5781 |
+
"step": 820
|
| 5782 |
+
},
|
| 5783 |
+
{
|
| 5784 |
+
"epoch": 0.7651444547996272,
|
| 5785 |
+
"grad_norm": 0.13230851292610168,
|
| 5786 |
+
"learning_rate": 1.4239858643037751e-05,
|
| 5787 |
+
"loss": 0.0025,
|
| 5788 |
+
"step": 821
|
| 5789 |
+
},
|
| 5790 |
+
{
|
| 5791 |
+
"epoch": 0.766076421248835,
|
| 5792 |
+
"grad_norm": 0.42541301250457764,
|
| 5793 |
+
"learning_rate": 1.4132710114068426e-05,
|
| 5794 |
+
"loss": 0.0078,
|
| 5795 |
+
"step": 822
|
| 5796 |
+
},
|
| 5797 |
+
{
|
| 5798 |
+
"epoch": 0.7670083876980429,
|
| 5799 |
+
"grad_norm": 1.640199899673462,
|
| 5800 |
+
"learning_rate": 1.4025899842026442e-05,
|
| 5801 |
+
"loss": 0.0358,
|
| 5802 |
+
"step": 823
|
| 5803 |
+
},
|
| 5804 |
+
{
|
| 5805 |
+
"epoch": 0.7679403541472507,
|
| 5806 |
+
"grad_norm": 3.3744680881500244,
|
| 5807 |
+
"learning_rate": 1.3919428834217163e-05,
|
| 5808 |
+
"loss": 0.066,
|
| 5809 |
+
"step": 824
|
| 5810 |
+
},
|
| 5811 |
+
{
|
| 5812 |
+
"epoch": 0.7688723205964585,
|
| 5813 |
+
"grad_norm": 4.397242069244385,
|
| 5814 |
+
"learning_rate": 1.3813298094746491e-05,
|
| 5815 |
+
"loss": 0.1446,
|
| 5816 |
+
"step": 825
|
| 5817 |
+
},
|
| 5818 |
+
{
|
| 5819 |
+
"epoch": 0.7698042870456664,
|
| 5820 |
+
"grad_norm": 1.2552555799484253,
|
| 5821 |
+
"learning_rate": 1.3707508624511262e-05,
|
| 5822 |
+
"loss": 0.0247,
|
| 5823 |
+
"step": 826
|
| 5824 |
+
},
|
| 5825 |
+
{
|
| 5826 |
+
"epoch": 0.7707362534948742,
|
| 5827 |
+
"grad_norm": 0.9916906356811523,
|
| 5828 |
+
"learning_rate": 1.36020614211899e-05,
|
| 5829 |
+
"loss": 0.033,
|
| 5830 |
+
"step": 827
|
| 5831 |
+
},
|
| 5832 |
+
{
|
| 5833 |
+
"epoch": 0.771668219944082,
|
| 5834 |
+
"grad_norm": 1.0547466278076172,
|
| 5835 |
+
"learning_rate": 1.349695747923298e-05,
|
| 5836 |
+
"loss": 0.0706,
|
| 5837 |
+
"step": 828
|
| 5838 |
+
},
|
| 5839 |
+
{
|
| 5840 |
+
"epoch": 0.7726001863932899,
|
| 5841 |
+
"grad_norm": 0.8943658471107483,
|
| 5842 |
+
"learning_rate": 1.339219778985385e-05,
|
| 5843 |
+
"loss": 0.0239,
|
| 5844 |
+
"step": 829
|
| 5845 |
+
},
|
| 5846 |
+
{
|
| 5847 |
+
"epoch": 0.7735321528424977,
|
| 5848 |
+
"grad_norm": 1.5261614322662354,
|
| 5849 |
+
"learning_rate": 1.3287783341019278e-05,
|
| 5850 |
+
"loss": 0.0195,
|
| 5851 |
+
"step": 830
|
| 5852 |
+
},
|
| 5853 |
+
{
|
| 5854 |
+
"epoch": 0.7744641192917054,
|
| 5855 |
+
"grad_norm": 0.11581642925739288,
|
| 5856 |
+
"learning_rate": 1.3183715117440142e-05,
|
| 5857 |
+
"loss": 0.0019,
|
| 5858 |
+
"step": 831
|
| 5859 |
+
},
|
| 5860 |
+
{
|
| 5861 |
+
"epoch": 0.7753960857409133,
|
| 5862 |
+
"grad_norm": 0.190779909491539,
|
| 5863 |
+
"learning_rate": 1.3079994100562159e-05,
|
| 5864 |
+
"loss": 0.0051,
|
| 5865 |
+
"step": 832
|
| 5866 |
+
},
|
| 5867 |
+
{
|
| 5868 |
+
"epoch": 0.7763280521901211,
|
| 5869 |
+
"grad_norm": 0.7171970009803772,
|
| 5870 |
+
"learning_rate": 1.2976621268556571e-05,
|
| 5871 |
+
"loss": 0.0258,
|
| 5872 |
+
"step": 833
|
| 5873 |
+
},
|
| 5874 |
+
{
|
| 5875 |
+
"epoch": 0.777260018639329,
|
| 5876 |
+
"grad_norm": 1.5177792310714722,
|
| 5877 |
+
"learning_rate": 1.2873597596311027e-05,
|
| 5878 |
+
"loss": 0.0589,
|
| 5879 |
+
"step": 834
|
| 5880 |
+
},
|
| 5881 |
+
{
|
| 5882 |
+
"epoch": 0.7781919850885368,
|
| 5883 |
+
"grad_norm": 0.1316295564174652,
|
| 5884 |
+
"learning_rate": 1.2770924055420258e-05,
|
| 5885 |
+
"loss": 0.0026,
|
| 5886 |
+
"step": 835
|
| 5887 |
+
},
|
| 5888 |
+
{
|
| 5889 |
+
"epoch": 0.7791239515377446,
|
| 5890 |
+
"grad_norm": 0.5268474221229553,
|
| 5891 |
+
"learning_rate": 1.2668601614177017e-05,
|
| 5892 |
+
"loss": 0.0101,
|
| 5893 |
+
"step": 836
|
| 5894 |
+
},
|
| 5895 |
+
{
|
| 5896 |
+
"epoch": 0.7800559179869525,
|
| 5897 |
+
"grad_norm": 0.8330385684967041,
|
| 5898 |
+
"learning_rate": 1.2566631237562893e-05,
|
| 5899 |
+
"loss": 0.0351,
|
| 5900 |
+
"step": 837
|
| 5901 |
+
},
|
| 5902 |
+
{
|
| 5903 |
+
"epoch": 0.7809878844361603,
|
| 5904 |
+
"grad_norm": 0.4103114902973175,
|
| 5905 |
+
"learning_rate": 1.246501388723923e-05,
|
| 5906 |
+
"loss": 0.009,
|
| 5907 |
+
"step": 838
|
| 5908 |
+
},
|
| 5909 |
+
{
|
| 5910 |
+
"epoch": 0.7819198508853681,
|
| 5911 |
+
"grad_norm": 0.7705252766609192,
|
| 5912 |
+
"learning_rate": 1.2363750521538064e-05,
|
| 5913 |
+
"loss": 0.0211,
|
| 5914 |
+
"step": 839
|
| 5915 |
+
},
|
| 5916 |
+
{
|
| 5917 |
+
"epoch": 0.782851817334576,
|
| 5918 |
+
"grad_norm": 1.3898861408233643,
|
| 5919 |
+
"learning_rate": 1.2262842095453065e-05,
|
| 5920 |
+
"loss": 0.0678,
|
| 5921 |
+
"step": 840
|
| 5922 |
+
},
|
| 5923 |
+
{
|
| 5924 |
+
"epoch": 0.7837837837837838,
|
| 5925 |
+
"grad_norm": 2.382798910140991,
|
| 5926 |
+
"learning_rate": 1.2162289560630524e-05,
|
| 5927 |
+
"loss": 0.0201,
|
| 5928 |
+
"step": 841
|
| 5929 |
+
},
|
| 5930 |
+
{
|
| 5931 |
+
"epoch": 0.7847157502329916,
|
| 5932 |
+
"grad_norm": 0.8590038418769836,
|
| 5933 |
+
"learning_rate": 1.2062093865360458e-05,
|
| 5934 |
+
"loss": 0.0196,
|
| 5935 |
+
"step": 842
|
| 5936 |
+
},
|
| 5937 |
+
{
|
| 5938 |
+
"epoch": 0.7856477166821995,
|
| 5939 |
+
"grad_norm": 0.4806217849254608,
|
| 5940 |
+
"learning_rate": 1.1962255954567537e-05,
|
| 5941 |
+
"loss": 0.0081,
|
| 5942 |
+
"step": 843
|
| 5943 |
+
},
|
| 5944 |
+
{
|
| 5945 |
+
"epoch": 0.7865796831314072,
|
| 5946 |
+
"grad_norm": 0.6536293625831604,
|
| 5947 |
+
"learning_rate": 1.1862776769802276e-05,
|
| 5948 |
+
"loss": 0.01,
|
| 5949 |
+
"step": 844
|
| 5950 |
+
},
|
| 5951 |
+
{
|
| 5952 |
+
"epoch": 0.7875116495806151,
|
| 5953 |
+
"grad_norm": 1.8112365007400513,
|
| 5954 |
+
"learning_rate": 1.1763657249232107e-05,
|
| 5955 |
+
"loss": 0.0299,
|
| 5956 |
+
"step": 845
|
| 5957 |
+
},
|
| 5958 |
+
{
|
| 5959 |
+
"epoch": 0.7884436160298229,
|
| 5960 |
+
"grad_norm": 0.0851447582244873,
|
| 5961 |
+
"learning_rate": 1.1664898327632551e-05,
|
| 5962 |
+
"loss": 0.0015,
|
| 5963 |
+
"step": 846
|
| 5964 |
+
},
|
| 5965 |
+
{
|
| 5966 |
+
"epoch": 0.7893755824790307,
|
| 5967 |
+
"grad_norm": 3.3141989707946777,
|
| 5968 |
+
"learning_rate": 1.1566500936378388e-05,
|
| 5969 |
+
"loss": 0.0935,
|
| 5970 |
+
"step": 847
|
| 5971 |
+
},
|
| 5972 |
+
{
|
| 5973 |
+
"epoch": 0.7903075489282386,
|
| 5974 |
+
"grad_norm": 0.655958354473114,
|
| 5975 |
+
"learning_rate": 1.146846600343488e-05,
|
| 5976 |
+
"loss": 0.0092,
|
| 5977 |
+
"step": 848
|
| 5978 |
+
},
|
| 5979 |
+
{
|
| 5980 |
+
"epoch": 0.7912395153774464,
|
| 5981 |
+
"grad_norm": 2.1537182331085205,
|
| 5982 |
+
"learning_rate": 1.1370794453349037e-05,
|
| 5983 |
+
"loss": 0.0292,
|
| 5984 |
+
"step": 849
|
| 5985 |
+
},
|
| 5986 |
+
{
|
| 5987 |
+
"epoch": 0.7921714818266542,
|
| 5988 |
+
"grad_norm": 1.5979948043823242,
|
| 5989 |
+
"learning_rate": 1.1273487207240847e-05,
|
| 5990 |
+
"loss": 0.0682,
|
| 5991 |
+
"step": 850
|
| 5992 |
+
},
|
| 5993 |
+
{
|
| 5994 |
+
"epoch": 0.7931034482758621,
|
| 5995 |
+
"grad_norm": 0.32921919226646423,
|
| 5996 |
+
"learning_rate": 1.1176545182794674e-05,
|
| 5997 |
+
"loss": 0.0051,
|
| 5998 |
+
"step": 851
|
| 5999 |
+
},
|
| 6000 |
+
{
|
| 6001 |
+
"epoch": 0.7940354147250699,
|
| 6002 |
+
"grad_norm": 0.3104652762413025,
|
| 6003 |
+
"learning_rate": 1.1079969294250514e-05,
|
