Training in progress, step 1434, 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 3664864144
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:309b5b8c9ac4b830a70f6fab0898dedbcf22d89e2c8dd2e39b1078a8bfd07917
|
| 3 |
size 3664864144
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 7330126728
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0419df361bfd3bf52439d41108b5b608f9c17add6c391f3cd21793e59a0a7bf7
|
| 3 |
size 7330126728
|
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 sha256:821bb6d17fcad1256b8d0e158f027f5e92af17c9455cff930641be7720551348
|
| 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 sha256:3234d47291be091ba5e96799eacd38272127c97f6037585624843f0ee85c4d82
|
| 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": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -8372,6 +8372,1679 @@
|
|
| 8372 |
"learning_rate": 1.3790573988406074e-05,
|
| 8373 |
"loss": 0.0,
|
| 8374 |
"step": 1195
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8375 |
}
|
| 8376 |
],
|
| 8377 |
"logging_steps": 1,
|
|
@@ -8391,7 +10064,7 @@
|
|
| 8391 |
"attributes": {}
|
| 8392 |
}
|
| 8393 |
},
|
| 8394 |
-
"total_flos":
|
| 8395 |
"train_batch_size": 4,
|
| 8396 |
"trial_name": null,
|
| 8397 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9281553398058252,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 1434,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 8372 |
"learning_rate": 1.3790573988406074e-05,
|
| 8373 |
"loss": 0.0,
|
| 8374 |
"step": 1195
|
| 8375 |
+
},
|
| 8376 |
+
{
|
| 8377 |
+
"epoch": 0.7741100323624596,
|
| 8378 |
+
"grad_norm": 2.7803118427982554e-05,
|
| 8379 |
+
"learning_rate": 1.3715696000155614e-05,
|
| 8380 |
+
"loss": 0.0,
|
| 8381 |
+
"step": 1196
|
| 8382 |
+
},
|
| 8383 |
+
{
|
| 8384 |
+
"epoch": 0.7747572815533981,
|
| 8385 |
+
"grad_norm": 2.7793254048447125e-05,
|
| 8386 |
+
"learning_rate": 1.3640989519231629e-05,
|
| 8387 |
+
"loss": 0.0,
|
| 8388 |
+
"step": 1197
|
| 8389 |
+
},
|
| 8390 |
+
{
|
| 8391 |
+
"epoch": 0.7754045307443366,
|
| 8392 |
+
"grad_norm": 2.8403444957803003e-05,
|
| 8393 |
+
"learning_rate": 1.3566454898754016e-05,
|
| 8394 |
+
"loss": 0.0,
|
| 8395 |
+
"step": 1198
|
| 8396 |
+
},
|
| 8397 |
+
{
|
| 8398 |
+
"epoch": 0.7760517799352751,
|
| 8399 |
+
"grad_norm": 3.0428616810240783e-05,
|
| 8400 |
+
"learning_rate": 1.3492092491030279e-05,
|
| 8401 |
+
"loss": 0.0,
|
| 8402 |
+
"step": 1199
|
| 8403 |
+
},
|
| 8404 |
+
{
|
| 8405 |
+
"epoch": 0.7766990291262136,
|
| 8406 |
+
"grad_norm": 2.8647311410168186e-05,
|
| 8407 |
+
"learning_rate": 1.3417902647553948e-05,
|
| 8408 |
+
"loss": 0.0,
|
| 8409 |
+