| 6004 |
+
"loss": 0.0053,
|
| 6005 |
+
"step": 852
|
| 6006 |
+
},
|
| 6007 |
+
{
|
| 6008 |
+
"epoch": 0.7949673811742777,
|
| 6009 |
+
"grad_norm": 0.23145949840545654,
|
| 6010 |
+
"learning_rate": 1.0983760452395413e-05,
|
| 6011 |
+
"loss": 0.0033,
|
| 6012 |
+
"step": 853
|
| 6013 |
+
},
|
| 6014 |
+
{
|
| 6015 |
+
"epoch": 0.7958993476234856,
|
| 6016 |
+
"grad_norm": 0.44747766852378845,
|
| 6017 |
+
"learning_rate": 1.0887919564554894e-05,
|
| 6018 |
+
"loss": 0.0089,
|
| 6019 |
+
"step": 854
|
| 6020 |
+
},
|
| 6021 |
+
{
|
| 6022 |
+
"epoch": 0.7968313140726934,
|
| 6023 |
+
"grad_norm": 2.918243169784546,
|
| 6024 |
+
"learning_rate": 1.079244753458437e-05,
|
| 6025 |
+
"loss": 0.0767,
|
| 6026 |
+
"step": 855
|
| 6027 |
+
},
|
| 6028 |
+
{
|
| 6029 |
+
"epoch": 0.7977632805219013,
|
| 6030 |
+
"grad_norm": 0.6843606233596802,
|
| 6031 |
+
"learning_rate": 1.0697345262860636e-05,
|
| 6032 |
+
"loss": 0.0144,
|
| 6033 |
+
"step": 856
|
| 6034 |
+
},
|
| 6035 |
+
{
|
| 6036 |
+
"epoch": 0.798695246971109,
|
| 6037 |
+
"grad_norm": 0.1016877144575119,
|
| 6038 |
+
"learning_rate": 1.0602613646273374e-05,
|
| 6039 |
+
"loss": 0.0033,
|
| 6040 |
+
"step": 857
|
| 6041 |
+
},
|
| 6042 |
+
{
|
| 6043 |
+
"epoch": 0.7996272134203168,
|
| 6044 |
+
"grad_norm": 0.7905264496803284,
|
| 6045 |
+
"learning_rate": 1.0508253578216693e-05,
|
| 6046 |
+
"loss": 0.0268,
|
| 6047 |
+
"step": 858
|
| 6048 |
+
},
|
| 6049 |
+
{
|
| 6050 |
+
"epoch": 0.8005591798695247,
|
| 6051 |
+
"grad_norm": 1.3822872638702393,
|
| 6052 |
+
"learning_rate": 1.0414265948580692e-05,
|
| 6053 |
+
"loss": 0.0685,
|
| 6054 |
+
"step": 859
|
| 6055 |
+
},
|
| 6056 |
+
{
|
| 6057 |
+
"epoch": 0.8014911463187325,
|
| 6058 |
+
"grad_norm": 0.5625484585762024,
|
| 6059 |
+
"learning_rate": 1.032065164374313e-05,
|
| 6060 |
+
"loss": 0.0123,
|
| 6061 |
+
"step": 860
|
| 6062 |
+
},
|
| 6063 |
+
{
|
| 6064 |
+
"epoch": 0.8024231127679403,
|
| 6065 |
+
"grad_norm": 0.05632966756820679,
|
| 6066 |
+
"learning_rate": 1.0227411546560962e-05,
|
| 6067 |
+
"loss": 0.001,
|
| 6068 |
+
"step": 861
|
| 6069 |
+
},
|
| 6070 |
+
{
|
| 6071 |
+
"epoch": 0.8033550792171482,
|
| 6072 |
+
"grad_norm": 0.7446116805076599,
|
| 6073 |
+
"learning_rate": 1.01345465363621e-05,
|
| 6074 |
+
"loss": 0.0202,
|
| 6075 |
+
"step": 862
|
| 6076 |
+
},
|
| 6077 |
+
{
|
| 6078 |
+
"epoch": 0.804287045666356,
|
| 6079 |
+
"grad_norm": 0.09283380210399628,
|
| 6080 |
+
"learning_rate": 1.0042057488937067e-05,
|
| 6081 |
+
"loss": 0.0016,
|
| 6082 |
+
"step": 863
|
| 6083 |
+
},
|
| 6084 |
+
{
|
| 6085 |
+
"epoch": 0.8052190121155638,
|
| 6086 |
+
"grad_norm": 0.27069157361984253,
|
| 6087 |
+
"learning_rate": 9.949945276530781e-06,
|
| 6088 |
+
"loss": 0.0053,
|
| 6089 |
+
"step": 864
|
| 6090 |
+
},
|
| 6091 |
+
{
|
| 6092 |
+
"epoch": 0.8061509785647717,
|
| 6093 |
+
"grad_norm": 0.8382067680358887,
|
| 6094 |
+
"learning_rate": 9.858210767834292e-06,
|
| 6095 |
+
"loss": 0.0226,
|
| 6096 |
+
"step": 865
|
| 6097 |
+
},
|
| 6098 |
+
{
|
| 6099 |
+
"epoch": 0.8070829450139795,
|
| 6100 |
+
"grad_norm": 1.0016965866088867,
|
| 6101 |
+
"learning_rate": 9.766854827976617e-06,
|
| 6102 |
+
"loss": 0.041,
|
| 6103 |
+
"step": 866
|
| 6104 |
+
},
|
| 6105 |
+
{
|
| 6106 |
+
"epoch": 0.8080149114631874,
|
| 6107 |
+
"grad_norm": 1.245935320854187,
|
| 6108 |
+
"learning_rate": 9.675878318516546e-06,
|
| 6109 |
+
"loss": 0.0404,
|
| 6110 |
+
"step": 867
|
| 6111 |
+
},
|
| 6112 |
+
{
|
| 6113 |
+
"epoch": 0.8089468779123952,
|
| 6114 |
+
"grad_norm": 0.5361421704292297,
|
| 6115 |
+
"learning_rate": 9.58528209743459e-06,
|
| 6116 |
+
"loss": 0.0076,
|
| 6117 |
+
"step": 868
|
| 6118 |
+
},
|
| 6119 |
+
{
|
| 6120 |
+
"epoch": 0.809878844361603,
|
| 6121 |
+
"grad_norm": 0.09631752967834473,
|
| 6122 |
+
"learning_rate": 9.495067019124792e-06,
|
| 6123 |
+
"loss": 0.0028,
|
| 6124 |
+
"step": 869
|
| 6125 |
+
},
|
| 6126 |
+
{
|
| 6127 |
+
"epoch": 0.8108108108108109,
|
| 6128 |
+
"grad_norm": 0.6793972253799438,
|
| 6129 |
+
"learning_rate": 9.405233934386726e-06,
|
| 6130 |
+
"loss": 0.0057,
|
| 6131 |
+
"step": 870
|
| 6132 |
+
},
|
| 6133 |
+
{
|
| 6134 |
+
"epoch": 0.8117427772600186,
|
| 6135 |
+
"grad_norm": 3.4450275897979736,
|
| 6136 |
+
"learning_rate": 9.315783690417479e-06,
|
| 6137 |
+
"loss": 0.053,
|
| 6138 |
+
"step": 871
|
| 6139 |
+
},
|
| 6140 |
+
{
|
| 6141 |
+
"epoch": 0.8126747437092264,
|
| 6142 |
+
"grad_norm": 1.280646800994873,
|
| 6143 |
+
"learning_rate": 9.226717130803636e-06,
|
| 6144 |
+
"loss": 0.042,
|
| 6145 |
+
"step": 872
|
| 6146 |
+
},
|
| 6147 |
+
{
|
| 6148 |
+
"epoch": 0.8136067101584343,
|
| 6149 |
+
"grad_norm": 1.587612509727478,
|
| 6150 |
+
"learning_rate": 9.138035095513336e-06,
|
| 6151 |
+
"loss": 0.0671,
|
| 6152 |
+
"step": 873
|
| 6153 |
+
},
|
| 6154 |
+
{
|
| 6155 |
+
"epoch": 0.8145386766076421,
|
| 6156 |
+
"grad_norm": 1.265060305595398,
|
| 6157 |
+
"learning_rate": 9.049738420888349e-06,
|
| 6158 |
+
"loss": 0.014,
|
| 6159 |
+
"step": 874
|
| 6160 |
+
},
|
| 6161 |
+
{
|
| 6162 |
+
"epoch": 0.8154706430568499,
|
| 6163 |
+
"grad_norm": 1.8084266185760498,
|
| 6164 |
+
"learning_rate": 8.961827939636196e-06,
|
| 6165 |
+
"loss": 0.0891,
|
| 6166 |
+
"step": 875
|
| 6167 |
+
},
|
| 6168 |
+
{
|
| 6169 |
+
"epoch": 0.8164026095060578,
|
| 6170 |
+
"grad_norm": 0.39547011256217957,
|
| 6171 |
+
"learning_rate": 8.87430448082227e-06,
|
| 6172 |
+
"loss": 0.0022,
|
| 6173 |
+
"step": 876
|
| 6174 |
+
},
|
| 6175 |
+
{
|
| 6176 |
+
"epoch": 0.8173345759552656,
|
| 6177 |
+
"grad_norm": 0.3031284213066101,
|
| 6178 |
+
"learning_rate": 8.787168869862066e-06,
|
| 6179 |
+
"loss": 0.0066,
|
| 6180 |
+
"step": 877
|
| 6181 |
+
},
|
| 6182 |
+
{
|
| 6183 |
+
"epoch": 0.8182665424044734,
|
| 6184 |
+
"grad_norm": 0.15621918439865112,
|
| 6185 |
+
"learning_rate": 8.700421928513352e-06,
|
| 6186 |
+
"loss": 0.0032,
|
| 6187 |
+
"step": 878
|
| 6188 |
+
},
|
| 6189 |
+
{
|
| 6190 |
+
"epoch": 0.8191985088536813,
|
| 6191 |
+
"grad_norm": 0.22168925404548645,
|
| 6192 |
+
"learning_rate": 8.614064474868421e-06,
|
| 6193 |
+
"loss": 0.0042,
|
| 6194 |
+
"step": 879
|
| 6195 |
+
},
|
| 6196 |
+
{
|
| 6197 |
+
"epoch": 0.8201304753028891,
|
| 6198 |
+
"grad_norm": 1.5389249324798584,
|
| 6199 |
+
"learning_rate": 8.528097323346407e-06,
|
| 6200 |
+
"loss": 0.0427,
|
| 6201 |
+
"step": 880
|
| 6202 |
+
},
|
| 6203 |
+
{
|
| 6204 |
+
"epoch": 0.821062441752097,
|
| 6205 |
+
"grad_norm": 1.5170341730117798,
|
| 6206 |
+
"learning_rate": 8.442521284685573e-06,
|
| 6207 |
+
"loss": 0.0335,
|
| 6208 |
+
"step": 881
|
| 6209 |
+
},
|
| 6210 |
+
{
|
| 6211 |
+
"epoch": 0.8219944082013048,
|
| 6212 |
+
"grad_norm": 0.08587615191936493,
|
| 6213 |
+
"learning_rate": 8.357337165935675e-06,
|
| 6214 |
+
"loss": 0.0023,
|
| 6215 |
+
"step": 882
|
| 6216 |
+
},
|
| 6217 |
+
{
|
| 6218 |
+
"epoch": 0.8229263746505125,
|
| 6219 |
+
"grad_norm": 0.17892524600028992,
|
| 6220 |
+
"learning_rate": 8.27254577045039e-06,
|
| 6221 |
+
"loss": 0.0027,
|
| 6222 |
+
"step": 883
|
| 6223 |
+
},
|
| 6224 |
+
{
|
| 6225 |
+
"epoch": 0.8238583410997204,
|
| 6226 |
+
"grad_norm": 0.17656829953193665,
|
| 6227 |
+
"learning_rate": 8.188147897879667e-06,