"step": 1200
|
| 8410 |
+
},
|
| 8411 |
+
{
|
| 8412 |
+
"epoch": 0.7773462783171521,
|
| 8413 |
+
"grad_norm": 3.5604185541160405e-05,
|
| 8414 |
+
"learning_rate": 1.3343885719002896e-05,
|
| 8415 |
+
"loss": 0.0,
|
| 8416 |
+
"step": 1201
|
| 8417 |
+
},
|
| 8418 |
+
{
|
| 8419 |
+
"epoch": 0.7779935275080906,
|
| 8420 |
+
"grad_norm": 3.1396841222885996e-05,
|
| 8421 |
+
"learning_rate": 1.3270042055237609e-05,
|
| 8422 |
+
"loss": 0.0,
|
| 8423 |
+
"step": 1202
|
| 8424 |
+
},
|
| 8425 |
+
{
|
| 8426 |
+
"epoch": 0.7786407766990291,
|
| 8427 |
+
"grad_norm": 3.281456883996725e-05,
|
| 8428 |
+
"learning_rate": 1.3196372005299656e-05,
|
| 8429 |
+
"loss": 0.0,
|
| 8430 |
+
"step": 1203
|
| 8431 |
+
},
|
| 8432 |
+
{
|
| 8433 |
+
"epoch": 0.7792880258899676,
|
| 8434 |
+
"grad_norm": 2.9937604267615825e-05,
|
| 8435 |
+
"learning_rate": 1.3122875917409916e-05,
|
| 8436 |
+
"loss": 0.0,
|
| 8437 |
+
"step": 1204
|
| 8438 |
+
},
|
| 8439 |
+
{
|
| 8440 |
+
"epoch": 0.7799352750809061,
|
| 8441 |
+
"grad_norm": 2.828240758390166e-05,
|
| 8442 |
+
"learning_rate": 1.3049554138967051e-05,
|
| 8443 |
+
"loss": 0.0,
|
| 8444 |
+
"step": 1205
|
| 8445 |
+
},
|
| 8446 |
+
{
|
| 8447 |
+
"epoch": 0.7805825242718447,
|
| 8448 |
+
"grad_norm": 3.450288932071999e-05,
|
| 8449 |
+
"learning_rate": 1.2976407016545744e-05,
|
| 8450 |
+
"loss": 0.0,
|
| 8451 |
+
"step": 1206
|
| 8452 |
+
},
|
| 8453 |
+
{
|
| 8454 |
+
"epoch": 0.7812297734627832,
|
| 8455 |
+
"grad_norm": 3.089702659053728e-05,
|
| 8456 |
+
"learning_rate": 1.2903434895895183e-05,
|
| 8457 |
+
"loss": 0.0,
|
| 8458 |
+
"step": 1207
|
| 8459 |
+
},
|
| 8460 |
+
{
|
| 8461 |
+
"epoch": 0.7818770226537217,
|
| 8462 |
+
"grad_norm": 3.3058688131859526e-05,
|
| 8463 |
+
"learning_rate": 1.2830638121937289e-05,
|
| 8464 |
+
"loss": 0.0,
|
| 8465 |
+
"step": 1208
|
| 8466 |
+
},
|
| 8467 |
+
{
|
| 8468 |
+
"epoch": 0.7825242718446602,
|
| 8469 |
+
"grad_norm": 3.49712063325569e-05,
|
| 8470 |
+
"learning_rate": 1.275801703876524e-05,
|
| 8471 |
+
"loss": 0.0,
|
| 8472 |
+
"step": 1209
|
| 8473 |
+
},
|
| 8474 |
+
{
|
| 8475 |
+
"epoch": 0.7831715210355987,
|
| 8476 |
+
"grad_norm": 3.258398282923736e-05,
|
| 8477 |
+
"learning_rate": 1.2685571989641698e-05,
|
| 8478 |
+
"loss": 0.0,
|
| 8479 |
+
"step": 1210
|
| 8480 |
+
},
|
| 8481 |
+
{
|
| 8482 |
+
"epoch": 0.7838187702265372,
|
| 8483 |
+
"grad_norm": 3.234569157939404e-05,
|
| 8484 |
+
"learning_rate": 1.2613303316997288e-05,
|
| 8485 |
+
"loss": 0.0,
|
| 8486 |
+
"step": 1211
|
| 8487 |
+
},
|
| 8488 |
+
{
|
| 8489 |
+
"epoch": 0.7844660194174757,
|
| 8490 |
+
"grad_norm": 2.958671575470362e-05,
|
| 8491 |
+
"learning_rate": 1.2541211362428962e-05,
|
| 8492 |
+
"loss": 0.0,
|
| 8493 |
+
"step": 1212
|
| 8494 |
+
},
|
| 8495 |
+
{
|
| 8496 |
+
"epoch": 0.7851132686084142,
|
| 8497 |
+
"grad_norm": 2.8513155484688468e-05,
|
| 8498 |
+
"learning_rate": 1.246929646669831e-05,
|
| 8499 |
+
"loss": 0.0,
|
| 8500 |
+
"step": 1213
|
| 8501 |
+
},
|
| 8502 |
+
{
|
| 8503 |
+
"epoch": 0.7857605177993527,
|
| 8504 |
+
"grad_norm": 3.2108306186273694e-05,
|
| 8505 |
+
"learning_rate": 1.2397558969730077e-05,
|
| 8506 |
+
"loss": 0.0,
|
| 8507 |
+
"step": 1214
|
| 8508 |
+
},
|
| 8509 |
+
{
|
| 8510 |
+
"epoch": 0.7864077669902912,
|
| 8511 |
+
"grad_norm": 3.305998325231485e-05,
|
| 8512 |
+
"learning_rate": 1.2325999210610423e-05,
|
| 8513 |
+
"loss": 0.0,
|
| 8514 |
+
"step": 1215
|
| 8515 |
+
},
|
| 8516 |
+
{
|
| 8517 |
+
"epoch": 0.7870550161812297,
|
| 8518 |
+
"grad_norm": 2.8868706067441963e-05,
|
| 8519 |
+
"learning_rate": 1.2254617527585433e-05,
|
| 8520 |
+
"loss": 0.0,
|
| 8521 |
+
"step": 1216
|
| 8522 |
+
},
|
| 8523 |
+
{
|
| 8524 |
+
"epoch": 0.7877022653721683,
|
| 8525 |
+
"grad_norm": 3.401190770091489e-05,
|
| 8526 |
+
"learning_rate": 1.2183414258059417e-05,
|
| 8527 |
+
"loss": 0.0,
|
| 8528 |
+
"step": 1217
|
| 8529 |
+
},
|
| 8530 |
+
{
|
| 8531 |
+
"epoch": 0.7883495145631068,
|
| 8532 |
+
"grad_norm": 2.9487531719496474e-05,
|
| 8533 |
+
"learning_rate": 1.2112389738593427e-05,
|
| 8534 |
+
"loss": 0.0,
|
| 8535 |
+
"step": 1218
|
| 8536 |
+
},
|
| 8537 |
+
{
|
| 8538 |
+
"epoch": 0.7889967637540453,
|
| 8539 |
+
"grad_norm": 3.0200997571228072e-05,
|
| 8540 |
+
"learning_rate": 1.2041544304903552e-05,
|
| 8541 |
+
"loss": 0.0,
|
| 8542 |
+
"step": 1219
|
| 8543 |
+
},
|
| 8544 |
+
{
|
| 8545 |
+
"epoch": 0.7896440129449838,
|
| 8546 |
+
"grad_norm": 2.9967310183565132e-05,
|
| 8547 |
+
"learning_rate": 1.1970878291859423e-05,
|
| 8548 |
+
"loss": 0.0,
|
| 8549 |
+
"step": 1220
|
| 8550 |
+
},
|
| 8551 |
+
{
|
| 8552 |
+
"epoch": 0.7902912621359224,
|
| 8553 |
+
"grad_norm": 3.139051841571927e-05,
|
| 8554 |
+
"learning_rate": 1.190039203348256e-05,
|
| 8555 |
+
"loss": 0.0,
|
| 8556 |
+
"step": 1221
|
| 8557 |
+
},
|
| 8558 |
+
{
|
| 8559 |
+
"epoch": 0.7909385113268609,
|
| 8560 |
+
"grad_norm": 3.090596874244511e-05,
|
| 8561 |
+
"learning_rate": 1.183008586294485e-05,
|
| 8562 |
+
"loss": 0.0,
|
| 8563 |
+
"step": 1222
|
| 8564 |
+
},
|
| 8565 |
+
{
|
| 8566 |
+
"epoch": 0.7915857605177994,
|
| 8567 |
+
"grad_norm": 3.282758916611783e-05,
|
| 8568 |
+
"learning_rate": 1.1759960112566953e-05,
|
| 8569 |
+
"loss": 0.0,
|
| 8570 |
+
"step": 1223
|
| 8571 |
+
},
|
| 8572 |
+
{