|
| 6228 |
+
"loss": 0.0046,
|
| 6229 |
+
"step": 884
|
| 6230 |
+
},
|
| 6231 |
+
{
|
| 6232 |
+
"epoch": 0.8247903075489282,
|
| 6233 |
+
"grad_norm": 0.5518335700035095,
|
| 6234 |
+
"learning_rate": 8.104144344162229e-06,
|
| 6235 |
+
"loss": 0.0124,
|
| 6236 |
+
"step": 885
|
| 6237 |
+
},
|
| 6238 |
+
{
|
| 6239 |
+
"epoch": 0.825722273998136,
|
| 6240 |
+
"grad_norm": 0.30466359853744507,
|
| 6241 |
+
"learning_rate": 8.02053590151805e-06,
|
| 6242 |
+
"loss": 0.0077,
|
| 6243 |
+
"step": 886
|
| 6244 |
+
},
|
| 6245 |
+
{
|
| 6246 |
+
"epoch": 0.8266542404473439,
|
| 6247 |
+
"grad_norm": 0.7674888372421265,
|
| 6248 |
+
"learning_rate": 7.937323358440935e-06,
|
| 6249 |
+
"loss": 0.0248,
|
| 6250 |
+
"step": 887
|
| 6251 |
+
},
|
| 6252 |
+
{
|
| 6253 |
+
"epoch": 0.8275862068965517,
|
| 6254 |
+
"grad_norm": 0.055480506271123886,
|
| 6255 |
+
"learning_rate": 7.854507499691005e-06,
|
| 6256 |
+
"loss": 0.0011,
|
| 6257 |
+
"step": 888
|
| 6258 |
+
},
|
| 6259 |
+
{
|
| 6260 |
+
"epoch": 0.8285181733457595,
|
| 6261 |
+
"grad_norm": 0.22888953983783722,
|
| 6262 |
+
"learning_rate": 7.772089106287344e-06,
|
| 6263 |
+
"loss": 0.0029,
|
| 6264 |
+
"step": 889
|
| 6265 |
+
},
|
| 6266 |
+
{
|
| 6267 |
+
"epoch": 0.8294501397949674,
|
| 6268 |
+
"grad_norm": 0.0735834464430809,
|
| 6269 |
+
"learning_rate": 7.690068955500624e-06,
|
| 6270 |
+
"loss": 0.0015,
|
| 6271 |
+
"step": 890
|
| 6272 |
+
},
|
| 6273 |
+
{
|
| 6274 |
+
"epoch": 0.8303821062441752,
|
| 6275 |
+
"grad_norm": 1.0018421411514282,
|
| 6276 |
+
"learning_rate": 7.608447820845771e-06,
|
| 6277 |
+
"loss": 0.0309,
|
| 6278 |
+
"step": 891
|
| 6279 |
+
},
|
| 6280 |
+
{
|
| 6281 |
+
"epoch": 0.8313140726933831,
|
| 6282 |
+
"grad_norm": 1.2352806329727173,
|
| 6283 |
+
"learning_rate": 7.527226472074678e-06,
|
| 6284 |
+
"loss": 0.0362,
|
| 6285 |
+
"step": 892
|
| 6286 |
+
},
|
| 6287 |
+
{
|
| 6288 |
+
"epoch": 0.8322460391425909,
|
| 6289 |
+
"grad_norm": 0.9229680299758911,
|
| 6290 |
+
"learning_rate": 7.446405675168938e-06,
|
| 6291 |
+
"loss": 0.0236,
|
| 6292 |
+
"step": 893
|
| 6293 |
+
},
|
| 6294 |
+
{
|
| 6295 |
+
"epoch": 0.8331780055917987,
|
| 6296 |
+
"grad_norm": 0.12361752986907959,
|
| 6297 |
+
"learning_rate": 7.365986192332625e-06,
|
| 6298 |
+
"loss": 0.0015,
|
| 6299 |
+
"step": 894
|
| 6300 |
+
},
|
| 6301 |
+
{
|
| 6302 |
+
"epoch": 0.8341099720410066,
|
| 6303 |
+
"grad_norm": 0.18358244001865387,
|
| 6304 |
+
"learning_rate": 7.285968781985092e-06,
|
| 6305 |
+
"loss": 0.0041,
|
| 6306 |
+
"step": 895
|
| 6307 |
+
},
|
| 6308 |
+
{
|
| 6309 |
+
"epoch": 0.8350419384902144,
|
| 6310 |
+
"grad_norm": 1.7125695943832397,
|
| 6311 |
+
"learning_rate": 7.206354198753862e-06,
|
| 6312 |
+
"loss": 0.0712,
|
| 6313 |
+
"step": 896
|
| 6314 |
+
},
|
| 6315 |
+
{
|
| 6316 |
+
"epoch": 0.8359739049394221,
|
| 6317 |
+
"grad_norm": 2.4467711448669434,
|
| 6318 |
+
"learning_rate": 7.127143193467445e-06,
|
| 6319 |
+
"loss": 0.0988,
|
| 6320 |
+
"step": 897
|
| 6321 |
+
},
|
| 6322 |
+
{
|
| 6323 |
+
"epoch": 0.83690587138863,
|
| 6324 |
+
"grad_norm": 2.2507357597351074,
|
| 6325 |
+
"learning_rate": 7.048336513148307e-06,
|
| 6326 |
+
"loss": 0.0513,
|
| 6327 |
+
"step": 898
|
| 6328 |
+
},
|
| 6329 |
+
{
|
| 6330 |
+
"epoch": 0.8378378378378378,
|
| 6331 |
+
"grad_norm": 2.246809720993042,
|
| 6332 |
+
"learning_rate": 6.969934901005809e-06,
|
| 6333 |
+
"loss": 0.0265,
|
| 6334 |
+
"step": 899
|
| 6335 |
+
},
|
| 6336 |
+
{
|
| 6337 |
+
"epoch": 0.8387698042870456,
|
| 6338 |
+
"grad_norm": 1.8378955125808716,
|
| 6339 |
+
"learning_rate": 6.8919390964291895e-06,
|
| 6340 |
+
"loss": 0.1061,
|
| 6341 |
+
"step": 900
|
| 6342 |
+
},
|
| 6343 |
+
{
|
| 6344 |
+
"epoch": 0.8397017707362535,
|
| 6345 |
+
"grad_norm": 0.47382405400276184,
|
| 6346 |
+
"learning_rate": 6.814349834980621e-06,
|
| 6347 |
+
"loss": 0.011,
|
| 6348 |
+
"step": 901
|
| 6349 |
+
},
|
| 6350 |
+
{
|
| 6351 |
+
"epoch": 0.8406337371854613,
|
| 6352 |
+
"grad_norm": 1.1276845932006836,
|
| 6353 |
+
"learning_rate": 6.7371678483882264e-06,
|
| 6354 |
+
"loss": 0.025,
|
| 6355 |
+
"step": 902
|
| 6356 |
+
},
|
| 6357 |
+
{
|
| 6358 |
+
"epoch": 0.8415657036346692,
|
| 6359 |
+
"grad_norm": 0.5048341155052185,
|
| 6360 |
+
"learning_rate": 6.660393864539222e-06,
|
| 6361 |
+
"loss": 0.0053,
|
| 6362 |
+
"step": 903
|
| 6363 |
+
},
|
| 6364 |
+
{
|
| 6365 |
+
"epoch": 0.842497670083877,
|
| 6366 |
+
"grad_norm": 0.9290698766708374,
|
| 6367 |
+
"learning_rate": 6.58402860747302e-06,
|
| 6368 |
+
"loss": 0.0501,
|
| 6369 |
+
"step": 904
|
| 6370 |
+
},
|
| 6371 |
+
{
|
| 6372 |
+
"epoch": 0.8434296365330848,
|
| 6373 |
+
"grad_norm": 1.0195822715759277,
|
| 6374 |
+
"learning_rate": 6.508072797374454e-06,
|
| 6375 |
+
"loss": 0.0182,
|
| 6376 |
+
"step": 905
|
| 6377 |
+
},
|
| 6378 |
+
{
|
| 6379 |
+
"epoch": 0.8443616029822927,
|
| 6380 |
+
"grad_norm": 0.4818008244037628,
|
| 6381 |
+
"learning_rate": 6.4325271505669025e-06,
|
| 6382 |
+
"loss": 0.0074,
|
| 6383 |
+
"step": 906
|
| 6384 |
+
},
|
| 6385 |
+
{
|
| 6386 |
+
"epoch": 0.8452935694315005,
|
| 6387 |
+
"grad_norm": 0.051283132284879684,
|
| 6388 |
+
"learning_rate": 6.3573923795056e-06,
|
| 6389 |
+
"loss": 0.0011,
|
| 6390 |
+
"step": 907
|
| 6391 |
+
},
|
| 6392 |
+
{
|
| 6393 |
+
"epoch": 0.8462255358807083,
|
| 6394 |
+
"grad_norm": 1.3800734281539917,
|
| 6395 |
+
"learning_rate": 6.282669192770896e-06,
|
| 6396 |
+
"loss": 0.0442,
|
| 6397 |
+
"step": 908
|
| 6398 |
+
},
|
| 6399 |
+
{
|
| 6400 |
+
"epoch": 0.8471575023299162,
|
| 6401 |
+
"grad_norm": 1.0327776670455933,
|
| 6402 |
+
"learning_rate": 6.208358295061572e-06,
|
| 6403 |
+
"loss": 0.037,
|
| 6404 |
+
"step": 909
|
| 6405 |
+
},
|
| 6406 |
+
{
|
| 6407 |
+
"epoch": 0.848089468779124,
|
| 6408 |
+
"grad_norm": 0.1635047346353531,
|
| 6409 |
+
"learning_rate": 6.134460387188206e-06,
|
| 6410 |
+
"loss": 0.003,
|
| 6411 |
+
"step": 910
|
| 6412 |
+
},
|
| 6413 |
+
{
|
| 6414 |
+
"epoch": 0.8490214352283317,
|
| 6415 |
+
"grad_norm": 0.17982259392738342,
|
| 6416 |
+
"learning_rate": 6.060976166066546e-06,
|
| 6417 |
+
"loss": 0.0032,
|
| 6418 |
+
"step": 911
|
| 6419 |
+
},
|
| 6420 |
+
{
|
| 6421 |
+
"epoch": 0.8499534016775396,
|
| 6422 |
+
"grad_norm": 0.2328854650259018,
|
| 6423 |
+
"learning_rate": 5.9879063247109405e-06,
|
| 6424 |
+
"loss": 0.0049,
|
| 6425 |
+
"step": 912
|
| 6426 |
+
},
|
| 6427 |
+
{
|
| 6428 |
+
"epoch": 0.8508853681267474,
|
| 6429 |
+
"grad_norm": 1.2415688037872314,
|
| 6430 |
+
"learning_rate": 5.915251552227829e-06,
|
| 6431 |
+
"loss": 0.0307,
|
| 6432 |
+
"step": 913
|
| 6433 |
+
},
|
| 6434 |
+
{
|
| 6435 |
+
"epoch": 0.8518173345759553,
|
| 6436 |
+
"grad_norm": 0.3968764841556549,
|
| 6437 |
+
"learning_rate": 5.8430125338092115e-06,
|
| 6438 |
+
"loss": 0.0087,
|
| 6439 |
+
"step": 914
|
| 6440 |
+
},
|
| 6441 |
+
{
|
| 6442 |
+
"epoch": 0.8527493010251631,
|
| 6443 |
+
"grad_norm": 1.2752315998077393,
|
| 6444 |
+
"learning_rate": 5.771189950726191e-06,
|
| 6445 |
+
"loss": 0.0312,
|
| 6446 |
+
"step": 915
|
| 6447 |
+
},
|
| 6448 |
+
{
|
| 6449 |
+
"epoch": 0.8536812674743709,
|
| 6450 |
+
"grad_norm": 0.7192822098731995,
|
| 6451 |
+