|
| 8573 |
+
"epoch": 0.7922330097087379,
|
| 8574 |
+
"grad_norm": 2.9495713533833623e-05,
|
| 8575 |
+
"learning_rate": 1.1690015113816689e-05,
|
| 8576 |
+
"loss": 0.0,
|
| 8577 |
+
"step": 1224
|
| 8578 |
+
},
|
| 8579 |
+
{
|
| 8580 |
+
"epoch": 0.7928802588996764,
|
| 8581 |
+
"grad_norm": 3.188411210430786e-05,
|
| 8582 |
+
"learning_rate": 1.1620251197307535e-05,
|
| 8583 |
+
"loss": 0.0,
|
| 8584 |
+
"step": 1225
|
| 8585 |
+
},
|
| 8586 |
+
{
|
| 8587 |
+
"epoch": 0.7935275080906149,
|
| 8588 |
+
"grad_norm": 3.187660331605002e-05,
|
| 8589 |
+
"learning_rate": 1.1550668692797062e-05,
|
| 8590 |
+
"loss": 0.0,
|
| 8591 |
+
"step": 1226
|
| 8592 |
+
},
|
| 8593 |
+
{
|
| 8594 |
+
"epoch": 0.7941747572815534,
|
| 8595 |
+
"grad_norm": 3.0207180316210724e-05,
|
| 8596 |
+
"learning_rate": 1.148126792918527e-05,
|
| 8597 |
+
"loss": 0.0,
|
| 8598 |
+
"step": 1227
|
| 8599 |
+
},
|
| 8600 |
+
{
|
| 8601 |
+
"epoch": 0.7948220064724919,
|
| 8602 |
+
"grad_norm": 2.8910977562190965e-05,
|
| 8603 |
+
"learning_rate": 1.1412049234513178e-05,
|
| 8604 |
+
"loss": 0.0,
|
| 8605 |
+
"step": 1228
|
| 8606 |
+
},
|
| 8607 |
+
{
|
| 8608 |
+
"epoch": 0.7954692556634304,
|
| 8609 |
+
"grad_norm": 3.139706313959323e-05,
|
| 8610 |
+
"learning_rate": 1.1343012935961206e-05,
|
| 8611 |
+
"loss": 0.0,
|
| 8612 |
+
"step": 1229
|
| 8613 |
+
},
|
| 8614 |
+
{
|
| 8615 |
+
"epoch": 0.7961165048543689,
|
| 8616 |
+
"grad_norm": 2.9498503863578662e-05,
|
| 8617 |
+
"learning_rate": 1.1274159359847591e-05,
|
| 8618 |
+
"loss": 0.0,
|
| 8619 |
+
"step": 1230
|
| 8620 |
+
},
|
| 8621 |
+
{
|
| 8622 |
+
"epoch": 0.7967637540453074,
|
| 8623 |
+
"grad_norm": 2.889409734052606e-05,
|
| 8624 |
+
"learning_rate": 1.1205488831626936e-05,
|
| 8625 |
+
"loss": 0.0,
|
| 8626 |
+
"step": 1231
|
| 8627 |
+
},
|
| 8628 |
+
{
|
| 8629 |
+
"epoch": 0.7974110032362459,
|
| 8630 |
+
"grad_norm": 2.7833795684273355e-05,
|
| 8631 |
+
"learning_rate": 1.1137001675888564e-05,
|
| 8632 |
+
"loss": 0.0,
|
| 8633 |
+
"step": 1232
|
| 8634 |
+
},
|
| 8635 |
+
{
|
| 8636 |
+
"epoch": 0.7980582524271844,
|
| 8637 |
+
"grad_norm": 2.9727289074799046e-05,
|
| 8638 |
+
"learning_rate": 1.1068698216355084e-05,
|
| 8639 |
+
"loss": 0.0,
|
| 8640 |
+
"step": 1233
|
| 8641 |
+
},
|
| 8642 |
+
{
|
| 8643 |
+
"epoch": 0.7987055016181229,
|
| 8644 |
+
"grad_norm": 2.9720979000558145e-05,
|
| 8645 |
+
"learning_rate": 1.100057877588081e-05,
|
| 8646 |
+
"loss": 0.0,
|
| 8647 |
+
"step": 1234
|
| 8648 |
+
},
|
| 8649 |
+
{
|
| 8650 |
+
"epoch": 0.7993527508090615,
|
| 8651 |
+
"grad_norm": 3.0549646908184513e-05,
|
| 8652 |
+
"learning_rate": 1.0932643676450205e-05,
|
| 8653 |
+
"loss": 0.0,
|
| 8654 |
+
"step": 1235
|
| 8655 |
+
},
|
| 8656 |
+
{
|
| 8657 |
+
"epoch": 0.8,
|
| 8658 |
+
"grad_norm": 2.8757727704942226e-05,
|
| 8659 |
+
"learning_rate": 1.0864893239176438e-05,
|
| 8660 |
+
"loss": 0.0,
|
| 8661 |
+
"step": 1236
|
| 8662 |
+
},
|
| 8663 |
+
{
|
| 8664 |
+
"epoch": 0.8006472491909385,
|
| 8665 |
+
"grad_norm": 2.9481703677447513e-05,
|
| 8666 |
+
"learning_rate": 1.0797327784299837e-05,
|
| 8667 |
+
"loss": 0.0,
|
| 8668 |
+
"step": 1237
|
| 8669 |
+
},
|
| 8670 |
+
{
|
| 8671 |
+
"epoch": 0.801294498381877,
|
| 8672 |
+
"grad_norm": 3.140431363135576e-05,
|
| 8673 |
+
"learning_rate": 1.0729947631186276e-05,
|
| 8674 |
+
"loss": 0.0,
|
| 8675 |
+
"step": 1238
|
| 8676 |
+
},
|
| 8677 |
+
{
|
| 8678 |
+
"epoch": 0.8019417475728156,
|
| 8679 |
+
"grad_norm": 2.90126190520823e-05,
|
| 8680 |
+
"learning_rate": 1.066275309832584e-05,
|
| 8681 |
+
"loss": 0.0,
|
| 8682 |
+
"step": 1239
|
| 8683 |
+
},
|
| 8684 |
+
{
|
| 8685 |
+
"epoch": 0.8025889967637541,
|
| 8686 |
+
"grad_norm": 2.924008913396392e-05,
|
| 8687 |
+
"learning_rate": 1.0595744503331207e-05,
|
| 8688 |
+
"loss": 0.0,
|
| 8689 |
+
"step": 1240
|
| 8690 |
+
},
|
| 8691 |
+
{
|
| 8692 |
+
"epoch": 0.8032362459546926,
|
| 8693 |
+
"grad_norm": 3.188064874848351e-05,
|
| 8694 |
+
"learning_rate": 1.0528922162936134e-05,
|
| 8695 |
+
"loss": 0.0,
|
| 8696 |
+
"step": 1241
|
| 8697 |
+
},
|
| 8698 |
+
{
|
| 8699 |
+
"epoch": 0.8038834951456311,
|
| 8700 |
+
"grad_norm": 2.780007525871042e-05,
|
| 8701 |
+
"learning_rate": 1.0462286392994058e-05,
|
| 8702 |
+
"loss": 0.0,
|
| 8703 |
+
"step": 1242
|
| 8704 |
+
},
|
| 8705 |
+
{
|
| 8706 |
+
"epoch": 0.8045307443365696,
|
| 8707 |
+
"grad_norm": 2.7808411687146872e-05,
|
| 8708 |
+
"learning_rate": 1.0395837508476486e-05,
|
| 8709 |
+
"loss": 0.0,
|
| 8710 |
+
"step": 1243
|
| 8711 |
+
},
|
| 8712 |
+
{
|
| 8713 |
+
"epoch": 0.8051779935275081,
|
| 8714 |
+
"grad_norm": 2.9482629543053918e-05,
|
| 8715 |
+
"learning_rate": 1.0329575823471588e-05,
|
| 8716 |
+
"loss": 0.0,
|
| 8717 |
+
"step": 1244
|
| 8718 |
+
},
|
| 8719 |
+
{
|
| 8720 |
+
"epoch": 0.8058252427184466,
|
| 8721 |
+
"grad_norm": 2.779496571747586e-05,
|
| 8722 |
+
"learning_rate": 1.0263501651182706e-05,
|
| 8723 |
+
"loss": 0.0,
|
| 8724 |
+
"step": 1245
|
| 8725 |
+
},
|
| 8726 |
+
{
|
| 8727 |
+
"epoch": 0.8064724919093851,
|
| 8728 |
+
"grad_norm": 2.971193680423312e-05,
|
| 8729 |
+
"learning_rate": 1.0197615303926794e-05,
|
| 8730 |
+
"loss": 0.0,
|
| 8731 |
+
"step": 1246
|
| 8732 |
+
},
|
| 8733 |
+
{
|
| 8734 |
+
"epoch": 0.8071197411003236,
|
| 8735 |
+