"learning_rate": 5.699784480322568e-06,
|
| 6452 |
+
"loss": 0.0108,
|
| 6453 |
+
"step": 916
|
| 6454 |
+
},
|
| 6455 |
+
{
|
| 6456 |
+
"epoch": 0.8546132339235788,
|
| 6457 |
+
"grad_norm": 0.34972473978996277,
|
| 6458 |
+
"learning_rate": 5.628796796008434e-06,
|
| 6459 |
+
"loss": 0.0029,
|
| 6460 |
+
"step": 917
|
| 6461 |
+
},
|
| 6462 |
+
{
|
| 6463 |
+
"epoch": 0.8555452003727866,
|
| 6464 |
+
"grad_norm": 0.3486124277114868,
|
| 6465 |
+
"learning_rate": 5.558227567253832e-06,
|
| 6466 |
+
"loss": 0.0051,
|
| 6467 |
+
"step": 918
|
| 6468 |
+
},
|
| 6469 |
+
{
|
| 6470 |
+
"epoch": 0.8564771668219944,
|
| 6471 |
+
"grad_norm": 0.3966856300830841,
|
| 6472 |
+
"learning_rate": 5.488077459582425e-06,
|
| 6473 |
+
"loss": 0.0074,
|
| 6474 |
+
"step": 919
|
| 6475 |
+
},
|
| 6476 |
+
{
|
| 6477 |
+
"epoch": 0.8574091332712023,
|
| 6478 |
+
"grad_norm": 0.9853683114051819,
|
| 6479 |
+
"learning_rate": 5.418347134565249e-06,
|
| 6480 |
+
"loss": 0.0151,
|
| 6481 |
+
"step": 920
|
| 6482 |
+
},
|
| 6483 |
+
{
|
| 6484 |
+
"epoch": 0.8583410997204101,
|
| 6485 |
+
"grad_norm": 0.6727381348609924,
|
| 6486 |
+
"learning_rate": 5.349037249814442e-06,
|
| 6487 |
+
"loss": 0.0114,
|
| 6488 |
+
"step": 921
|
| 6489 |
+
},
|
| 6490 |
+
{
|
| 6491 |
+
"epoch": 0.8592730661696178,
|
| 6492 |
+
"grad_norm": 1.942173719406128,
|
| 6493 |
+
"learning_rate": 5.28014845897708e-06,
|
| 6494 |
+
"loss": 0.0432,
|
| 6495 |
+
"step": 922
|
| 6496 |
+
},
|
| 6497 |
+
{
|
| 6498 |
+
"epoch": 0.8602050326188257,
|
| 6499 |
+
"grad_norm": 0.726831316947937,
|
| 6500 |
+
"learning_rate": 5.211681411728969e-06,
|
| 6501 |
+
"loss": 0.0217,
|
| 6502 |
+
"step": 923
|
| 6503 |
+
},
|
| 6504 |
+
{
|
| 6505 |
+
"epoch": 0.8611369990680335,
|
| 6506 |
+
"grad_norm": 1.3569886684417725,
|
| 6507 |
+
"learning_rate": 5.143636753768549e-06,
|
| 6508 |
+
"loss": 0.0124,
|
| 6509 |
+
"step": 924
|
| 6510 |
+
},
|
| 6511 |
+
{
|
| 6512 |
+
"epoch": 0.8620689655172413,
|
| 6513 |
+
"grad_norm": 1.4848052263259888,
|
| 6514 |
+
"learning_rate": 5.076015126810784e-06,
|
| 6515 |
+
"loss": 0.0913,
|
| 6516 |
+
"step": 925
|
| 6517 |
+
},
|
| 6518 |
+
{
|
| 6519 |
+
"epoch": 0.8630009319664492,
|
| 6520 |
+
"grad_norm": 0.08492793142795563,
|
| 6521 |
+
"learning_rate": 5.008817168581137e-06,
|
| 6522 |
+
"loss": 0.0021,
|
| 6523 |
+
"step": 926
|
| 6524 |
+
},
|
| 6525 |
+
{
|
| 6526 |
+
"epoch": 0.863932898415657,
|
| 6527 |
+
"grad_norm": 0.3844028115272522,
|
| 6528 |
+
"learning_rate": 4.94204351280953e-06,
|
| 6529 |
+
"loss": 0.0082,
|
| 6530 |
+
"step": 927
|
| 6531 |
+
},
|
| 6532 |
+
{
|
| 6533 |
+
"epoch": 0.8648648648648649,
|
| 6534 |
+
"grad_norm": 0.3236229717731476,
|
| 6535 |
+
"learning_rate": 4.8756947892243725e-06,
|
| 6536 |
+
"loss": 0.0026,
|
| 6537 |
+
"step": 928
|
| 6538 |
+
},
|
| 6539 |
+
{
|
| 6540 |
+
"epoch": 0.8657968313140727,
|
| 6541 |
+
"grad_norm": 0.13389159739017487,
|
| 6542 |
+
"learning_rate": 4.809771623546627e-06,
|
| 6543 |
+
"loss": 0.0024,
|
| 6544 |
+
"step": 929
|
| 6545 |
+
},
|
| 6546 |
+
{
|
| 6547 |
+
"epoch": 0.8667287977632805,
|
| 6548 |
+
"grad_norm": 0.4836329519748688,
|
| 6549 |
+
"learning_rate": 4.744274637483936e-06,
|
| 6550 |
+
"loss": 0.0089,
|
| 6551 |
+
"step": 930
|
| 6552 |
+
},
|
| 6553 |
+
{
|
| 6554 |
+
"epoch": 0.8676607642124884,
|
| 6555 |
+
"grad_norm": 0.7628198862075806,
|
| 6556 |
+
"learning_rate": 4.6792044487247e-06,
|
| 6557 |
+
"loss": 0.0091,
|
| 6558 |
+
"step": 931
|
| 6559 |
+
},
|
| 6560 |
+
{
|
| 6561 |
+
"epoch": 0.8685927306616962,
|
| 6562 |
+
"grad_norm": 0.9907669425010681,
|
| 6563 |
+
"learning_rate": 4.614561670932288e-06,
|
| 6564 |
+
"loss": 0.0322,
|
| 6565 |
+
"step": 932
|
| 6566 |
+
},
|
| 6567 |
+
{
|
| 6568 |
+
"epoch": 0.869524697110904,
|
| 6569 |
+
"grad_norm": 0.8017989993095398,
|
| 6570 |
+
"learning_rate": 4.550346913739256e-06,
|
| 6571 |
+
"loss": 0.0208,
|
| 6572 |
+
"step": 933
|
| 6573 |
+
},
|
| 6574 |
+
{
|
| 6575 |
+
"epoch": 0.8704566635601119,
|
| 6576 |
+
"grad_norm": 0.680810272693634,
|
| 6577 |
+
"learning_rate": 4.486560782741578e-06,
|
| 6578 |
+
"loss": 0.0122,
|
| 6579 |
+
"step": 934
|
| 6580 |
+
},
|
| 6581 |
+
{
|
| 6582 |
+
"epoch": 0.8713886300093197,
|
| 6583 |
+
"grad_norm": 2.3900279998779297,
|
| 6584 |
+
"learning_rate": 4.423203879492943e-06,
|
| 6585 |
+
"loss": 0.0444,
|
| 6586 |
+
"step": 935
|
| 6587 |
+
},
|
| 6588 |
+
{
|
| 6589 |
+
"epoch": 0.8723205964585274,
|
| 6590 |
+
"grad_norm": 0.30623510479927063,
|
| 6591 |
+
"learning_rate": 4.360276801499086e-06,
|
| 6592 |
+
"loss": 0.006,
|
| 6593 |
+
"step": 936
|
| 6594 |
+
},
|
| 6595 |
+
{
|
| 6596 |
+
"epoch": 0.8732525629077353,
|
| 6597 |
+
"grad_norm": 0.19743725657463074,
|
| 6598 |
+
"learning_rate": 4.29778014221216e-06,
|
| 6599 |
+
"loss": 0.004,
|
| 6600 |
+
"step": 937
|
| 6601 |
+
},
|
| 6602 |
+
{
|
| 6603 |
+
"epoch": 0.8741845293569431,
|
| 6604 |
+
"grad_norm": 0.06322675198316574,
|
| 6605 |
+
"learning_rate": 4.2357144910251e-06,
|
| 6606 |
+
"loss": 0.0012,
|
| 6607 |
+
"step": 938
|
| 6608 |
+
},
|
| 6609 |
+
{
|
| 6610 |
+
"epoch": 0.875116495806151,
|
| 6611 |
+
"grad_norm": 0.14598406851291656,
|
| 6612 |
+
"learning_rate": 4.174080433266136e-06,
|
| 6613 |
+
"loss": 0.0035,
|
| 6614 |
+
"step": 939
|
| 6615 |
+
},
|
| 6616 |
+
{
|
| 6617 |
+
"epoch": 0.8760484622553588,
|
| 6618 |
+
"grad_norm": 0.20462970435619354,
|
| 6619 |
+
"learning_rate": 4.112878550193194e-06,
|
| 6620 |
+
"loss": 0.0028,
|
| 6621 |
+
"step": 940
|
| 6622 |
+
},
|
| 6623 |
+
{
|
| 6624 |
+
"epoch": 0.8769804287045666,
|
| 6625 |
+
"grad_norm": 1.6550384759902954,
|
| 6626 |
+
"learning_rate": 4.05210941898847e-06,
|
| 6627 |
+
"loss": 0.0479,
|
| 6628 |
+
"step": 941
|
| 6629 |
+
},
|
| 6630 |
+
{
|
| 6631 |
+
"epoch": 0.8779123951537745,
|
| 6632 |
+
"grad_norm": 0.5954486131668091,
|
| 6633 |
+
"learning_rate": 3.991773612752952e-06,
|
| 6634 |
+
"loss": 0.0106,
|
| 6635 |
+
"step": 942
|
| 6636 |
+
},
|
| 6637 |
+
{
|
| 6638 |
+
"epoch": 0.8788443616029823,
|
| 6639 |
+
"grad_norm": 0.2953099310398102,
|
| 6640 |
+
"learning_rate": 3.9318717005010495e-06,
|
| 6641 |
+
"loss": 0.0049,
|
| 6642 |
+
"step": 943
|
| 6643 |
+
},
|
| 6644 |
+
{
|
| 6645 |
+
"epoch": 0.8797763280521901,
|
| 6646 |
+
"grad_norm": 0.3062129020690918,
|
| 6647 |
+
"learning_rate": 3.872404247155193e-06,
|
| 6648 |
+
"loss": 0.0024,
|
| 6649 |
+
"step": 944
|
| 6650 |
+
},
|
| 6651 |
+
{
|
| 6652 |
+
"epoch": 0.880708294501398,
|
| 6653 |
+
"grad_norm": 0.44278305768966675,
|
| 6654 |
+
"learning_rate": 3.8133718135405285e-06,
|
| 6655 |
+
"loss": 0.0084,
|
| 6656 |
+
"step": 945
|
| 6657 |
+
},
|
| 6658 |
+
{
|
| 6659 |
+
"epoch": 0.8816402609506058,
|
| 6660 |
+
"grad_norm": 0.15503761172294617,
|
| 6661 |
+
"learning_rate": 3.754774956379614e-06,
|
| 6662 |
+
"loss": 0.0025,
|
| 6663 |
+
"step": 946
|
| 6664 |
+
},
|
| 6665 |
+
{
|
| 6666 |
+
"epoch": 0.8825722273998136,
|
| 6667 |
+
"grad_norm": 0.15528859198093414,
|
| 6668 |
+
"learning_rate": 3.696614228287187e-06,
|
| 6669 |
+
"loss": 0.0022,
|
| 6670 |
+
"step": 947
|
| 6671 |
+
},
|
| 6672 |
+
{
|
| 6673 |
+
"epoch": 0.8835041938490215,
|
| 6674 |
+
"grad_norm": 0.3577142655849457,
|
| 6675 |
+
"learning_rate": 3.6388901777649477e-06,
|
| 6676 |
+
"loss": 0.006,
|
| 6677 |
+
"step": 948
|
| 6678 |
+
},
|
| 6679 |
+
{
|
| 6680 |
+
"epoch": 0.8844361602982292,
|
| 6681 |
+
"grad_norm": 1.3068500757217407,
|
| 6682 |
+
"learning_rate": 3.581603349196372e-06,
|
| 6683 |
+
"loss": 0.0105,
|
| 6684 |
+
"step": 949
|
| 6685 |
+
},
|
| 6686 |
+
{
|
| 6687 |
+
"epoch": 0.8853681267474371,
|
| 6688 |
+
"grad_norm": 1.0203229188919067,
|
| 6689 |
+
"learning_rate": 3.5247542828415747e-06,
|
| 6690 |
+
"loss": 0.0584,
|
| 6691 |
+
"step": 950
|
| 6692 |
+
},
|
| 6693 |
+
{
|
| 6694 |
+
"epoch": 0.8863000931966449,
|
| 6695 |
+
"grad_norm": 1.07167387008667,
|
| 6696 |
+
"learning_rate": 3.468343514832251e-06,
|
| 6697 |
+
"loss": 0.0127,
|
| 6698 |
+
"step": 951
|
| 6699 |
+
},
|
| 6700 |
+
{
|
| 6701 |
+
"epoch": 0.8872320596458527,
|
| 6702 |
+
"grad_norm": 0.17283280193805695,
|
| 6703 |
+
"learning_rate": 3.4123715771665786e-06,
|
| 6704 |
+
"loss": 0.0023,
|
| 6705 |
+
"step": 952
|
| 6706 |
+
},
|
| 6707 |
+
{
|
| 6708 |
+
"epoch": 0.8881640260950606,
|
| 6709 |
+
"grad_norm": 0.06770434975624084,
|
| 6710 |
+
"learning_rate": 3.356838997704226e-06,
|
| 6711 |
+
"loss": 0.0018,
|
| 6712 |
+
"step": 953
|
| 6713 |
+
},
|
| 6714 |
+
{
|
| 6715 |
+
"epoch": 0.8890959925442684,
|
| 6716 |
+
"grad_norm": 0.20179446041584015,
|
| 6717 |
+
"learning_rate": 3.3017463001613626e-06,
|
| 6718 |
+
"loss": 0.0036,
|
| 6719 |
+
"step": 954
|
| 6720 |
+
},
|
| 6721 |
+
{
|
| 6722 |
+
"epoch": 0.8900279589934762,
|
| 6723 |
+
"grad_norm": 1.307664394378662,
|
| 6724 |
+
"learning_rate": 3.2470940041057106e-06,
|
| 6725 |
+
"loss": 0.0415,
|
| 6726 |
+
"step": 955
|
| 6727 |
+
},
|
| 6728 |
+
{
|
| 6729 |
+
"epoch": 0.8909599254426841,
|
| 6730 |
+
"grad_norm": 0.3763882517814636,
|
| 6731 |
+
"learning_rate": 3.1928826249516987e-06,
|
| 6732 |
+
"loss": 0.008,
|
| 6733 |
+
"step": 956
|
| 6734 |
+
},
|
| 6735 |
+
{
|
| 6736 |
+
"epoch": 0.8918918918918919,
|
| 6737 |
+
"grad_norm": 0.4271828532218933,
|
| 6738 |
+
"learning_rate": 3.139112673955513e-06,
|
| 6739 |
+
"loss": 0.0088,
|
| 6740 |
+
"step": 957
|
| 6741 |
+
},
|
| 6742 |
+
{
|
| 6743 |
+
"epoch": 0.8928238583410997,
|
| 6744 |
+
"grad_norm": 0.5596781969070435,
|
| 6745 |
+
"learning_rate": 3.08578465821035e-06,
|
| 6746 |
+
"loss": 0.0064,
|
| 6747 |
+
"step": 958
|
| 6748 |
+
},
|
| 6749 |
+
{
|
| 6750 |
+
"epoch": 0.8937558247903076,
|
| 6751 |
+
"grad_norm": 0.9955370426177979,
|
| 6752 |
+
"learning_rate": 3.0328990806415934e-06,
|
| 6753 |
+
"loss": 0.0293,
|
| 6754 |
+
"step": 959
|
| 6755 |
+
},
|
| 6756 |
+
{
|
| 6757 |
+
"epoch": 0.8946877912395154,
|
| 6758 |
+
"grad_norm": 0.6744177341461182,
|
| 6759 |
+
"learning_rate": 2.9804564400020994e-06,
|
| 6760 |
+
"loss": 0.0115,
|
| 6761 |
+
"step": 960
|
| 6762 |
+
},
|
| 6763 |
+
{
|
| 6764 |
+
"epoch": 0.8956197576887233,
|
| 6765 |
+
"grad_norm": 1.0342707633972168,
|
| 6766 |
+
"learning_rate": 2.9284572308674596e-06,
|
| 6767 |
+
"loss": 0.0688,
|
| 6768 |
+
"step": 961
|
| 6769 |
+
},
|
| 6770 |
+
{
|
| 6771 |
+
"epoch": 0.896551724137931,
|
| 6772 |
+
"grad_norm": 0.36630138754844666,
|
| 6773 |
+
"learning_rate": 2.8769019436313715e-06,
|
| 6774 |
+
"loss": 0.0069,
|
| 6775 |
+
"step": 962
|
| 6776 |
+
},
|
| 6777 |
+
{
|
| 6778 |
+
"epoch": 0.8974836905871388,
|
| 6779 |
+
"grad_norm": 0.9983659982681274,
|
| 6780 |
+
"learning_rate": 2.825791064500993e-06,
|
| 6781 |
+
"loss": 0.032,
|
| 6782 |
+
"step": 963
|
| 6783 |
+
},
|
| 6784 |
+
{
|
| 6785 |
+
"epoch": 0.8984156570363467,
|
| 6786 |
+
"grad_norm": 0.27666184306144714,
|
| 6787 |
+
"learning_rate": 2.7751250754923573e-06,
|
| 6788 |
+
"loss": 0.0053,
|
| 6789 |
+
"step": 964
|
| 6790 |
+
},
|
| 6791 |
+
{
|
| 6792 |
+
"epoch": 0.8993476234855545,
|
| 6793 |
+
"grad_norm": 0.03350253403186798,
|
| 6794 |
+
"learning_rate": 2.7249044544258363e-06,
|
| 6795 |
+
"loss": 0.0009,
|
| 6796 |
+
"step": 965
|
| 6797 |
+
},
|
| 6798 |
+
{
|
| 6799 |
+
"epoch": 0.9002795899347623,
|
| 6800 |
+
"grad_norm": 0.501646876335144,
|
| 6801 |
+
"learning_rate": 2.675129674921639e-06,
|
| 6802 |
+
"loss": 0.0064,
|
| 6803 |
+
"step": 966
|
| 6804 |
+
},
|
| 6805 |
+
{
|
| 6806 |
+
"epoch": 0.9012115563839702,
|
| 6807 |
+
"grad_norm": 0.07862615585327148,
|
| 6808 |
+
"learning_rate": 2.625801206395312e-06,
|
| 6809 |
+
"loss": 0.0015,
|
| 6810 |
+
"step": 967
|
| 6811 |
+
},
|
| 6812 |
+
{
|
| 6813 |
+
"epoch": 0.902143522833178,
|
| 6814 |
+
"grad_norm": 0.849259078502655,
|
| 6815 |
+
"learning_rate": 2.576919514053355e-06,
|
| 6816 |
+
"loss": 0.0202,
|
| 6817 |
+
"step": 968
|
| 6818 |
+
},
|
| 6819 |
+
{
|
| 6820 |
+
"epoch": 0.9030754892823858,
|
| 6821 |
+
"grad_norm": 0.0796375498175621,
|
| 6822 |
+
"learning_rate": 2.528485058888813e-06,
|
| 6823 |
+
"loss": 0.0017,
|
| 6824 |
+
"step": 969
|
| 6825 |
+
},
|
| 6826 |
+
{
|
| 6827 |
+
"epoch": 0.9040074557315937,
|
| 6828 |
+
"grad_norm": 1.963661789894104,
|
| 6829 |
+
"learning_rate": 2.48049829767692e-06,
|
| 6830 |
+
"loss": 0.0798,
|
| 6831 |
+
"step": 970
|
| 6832 |
+
},
|
| 6833 |
+
{
|
| 6834 |
+
"epoch": 0.9049394221808015,
|
| 6835 |
+
"grad_norm": 0.7925097346305847,
|
| 6836 |
+
"learning_rate": 2.4329596829708144e-06,
|
| 6837 |
+
"loss": 0.0099,
|
| 6838 |
+
"step": 971
|
| 6839 |
+
},
|
| 6840 |
+
{
|
| 6841 |
+
"epoch": 0.9058713886300093,
|
| 6842 |
+
"grad_norm": 0.11906258761882782,
|
| 6843 |
+
"learning_rate": 2.385869663097251e-06,
|
| 6844 |
+
"loss": 0.002,
|
| 6845 |
+
"step": 972
|
| 6846 |
+
},
|
| 6847 |
+
{
|
| 6848 |
+
"epoch": 0.9068033550792172,
|
| 6849 |
+
"grad_norm": 1.8423750400543213,
|
| 6850 |
+
"learning_rate": 2.3392286821523724e-06,
|
| 6851 |
+
"loss": 0.0678,
|
| 6852 |
+
"step": 973
|
| 6853 |
+
},
|
| 6854 |
+
{
|
| 6855 |
+
"epoch": 0.907735321528425,
|
| 6856 |
+
"grad_norm": 0.6542656421661377,
|
| 6857 |
+
"learning_rate": 2.2930371799975594e-06,
|
| 6858 |
+
"loss": 0.0111,
|
| 6859 |
+
"step": 974
|
| 6860 |
+
},
|
| 6861 |
+
{
|
| 6862 |
+
"epoch": 0.9086672879776329,
|
| 6863 |
+
"grad_norm": 1.2632553577423096,
|
| 6864 |
+
"learning_rate": 2.2472955922552163e-06,
|
| 6865 |
+
"loss": 0.0662,
|
| 6866 |
+
"step": 975
|
| 6867 |
+
},
|
| 6868 |
+
{
|
| 6869 |
+
"epoch": 0.9095992544268406,
|
| 6870 |
+
"grad_norm": 0.9758872985839844,
|
| 6871 |
+
"learning_rate": 2.202004350304715e-06,
|
| 6872 |
+
"loss": 0.0104,
|
| 6873 |
+
"step": 976
|
| 6874 |
+
},
|
| 6875 |
+
{
|
| 6876 |
+
"epoch": 0.9105312208760484,
|
| 6877 |
+
"grad_norm": 0.615419864654541,
|
| 6878 |
+
"learning_rate": 2.157163881278312e-06,
|
| 6879 |
+
"loss": 0.0131,
|
| 6880 |
+
"step": 977
|
| 6881 |
+
},
|
| 6882 |
+
{
|
| 6883 |
+
"epoch": 0.9114631873252563,
|
| 6884 |
+
"grad_norm": 0.07674716413021088,
|
| 6885 |
+
"learning_rate": 2.112774608057111e-06,
|
| 6886 |
+
"loss": 0.001,
|
| 6887 |
+
"step": 978
|
| 6888 |
+
},
|
| 6889 |
+
{
|
| 6890 |
+
"epoch": 0.9123951537744641,
|
| 6891 |
+
"grad_norm": 0.2658063471317291,
|
| 6892 |
+
"learning_rate": 2.068836949267089e-06,
|
| 6893 |
+
"loss": 0.0017,
|
| 6894 |
+
"step": 979
|
| 6895 |
+
},
|
| 6896 |
+
{
|
| 6897 |
+
"epoch": 0.9133271202236719,
|
| 6898 |
+