"grad_norm": 2.7200567274121568e-05,
|
| 8736 |
+
"learning_rate": 1.0131917093133075e-05,
|
| 8737 |
+
"loss": 0.0,
|
| 8738 |
+
"step": 1247
|
| 8739 |
+
},
|
| 8740 |
+
{
|
| 8741 |
+
"epoch": 0.8077669902912621,
|
| 8742 |
+
"grad_norm": 2.7336840503267013e-05,
|
| 8743 |
+
"learning_rate": 1.0066407329341443e-05,
|
| 8744 |
+
"loss": 0.0,
|
| 8745 |
+
"step": 1248
|
| 8746 |
+
},
|
| 8747 |
+
{
|
| 8748 |
+
"epoch": 0.8084142394822007,
|
| 8749 |
+
"grad_norm": 3.140536500723101e-05,
|
| 8750 |
+
"learning_rate": 1.0001086322201048e-05,
|
| 8751 |
+
"loss": 0.0,
|
| 8752 |
+
"step": 1249
|
| 8753 |
+
},
|
| 8754 |
+
{
|
| 8755 |
+
"epoch": 0.8090614886731392,
|
| 8756 |
+
"grad_norm": 3.138635292998515e-05,
|
| 8757 |
+
"learning_rate": 9.935954380468859e-06,
|
| 8758 |
+
"loss": 0.0,
|
| 8759 |
+
"step": 1250
|
| 8760 |
+
},
|
| 8761 |
+
{
|
| 8762 |
+
"epoch": 0.8097087378640777,
|
| 8763 |
+
"grad_norm": 3.832921720459126e-05,
|
| 8764 |
+
"learning_rate": 9.87101181200818e-06,
|
| 8765 |
+
"loss": 0.0,
|
| 8766 |
+
"step": 1251
|
| 8767 |
+
},
|
| 8768 |
+
{
|
| 8769 |
+
"epoch": 0.8103559870550162,
|
| 8770 |
+
"grad_norm": 2.887231130443979e-05,
|
| 8771 |
+
"learning_rate": 9.806258923787154e-06,
|
| 8772 |
+
"loss": 0.0,
|
| 8773 |
+
"step": 1252
|
| 8774 |
+
},
|
| 8775 |
+
{
|
| 8776 |
+
"epoch": 0.8110032362459547,
|
| 8777 |
+
"grad_norm": 3.210567228961736e-05,
|
| 8778 |
+
"learning_rate": 9.7416960218774e-06,
|
| 8779 |
+
"loss": 0.0,
|
| 8780 |
+
"step": 1253
|
| 8781 |
+
},
|
| 8782 |
+
{
|
| 8783 |
+
"epoch": 0.8116504854368932,
|
| 8784 |
+
"grad_norm": 3.331467814859934e-05,
|
| 8785 |
+
"learning_rate": 9.67732341145246e-06,
|
| 8786 |
+
"loss": 0.0,
|
| 8787 |
+
"step": 1254
|
| 8788 |
+
},
|
| 8789 |
+
{
|
| 8790 |
+
"epoch": 0.8122977346278317,
|
| 8791 |
+
"grad_norm": 2.97059250442544e-05,
|
| 8792 |
+
"learning_rate": 9.613141396786462e-06,
|
| 8793 |
+
"loss": 0.0,
|
| 8794 |
+
"step": 1255
|
| 8795 |
+
},
|
| 8796 |
+
{
|
| 8797 |
+
"epoch": 0.8129449838187702,
|
| 8798 |
+
"grad_norm": 3.066575663979165e-05,
|
| 8799 |
+
"learning_rate": 9.549150281252633e-06,
|
| 8800 |
+
"loss": 0.0,
|
| 8801 |
+
"step": 1256
|
| 8802 |
+
},
|
| 8803 |
+
{
|
| 8804 |
+
"epoch": 0.8135922330097087,
|
| 8805 |
+
"grad_norm": 2.9601724236272275e-05,
|
| 8806 |
+
"learning_rate": 9.48535036732181e-06,
|
| 8807 |
+
"loss": 0.0,
|
| 8808 |
+
"step": 1257
|
| 8809 |
+
},
|
| 8810 |
+
{
|
| 8811 |
+
"epoch": 0.8142394822006472,
|
| 8812 |
+
"grad_norm": 2.846097959263716e-05,
|
| 8813 |
+
"learning_rate": 9.421741956561137e-06,
|
| 8814 |
+
"loss": 0.0,
|
| 8815 |
+
"step": 1258
|
| 8816 |