"grad_norm": 1.0758992433547974,
|
| 6899 |
+
"learning_rate": 2.0253513192751373e-06,
|
| 6900 |
+
"loss": 0.0118,
|
| 6901 |
+
"step": 980
|
| 6902 |
+
},
|
| 6903 |
+
{
|
| 6904 |
+
"epoch": 0.9142590866728798,
|
| 6905 |
+
"grad_norm": 0.5867937803268433,
|
| 6906 |
+
"learning_rate": 1.982318128185151e-06,
|
| 6907 |
+
"loss": 0.0053,
|
| 6908 |
+
"step": 981
|
| 6909 |
+
},
|
| 6910 |
+
{
|
| 6911 |
+
"epoch": 0.9151910531220876,
|
| 6912 |
+
"grad_norm": 0.7973155379295349,
|
| 6913 |
+
"learning_rate": 1.9397377818341944e-06,
|
| 6914 |
+
"loss": 0.0145,
|
| 6915 |
+
"step": 982
|
| 6916 |
+
},
|
| 6917 |
+
{
|
| 6918 |
+
"epoch": 0.9161230195712954,
|
| 6919 |
+
"grad_norm": 0.16484799981117249,
|
| 6920 |
+
"learning_rate": 1.8976106817886196e-06,
|
| 6921 |
+
"loss": 0.0033,
|
| 6922 |
+
"step": 983
|
| 6923 |
+
},
|
| 6924 |
+
{
|
| 6925 |
+
"epoch": 0.9170549860205033,
|
| 6926 |
+
"grad_norm": 0.21485017240047455,
|
| 6927 |
+
"learning_rate": 1.855937225340315e-06,
|
| 6928 |
+
"loss": 0.0038,
|
| 6929 |
+
"step": 984
|
| 6930 |
+
},
|
| 6931 |
+
{
|
| 6932 |
+
"epoch": 0.9179869524697111,
|
| 6933 |
+
"grad_norm": 0.0540773943066597,
|
| 6934 |
+
"learning_rate": 1.8147178055029579e-06,
|
| 6935 |
+
"loss": 0.0012,
|
| 6936 |
+
"step": 985
|
| 6937 |
+
},
|
| 6938 |
+
{
|
| 6939 |
+
"epoch": 0.918918918918919,
|
| 6940 |
+
"grad_norm": 0.10457666218280792,
|
| 6941 |
+
"learning_rate": 1.7739528110083004e-06,
|
| 6942 |
+
"loss": 0.0014,
|
| 6943 |
+
"step": 986
|
| 6944 |
+
},
|
| 6945 |
+
{
|
| 6946 |
+
"epoch": 0.9198508853681268,
|
| 6947 |
+
"grad_norm": 0.07089216262102127,
|
| 6948 |
+
"learning_rate": 1.7336426263024897e-06,
|
| 6949 |
+
"loss": 0.0012,
|
| 6950 |
+
"step": 987
|
| 6951 |
+
},
|
| 6952 |
+
{
|
| 6953 |
+
"epoch": 0.9207828518173345,
|
| 6954 |
+
"grad_norm": 0.041863564401865005,
|
| 6955 |
+
"learning_rate": 1.6937876315424705e-06,
|
| 6956 |
+
"loss": 0.0008,
|
| 6957 |
+
"step": 988
|
| 6958 |
+
},
|
| 6959 |
+
{
|
| 6960 |
+
"epoch": 0.9217148182665424,
|
| 6961 |
+
"grad_norm": 0.1343851238489151,
|
| 6962 |
+
"learning_rate": 1.6543882025923886e-06,
|
| 6963 |
+
"loss": 0.0024,
|
| 6964 |
+
"step": 989
|
| 6965 |
+
},
|
| 6966 |
+
{
|
| 6967 |
+
"epoch": 0.9226467847157502,
|
| 6968 |
+
"grad_norm": 0.04936739057302475,
|
| 6969 |
+
"learning_rate": 1.6154447110200255e-06,
|
| 6970 |
+
"loss": 0.0008,
|
| 6971 |
+
"step": 990
|
| 6972 |
+
},
|
| 6973 |
+
{
|
| 6974 |
+
"epoch": 0.923578751164958,
|
| 6975 |
+
"grad_norm": 0.6000242233276367,
|
| 6976 |
+
"learning_rate": 1.5769575240933422e-06,
|
| 6977 |
+
"loss": 0.0095,
|
| 6978 |
+
"step": 991
|
| 6979 |
+
},
|
| 6980 |
+
{
|
| 6981 |
+
"epoch": 0.9245107176141659,
|
| 6982 |
+
"grad_norm": 0.7469452619552612,
|
| 6983 |
+
"learning_rate": 1.5389270047769578e-06,
|
| 6984 |
+
"loss": 0.0163,
|
| 6985 |
+
"step": 992
|
| 6986 |
+
},
|
| 6987 |
+
{
|
| 6988 |
+
"epoch": 0.9254426840633737,
|
| 6989 |
+
"grad_norm": 0.09905730932950974,
|
| 6990 |
+
"learning_rate": 1.5013535117287647e-06,
|
| 6991 |
+
"loss": 0.0014,
|
| 6992 |
+
"step": 993
|
| 6993 |
+
},
|
| 6994 |
+
{
|
| 6995 |
+
"epoch": 0.9263746505125815,
|
| 6996 |
+
"grad_norm": 0.9133400917053223,
|
| 6997 |
+
"learning_rate": 1.4642373992965364e-06,
|
| 6998 |
+
"loss": 0.0117,
|
| 6999 |
+
"step": 994
|
| 7000 |
+
},
|
| 7001 |
+
{
|
| 7002 |
+
"epoch": 0.9273066169617894,
|
| 7003 |
+
"grad_norm": 1.5601509809494019,
|
| 7004 |
+
"learning_rate": 1.4275790175145908e-06,
|
| 7005 |
+
"loss": 0.0671,
|
| 7006 |
+
"step": 995
|
| 7007 |
+
},
|
| 7008 |
+
{
|
| 7009 |
+
"epoch": 0.9282385834109972,
|
| 7010 |
+
"grad_norm": 1.1153135299682617,
|
| 7011 |
+
"learning_rate": 1.3913787121004719e-06,
|
| 7012 |
+
"loss": 0.0184,
|
| 7013 |
+
"step": 996
|
| 7014 |
+
},
|
| 7015 |
+
{
|
| 7016 |
+
"epoch": 0.9291705498602051,
|
| 7017 |
+
"grad_norm": 1.7211925983428955,
|
| 7018 |
+
"learning_rate": 1.3556368244517114e-06,
|
| 7019 |
+
"loss": 0.0227,
|
| 7020 |
+
"step": 997
|
| 7021 |
+
},
|
| 7022 |
+
{
|
| 7023 |
+
"epoch": 0.9301025163094129,
|
| 7024 |
+
"grad_norm": 0.985527753829956,
|
| 7025 |
+
"learning_rate": 1.3203536916425841e-06,
|
| 7026 |
+
"loss": 0.0273,
|
| 7027 |
+
"step": 998
|
| 7028 |
+
},
|
| 7029 |
+
{
|
| 7030 |
+
"epoch": 0.9310344827586207,
|
| 7031 |
+
"grad_norm": 1.213148832321167,
|
| 7032 |
+
"learning_rate": 1.2855296464209688e-06,
|
| 7033 |
+
"loss": 0.0441,
|
| 7034 |
+
"step": 999
|
| 7035 |
+
},
|
| 7036 |
+
{
|
| 7037 |
+
"epoch": 0.9319664492078286,
|
| 7038 |
+
"grad_norm": 1.5918529033660889,
|
| 7039 |
+
"learning_rate": 1.2511650172051637e-06,
|
| 7040 |
+
"loss": 0.0944,
|
| 7041 |
+
"step": 1000
|
| 7042 |
+
},
|
| 7043 |
+
{
|
| 7044 |
+
"epoch": 0.9328984156570364,
|
| 7045 |
+
"grad_norm": 0.06392987817525864,
|
| 7046 |
+
"learning_rate": 1.217260128080816e-06,
|
| 7047 |
+
"loss": 0.0012,
|
| 7048 |
+
"step": 1001
|
| 7049 |
+
},
|
| 7050 |
+
{
|
| 7051 |
+
"epoch": 0.9338303821062441,
|
| 7052 |
+
"grad_norm": 1.1391026973724365,
|
| 7053 |
+
"learning_rate": 1.1838152987978578e-06,
|
| 7054 |
+
"loss": 0.0308,
|
| 7055 |
+
"step": 1002
|
| 7056 |
+
},
|
| 7057 |
+
{
|
| 7058 |
+
"epoch": 0.934762348555452,
|
| 7059 |
+
"grad_norm": 0.05295366048812866,
|
| 7060 |
+
"learning_rate": 1.1508308447674975e-06,
|
| 7061 |
+
"loss": 0.0008,
|
| 7062 |
+
"step": 1003
|
| 7063 |
+
},
|
| 7064 |
+
{
|
| 7065 |
+
"epoch": 0.9356943150046598,
|
| 7066 |
+
"grad_norm": 1.0572638511657715,
|
| 7067 |
+
"learning_rate": 1.1183070770592441e-06,
|
| 7068 |
+
"loss": 0.0262,
|
| 7069 |
+
"step": 1004
|
| 7070 |
+
},
|
| 7071 |
+
{
|
| 7072 |
+
"epoch": 0.9366262814538676,
|
| 7073 |
+
"grad_norm": 0.37860023975372314,
|
| 7074 |
+
"learning_rate": 1.0862443023979651e-06,
|
| 7075 |
+
"loss": 0.0069,
|
| 7076 |
+
"step": 1005
|
| 7077 |
+
},
|
| 7078 |
+
{
|
| 7079 |
+
"epoch": 0.9375582479030755,
|
| 7080 |
+
"grad_norm": 0.15000490844249725,
|
| 7081 |
+
"learning_rate": 1.0546428231609895e-06,
|
| 7082 |
+
"loss": 0.0035,
|
| 7083 |
+
"step": 1006
|
| 7084 |
+
},
|
| 7085 |
+
{
|
| 7086 |
+
"epoch": 0.9384902143522833,
|
| 7087 |
+
"grad_norm": 0.05710078775882721,
|
| 7088 |
+
"learning_rate": 1.0235029373752758e-06,
|
| 7089 |
+
"loss": 0.001,
|
| 7090 |
+
"step": 1007
|
| 7091 |
+
},
|
| 7092 |
+
{
|
| 7093 |
+
"epoch": 0.9394221808014911,
|
| 7094 |
+
"grad_norm": 0.5820727348327637,
|
| 7095 |
+
"learning_rate": 9.928249387145982e-07,
|
| 7096 |
+
"loss": 0.0101,
|
| 7097 |
+
"step": 1008
|
| 7098 |
+
},
|
| 7099 |
+
{
|
| 7100 |
+
"epoch": 0.940354147250699,
|
| 7101 |
+
"grad_norm": 0.19106589257717133,
|
| 7102 |
+
"learning_rate": 9.6260911649676e-07,
|
| 7103 |
+
"loss": 0.0041,
|
| 7104 |
+
"step": 1009
|
| 7105 |
+
},
|
| 7106 |
+
{
|
| 7107 |
+
"epoch": 0.9412861136999068,
|
| 7108 |
+
"grad_norm": 0.3644753396511078,
|
| 7109 |
+
"learning_rate": 9.32855755680867e-07,
|
| 7110 |
+
"loss": 0.0098,
|
| 7111 |
+
"step": 1010
|
| 7112 |
+
},
|
| 7113 |
+
{
|
| 7114 |
+
"epoch": 0.9422180801491147,
|
| 7115 |
+
"grad_norm": 0.2526366114616394,
|
| 7116 |
+
"learning_rate": 9.035651368646648e-07,
|
| 7117 |
+
"loss": 0.0055,
|
| 7118 |
+
"step": 1011
|
| 7119 |
+
},
|
| 7120 |
+
{
|
| 7121 |
+
"epoch": 0.9431500465983225,
|
| 7122 |
+
"grad_norm": 0.6924790143966675,
|
| 7123 |
+
"learning_rate": 8.747375362818666e-07,
|
| 7124 |
+
"loss": 0.0188,
|
| 7125 |
+
"step": 1012
|
| 7126 |
+
},
|
| 7127 |
+
{
|
| 7128 |
+
"epoch": 0.9440820130475303,
|
| 7129 |
+
"grad_norm": 1.0197707414627075,
|
| 7130 |
+
"learning_rate": 8.463732257995571e-07,
|
| 7131 |
+
"loss": 0.0265,
|
| 7132 |
+
"step": 1013
|
| 7133 |
+
},
|
| 7134 |
+
{
|
| 7135 |
+
"epoch": 0.9450139794967382,
|
| 7136 |
+
"grad_norm": 0.5194602608680725,
|
| 7137 |
+
"learning_rate": 8.184724729156379e-07,
|
| 7138 |
+
"loss": 0.0094,
|
| 7139 |
+
"step": 1014
|
| 7140 |
+
},
|
| 7141 |
+
{
|
| 7142 |
+
"epoch": 0.9459459459459459,
|
| 7143 |
+
"grad_norm": 0.5754898190498352,
|
| 7144 |
+
"learning_rate": 7.910355407562742e-07,
|
| 7145 |
+
"loss": 0.0105,
|
| 7146 |
+
"step": 1015
|
| 7147 |
+
},
|
| 7148 |
+
{
|
| 7149 |
+
"epoch": 0.9468779123951537,
|
| 7150 |
+
"grad_norm": 0.28730660676956177,
|
| 7151 |
+
"learning_rate": 7.640626880734581e-07,
|
| 7152 |
+
"loss": 0.0034,
|
| 7153 |
+
"step": 1016
|
| 7154 |
+
},
|
| 7155 |
+
{
|
| 7156 |
+
"epoch": 0.9478098788443616,
|
| 7157 |
+
"grad_norm": 1.0178496837615967,
|
| 7158 |
+
"learning_rate": 7.375541692425326e-07,
|
| 7159 |
+
"loss": 0.0199,
|
| 7160 |
+
"step": 1017
|
| 7161 |
+
},
|
| 7162 |
+
{
|
| 7163 |
+
"epoch": 0.9487418452935694,
|
| 7164 |
+
"grad_norm": 0.2701060473918915,
|
| 7165 |
+
"learning_rate": 7.115102342598101e-07,
|
| 7166 |
+
"loss": 0.003,
|
| 7167 |
+
"step": 1018
|
| 7168 |
+
},
|
| 7169 |
+
{
|
| 7170 |
+
"epoch": 0.9496738117427772,
|
| 7171 |
+
"grad_norm": 0.5945216417312622,
|
| 7172 |
+
"learning_rate": 6.859311287402081e-07,
|
| 7173 |
+
"loss": 0.0105,
|
| 7174 |
+
"step": 1019
|
| 7175 |
+
},
|
| 7176 |
+
{
|
| 7177 |
+
"epoch": 0.9506057781919851,
|
| 7178 |
+
"grad_norm": 0.7214066982269287,
|
| 7179 |
+
"learning_rate": 6.608170939149283e-07,
|
| 7180 |
+
"loss": 0.0132,
|
| 7181 |
+
"step": 1020
|
| 7182 |
+
},
|
| 7183 |
+
{
|
| 7184 |
+
"epoch": 0.9515377446411929,
|
| 7185 |
+
"grad_norm": 0.11991623789072037,
|
| 7186 |
+
"learning_rate": 6.361683666291973e-07,
|
| 7187 |
+
"loss": 0.0031,
|
| 7188 |
+
"step": 1021
|
| 7189 |
+
},
|
| 7190 |
+
{
|
| 7191 |
+
"epoch": 0.9524697110904008,
|
| 7192 |
+
"grad_norm": 0.24525563418865204,
|
| 7193 |
+
"learning_rate": 6.119851793400189e-07,
|
| 7194 |
+
"loss": 0.0026,
|
| 7195 |
+
"step": 1022
|
| 7196 |
+
},
|
| 7197 |
+
{
|
| 7198 |
+
"epoch": 0.9534016775396086,
|
| 7199 |
+
"grad_norm": 1.5672634840011597,
|
| 7200 |
+
"learning_rate": 5.882677601139919e-07,
|
| 7201 |
+
"loss": 0.0275,
|
| 7202 |
+
"step": 1023
|
| 7203 |
+
},
|
| 7204 |
+
{
|
| 7205 |
+
"epoch": 0.9543336439888164,
|
| 7206 |
+
"grad_norm": 0.9412722587585449,
|
| 7207 |
+
"learning_rate": 5.650163326251345e-07,
|
| 7208 |
+
"loss": 0.0178,
|
| 7209 |
+
"step": 1024
|
| 7210 |
+
},
|
| 7211 |
+
{
|
| 7212 |
+
"epoch": 0.9552656104380243,
|
| 7213 |
+
"grad_norm": 0.39089635014533997,
|
| 7214 |
+
"learning_rate": 5.422311161528193e-07,
|
| 7215 |
+
"loss": 0.0224,
|
| 7216 |
+
"step": 1025
|
| 7217 |
+
},
|
| 7218 |
+
{
|
| 7219 |
+
"epoch": 0.9561975768872321,
|
| 7220 |
+
"grad_norm": 0.03957606479525566,
|
| 7221 |
+
"learning_rate": 5.199123255796634e-07,
|
| 7222 |
+
"loss": 0.0009,
|
| 7223 |
+
"step": 1026
|
| 7224 |
+
},
|
| 7225 |
+
{
|
| 7226 |
+
"epoch": 0.9571295433364398,
|
| 7227 |
+
"grad_norm": 1.1148029565811157,
|
| 7228 |
+
"learning_rate": 4.980601713895305e-07,
|
| 7229 |
+
"loss": 0.023,
|
| 7230 |
+
"step": 1027
|
| 7231 |
+
},
|
| 7232 |
+
{
|
| 7233 |
+
"epoch": 0.9580615097856477,
|
| 7234 |
+
"grad_norm": 0.12970255315303802,
|
| 7235 |
+
"learning_rate": 4.7667485966552685e-07,
|
| 7236 |
+
"loss": 0.0021,
|
| 7237 |
+
"step": 1028
|
| 7238 |
+
},
|
| 7239 |
+
{
|
| 7240 |
+
"epoch": 0.9589934762348555,
|
| 7241 |
+
"grad_norm": 0.9582773447036743,
|
| 7242 |
+
"learning_rate": 4.55756592088058e-07,
|
| 7243 |
+
"loss": 0.0134,
|
| 7244 |
+
"step": 1029
|
| 7245 |
+
},
|
| 7246 |
+
{
|
| 7247 |
+
"epoch": 0.9599254426840633,
|
| 7248 |
+
"grad_norm": 0.3440712094306946,
|
| 7249 |
+
"learning_rate": 4.353055659329419e-07,
|
| 7250 |
+
"loss": 0.0089,
|
| 7251 |
+
"step": 1030
|
| 7252 |
+
},
|
| 7253 |
+
{
|
| 7254 |
+
"epoch": 0.9608574091332712,
|
| 7255 |
+
"grad_norm": 0.0910550057888031,
|
| 7256 |
+
"learning_rate": 4.153219740695435e-07,
|
| 7257 |
+
"loss": 0.0022,
|
| 7258 |
+
"step": 1031
|
| 7259 |
+
},
|
| 7260 |
+
{
|
| 7261 |
+
"epoch": 0.961789375582479,
|
| 7262 |
+
"grad_norm": 0.5724707841873169,
|
| 7263 |
+
"learning_rate": 3.958060049589485e-07,
|
| 7264 |
+
"loss": 0.008,
|
| 7265 |
+
"step": 1032
|
| 7266 |
+
},
|
| 7267 |
+
{
|
| 7268 |
+
"epoch": 0.9627213420316869,
|
| 7269 |
+
"grad_norm": 0.7248005867004395,
|
| 7270 |
+
"learning_rate": 3.767578426521923e-07,
|
| 7271 |
+
"loss": 0.0209,
|
| 7272 |
+
"step": 1033
|
| 7273 |
+
},
|
| 7274 |
+
{
|
| 7275 |
+
"epoch": 0.9636533084808947,
|
| 7276 |
+
"grad_norm": 1.4421683549880981,
|
| 7277 |
+
"learning_rate": 3.5817766678850615e-07,
|
| 7278 |
+
"loss": 0.0059,
|
| 7279 |
+
"step": 1034
|
| 7280 |
+
},
|
| 7281 |
+
{
|
| 7282 |
+
"epoch": 0.9645852749301025,
|
| 7283 |
+
"grad_norm": 0.555047869682312,
|
| 7284 |
+
"learning_rate": 3.4006565259366295e-07,
|
| 7285 |
+
"loss": 0.0095,
|
| 7286 |
+
"step": 1035
|
| 7287 |
+
},
|
| 7288 |
+
{
|
| 7289 |
+
"epoch": 0.9655172413793104,
|
| 7290 |
+
"grad_norm": 0.6058024764060974,
|
| 7291 |
+
"learning_rate": 3.224219708782894e-07,
|
| 7292 |
+
"loss": 0.0113,
|
| 7293 |
+
"step": 1036
|
| 7294 |
+
},
|
| 7295 |
+
{
|
| 7296 |
+
"epoch": 0.9664492078285182,
|
| 7297 |
+
"grad_norm": 0.9548462629318237,
|
| 7298 |
+
"learning_rate": 3.052467880362675e-07,
|
| 7299 |
+
"loss": 0.0218,
|
| 7300 |
+
"step": 1037
|
| 7301 |
+
},
|
| 7302 |
+
{
|
| 7303 |
+
"epoch": 0.967381174277726,
|
| 7304 |
+
"grad_norm": 0.2721622586250305,
|
| 7305 |
+
"learning_rate": 2.88540266043158e-07,
|
| 7306 |
+
"loss": 0.0058,
|
| 7307 |
+
"step": 1038
|
| 7308 |
+
},
|
| 7309 |
+
{
|
| 7310 |
+
"epoch": 0.9683131407269339,
|
| 7311 |
+
"grad_norm": 0.2145017385482788,
|
| 7312 |
+
"learning_rate": 2.723025624546849e-07,
|
| 7313 |
+
"loss": 0.0034,
|
| 7314 |
+
"step": 1039
|
| 7315 |
+
},
|
| 7316 |
+
{
|
| 7317 |
+
"epoch": 0.9692451071761417,
|
| 7318 |
+
"grad_norm": 0.4106197655200958,
|
| 7319 |
+
"learning_rate": 2.5653383040524227e-07,
|
| 7320 |
+
"loss": 0.0055,
|
| 7321 |
+
"step": 1040
|
| 7322 |
+
},
|
| 7323 |
+
{
|
| 7324 |
+
"epoch": 0.9701770736253494,
|
| 7325 |
+
"grad_norm": 0.16109812259674072,
|
| 7326 |
+
"learning_rate": 2.412342186064565e-07,
|
| 7327 |
+
"loss": 0.0024,
|
| 7328 |
+
"step": 1041
|
| 7329 |
+
},
|
| 7330 |
+
{
|
| 7331 |
+
"epoch": 0.9711090400745573,
|
| 7332 |
+
"grad_norm": 1.1901835203170776,
|
| 7333 |
+
"learning_rate": 2.2640387134577058e-07,
|
| 7334 |
+
"loss": 0.0369,
|
| 7335 |
+
"step": 1042
|
| 7336 |
+
},
|
| 7337 |
+
{
|
| 7338 |
+
"epoch": 0.9720410065237651,
|
| 7339 |
+
"grad_norm": 0.8408852815628052,
|
| 7340 |
+
"learning_rate": 2.1204292848509555e-07,
|
| 7341 |
+
"loss": 0.019,
|
| 7342 |
+
"step": 1043
|
| 7343 |
+
},
|
| 7344 |
+
{
|
| 7345 |
+
"epoch": 0.972972972972973,
|
| 7346 |
+
"grad_norm": 0.14452603459358215,
|
| 7347 |
+
"learning_rate": 1.98151525459489e-07,
|
| 7348 |
+
"loss": 0.0023,
|
| 7349 |
+
"step": 1044
|
| 7350 |
+
},
|
| 7351 |
+
{
|
| 7352 |
+
"epoch": 0.9739049394221808,
|
| 7353 |
+
"grad_norm": 0.34769976139068604,
|
| 7354 |
+
"learning_rate": 1.8472979327587292e-07,
|
| 7355 |
+
"loss": 0.0066,
|
| 7356 |
+
"step": 1045
|
| 7357 |
+
},
|
| 7358 |
+
{
|
| 7359 |
+
"epoch": 0.9748369058713886,
|
| 7360 |
+
"grad_norm": 0.8973527550697327,
|
| 7361 |
+
"learning_rate": 1.717778585118013e-07,
|
| 7362 |
+
"loss": 0.0171,
|
| 7363 |
+
"step": 1046
|
| 7364 |
+
},
|
| 7365 |
+
{
|
| 7366 |
+
"epoch": 0.9757688723205965,
|
| 7367 |
+
"grad_norm": 0.06965801864862442,
|
| 7368 |
+
"learning_rate": 1.5929584331427216e-07,
|
| 7369 |
+
"loss": 0.0011,
|
| 7370 |
+
"step": 1047
|
| 7371 |
+
},
|
| 7372 |
+
{
|
| 7373 |
+
"epoch": 0.9767008387698043,
|
| 7374 |
+
"grad_norm": 0.5106223225593567,
|
| 7375 |
+
"learning_rate": 1.4728386539856754e-07,
|
| 7376 |
+
"loss": 0.0086,
|
| 7377 |
+
"step": 1048
|
| 7378 |
+
},
|
| 7379 |
+
{
|
| 7380 |
+
"epoch": 0.9776328052190121,
|
| 7381 |
+
"grad_norm": 1.0167614221572876,
|
| 7382 |
+
"learning_rate": 1.3574203804713747e-07,
|
| 7383 |
+
"loss": 0.0193,
|
| 7384 |
+
"step": 1049
|
| 7385 |
+
},
|
| 7386 |
+
{
|
| 7387 |
+
"epoch": 0.97856477166822,
|
| 7388 |
+
"grad_norm": 1.6071141958236694,
|
| 7389 |
+
"learning_rate": 1.246704701085566e-07,
|
| 7390 |
+
"loss": 0.0923,
|
| 7391 |
+
"step": 1050
|
| 7392 |
+
},
|
| 7393 |
+
{
|
| 7394 |
+
"epoch": 0.9794967381174278,
|
| 7395 |
+
"grad_norm": 0.06256900727748871,
|
| 7396 |
+
"learning_rate": 1.1406926599646373e-07,
|
| 7397 |
+
"loss": 0.001,
|
| 7398 |
+
"step": 1051
|
| 7399 |
+
},
|
| 7400 |
+
{
|
| 7401 |
+
"epoch": 0.9804287045666356,
|
| 7402 |
+
"grad_norm": 0.6707242727279663,
|
| 7403 |
+
"learning_rate": 1.0393852568860719e-07,
|
| 7404 |
+
"loss": 0.0157,
|
| 7405 |
+
"step": 1052
|
| 7406 |
+
},
|
| 7407 |
+
{
|
| 7408 |
+
"epoch": 0.9813606710158435,
|
| 7409 |
+
"grad_norm": 0.9473837614059448,
|
| 7410 |
+
"learning_rate": 9.427834472588992e-08,
|
| 7411 |
+
"loss": 0.0473,
|
| 7412 |
+
"step": 1053
|
| 7413 |
+
},
|
| 7414 |
+
{
|
| 7415 |
+
"epoch": 0.9822926374650512,
|
| 7416 |
+
"grad_norm": 0.40152788162231445,
|
| 7417 |
+
"learning_rate": 8.508881421145365e-08,
|
| 7418 |
+
"loss": 0.0091,
|
| 7419 |
+
"step": 1054
|
| 7420 |
+
},
|
| 7421 |
+
{
|
| 7422 |
+
"epoch": 0.983224603914259,
|
| 7423 |
+
"grad_norm": 0.11309801042079926,
|
| 7424 |
+
"learning_rate": 7.637002080985168e-08,
|
| 7425 |
+
"loss": 0.002,
|
| 7426 |
+
"step": 1055
|
| 7427 |
+
},
|
| 7428 |
+
{
|
| 7429 |
+
"epoch": 0.9841565703634669,
|
| 7430 |
+
"grad_norm": 0.1226629987359047,
|
| 7431 |
+
"learning_rate": 6.812204674619959e-08,
|
| 7432 |
+
"loss": 0.0025,
|
| 7433 |
+
"step": 1056
|
| 7434 |
+
},
|
| 7435 |
+
{
|
| 7436 |
+
"epoch": 0.9850885368126747,
|
| 7437 |
+
"grad_norm": 0.6275689601898193,
|
| 7438 |
+
"learning_rate": 6.034496980542037e-08,
|
| 7439 |
+
"loss": 0.0149,
|
| 7440 |
+
"step": 1057
|
| 7441 |
+
},
|
| 7442 |
+
{
|
| 7443 |
+
"epoch": 0.9860205032618826,
|
| 7444 |
+
"grad_norm": 0.6120526194572449,
|
| 7445 |
+
"learning_rate": 5.303886333151153e-08,
|
| 7446 |
+
"loss": 0.0088,
|
| 7447 |
+
"step": 1058
|
| 7448 |
+
},
|
| 7449 |
+
{
|
| 7450 |
+
"epoch": 0.9869524697110904,
|
| 7451 |
+
"grad_norm": 0.8378371000289917,
|
| 7452 |
+
"learning_rate": 4.620379622682358e-08,
|
| 7453 |
+
"loss": 0.0166,
|
| 7454 |
+
"step": 1059
|
| 7455 |
+
},
|
| 7456 |
+
{
|
| 7457 |
+
"epoch": 0.9878844361602982,
|
| 7458 |
+
"grad_norm": 0.06605886667966843,
|
| 7459 |
+
"learning_rate": 3.9839832951465986e-08,
|
| 7460 |
+
"loss": 0.0012,
|
| 7461 |
+
"step": 1060
|
| 7462 |
+
},
|
| 7463 |
+
{
|
| 7464 |
+
"epoch": 0.9888164026095061,
|
| 7465 |
+
"grad_norm": 0.9707759618759155,
|
| 7466 |
+
"learning_rate": 3.394703352263551e-08,
|
| 7467 |
+
"loss": 0.0434,
|
| 7468 |
+
"step": 1061
|
| 7469 |
+
},
|
| 7470 |
+
{
|
| 7471 |
+
"epoch": 0.9897483690587139,
|
| 7472 |
+
"grad_norm": 0.12164362519979477,
|
| 7473 |
+
"learning_rate": 2.8525453514099965e-08,
|
| 7474 |
+
"loss": 0.0018,
|
| 7475 |
+
"step": 1062
|
| 7476 |
+
},
|
| 7477 |
+
{
|
| 7478 |
+
"epoch": 0.9906803355079217,
|
| 7479 |
+
"grad_norm": 0.4129875898361206,
|
| 7480 |
+
"learning_rate": 2.3575144055643094e-08,
|
| 7481 |
+
"loss": 0.0093,
|
| 7482 |
+
"step": 1063
|
| 7483 |
+
},
|
| 7484 |
+
{
|
| 7485 |
+
"epoch": 0.9916123019571296,
|
| 7486 |
+
"grad_norm": 0.6688860654830933,
|
| 7487 |
+
"learning_rate": 1.9096151832609375e-08,
|
| 7488 |
+
"loss": 0.0118,
|
| 7489 |
+
"step": 1064
|
| 7490 |
+
},
|
| 7491 |
+
{
|
| 7492 |
+
"epoch": 0.9925442684063374,
|
| 7493 |
+
"grad_norm": 0.1448785960674286,
|
| 7494 |
+
"learning_rate": 1.5088519085437734e-08,
|
| 7495 |
+
"loss": 0.0021,
|
| 7496 |
+
"step": 1065
|
| 7497 |
+
},
|
| 7498 |
+
{
|
| 7499 |
+
"epoch": 0.9934762348555451,
|
| 7500 |
+
"grad_norm": 0.17577403783798218,
|
| 7501 |
+
"learning_rate": 1.1552283609272962e-08,
|
| 7502 |
+
"loss": 0.0036,
|
| 7503 |
+
"step": 1066
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 0.994408201304753,
|
| 7507 |
+
"grad_norm": 0.0788675993680954,
|
| 7508 |
+
"learning_rate": 8.487478753615997e-09,
|
| 7509 |
+
"loss": 0.0016,
|
| 7510 |
+
"step": 1067
|
| 7511 |
+
},
|
| 7512 |
+
{
|
| 7513 |
+
"epoch": 0.9953401677539608,
|
| 7514 |
+
"grad_norm": 0.3536447286605835,
|
| 7515 |
+
"learning_rate": 5.894133422001957e-09,
|
| 7516 |
+
"loss": 0.004,
|
| 7517 |
+
"step": 1068
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.9962721342031687,
|
| 7521 |
+
"grad_norm": 1.099785327911377,
|
| 7522 |
+
"learning_rate": 3.772272071722593e-09,
|
| 7523 |
+
"loss": 0.016,
|
| 7524 |
+
"step": 1069
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.9972041006523765,
|
| 7528 |
+
"grad_norm": 1.2353144884109497,
|
| 7529 |
+
"learning_rate": 2.1219147136264382e-09,
|
| 7530 |
+
"loss": 0.024,
|
| 7531 |
+
"step": 1070
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.9981360671015843,
|
| 7535 |
+
"grad_norm": 0.8307344317436218,
|
| 7536 |
+
"learning_rate": 9.43076911874563e-10,
|
| 7537 |
+
"loss": 0.0139,
|
| 7538 |
+
"step": 1071
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.9990680335507922,
|
| 7542 |
+
"grad_norm": 2.4767701625823975,
|
| 7543 |
+
"learning_rate": 2.3576978384065584e-10,
|
| 7544 |
+
"loss": 0.0363,
|
| 7545 |
+
"step": 1072
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 1.0,
|
| 7549 |
+
"grad_norm": 1.866561770439148,
|
| 7550 |
+
"learning_rate": 0.0,
|
| 7551 |
+
"loss": 0.0964,
|
| 7552 |
+
"step": 1073
|
| 7553 |
}
|
| 7554 |
],
|
| 7555 |
"logging_steps": 1,
|
|
|
|
| 7564 |
"should_evaluate": false,
|
| 7565 |
"should_log": false,
|
| 7566 |
"should_save": true,
|
| 7567 |
+
"should_training_stop": true
|
| 7568 |
},
|
| 7569 |
"attributes": {}
|
| 7570 |
}
|
| 7571 |
},
|
| 7572 |
+
"total_flos": 1.5471679295717376e+17,
|
| 7573 |
"train_batch_size": 2,
|
| 7574 |
"trial_name": null,
|
| 7575 |
"trial_params": null
|