+
},
|
| 8817 |
+
{
|
| 8818 |
+
"epoch": 0.8148867313915857,
|
| 8819 |
+
"grad_norm": 3.0182511181919836e-05,
|
| 8820 |
+
"learning_rate": 9.358325349632514e-06,
|
| 8821 |
+
"loss": 0.0,
|
| 8822 |
+
"step": 1259
|
| 8823 |
+
},
|
| 8824 |
+
{
|
| 8825 |
+
"epoch": 0.8155339805825242,
|
| 8826 |
+
"grad_norm": 3.1620755180483684e-05,
|
| 8827 |
+
"learning_rate": 9.295100846291238e-06,
|
| 8828 |
+
"loss": 0.0,
|
| 8829 |
+
"step": 1260
|
| 8830 |
+
},
|
| 8831 |
+
{
|
| 8832 |
+
"epoch": 0.8161812297734627,
|
| 8833 |
+
"grad_norm": 3.21077459375374e-05,
|
| 8834 |
+
"learning_rate": 9.232068745384603e-06,
|
| 8835 |
+
"loss": 0.0,
|
| 8836 |
+
"step": 1261
|
| 8837 |
+
},
|
| 8838 |
+
{
|
| 8839 |
+
"epoch": 0.8168284789644013,
|
| 8840 |
+
"grad_norm": 3.1390874937642366e-05,
|
| 8841 |
+
"learning_rate": 9.16922934485046e-06,
|
| 8842 |
+
"loss": 0.0,
|
| 8843 |
+
"step": 1262
|
| 8844 |
+
},
|
| 8845 |
+
{
|
| 8846 |
+
"epoch": 0.8174757281553398,
|
| 8847 |
+
"grad_norm": 3.0309969588415697e-05,
|
| 8848 |
+
"learning_rate": 9.106582941715768e-06,
|
| 8849 |
+
"loss": 0.0,
|
| 8850 |
+
"step": 1263
|
| 8851 |
+
},
|
| 8852 |
+
{
|
| 8853 |
+
"epoch": 0.8181229773462784,
|
| 8854 |
+
"grad_norm": 2.8271257178857923e-05,
|
| 8855 |
+
"learning_rate": 9.044129832095299e-06,
|
| 8856 |
+
"loss": 0.0,
|
| 8857 |
+
"step": 1264
|
| 8858 |
+
},
|
| 8859 |
+
{
|
| 8860 |
+
"epoch": 0.8187702265372169,
|
| 8861 |
+
"grad_norm": 7.29700768715702e-05,
|
| 8862 |
+
"learning_rate": 8.981870311190099e-06,
|
| 8863 |
+
"loss": 0.0,
|
| 8864 |
+
"step": 1265
|
| 8865 |
+
},
|
| 8866 |
+
{
|
| 8867 |
+
"epoch": 0.8194174757281554,
|
| 8868 |
+
"grad_norm": 3.4984001104021445e-05,
|
| 8869 |
+
"learning_rate": 8.91980467328623e-06,
|
| 8870 |
+
"loss": 0.0,
|
| 8871 |
+
"step": 1266
|
| 8872 |
+
},
|
| 8873 |
+
{
|
| 8874 |
+
"epoch": 0.8200647249190939,
|
| 8875 |
+
"grad_norm": 2.912871968874242e-05,
|
| 8876 |
+
"learning_rate": 8.857933211753289e-06,
|
| 8877 |
+
"loss": 0.0,
|
| 8878 |
+
"step": 1267
|
| 8879 |
+
},
|
| 8880 |
+
{
|
| 8881 |
+
"epoch": 0.8207119741100324,
|
| 8882 |
+
"grad_norm": 3.0196642910595983e-05,
|
| 8883 |
+
"learning_rate": 8.796256219043042e-06,
|
| 8884 |
+
"loss": 0.0,
|
| 8885 |
+
"step": 1268
|
| 8886 |
+
},
|
| 8887 |
+
{
|
| 8888 |
+
"epoch": 0.8213592233009709,
|
| 8889 |
+
"grad_norm": 2.9139126127120107e-05,
|
| 8890 |
+
"learning_rate": 8.734773986688032e-06,
|
| 8891 |
+
"loss": 0.0,
|
| 8892 |
+
"step": 1269
|
| 8893 |
+
},
|
| 8894 |
+
{
|
| 8895 |
+
"epoch": 0.8220064724919094,
|
| 8896 |
+
"grad_norm": 3.0069269996602088e-05,
|
| 8897 |
+
"learning_rate": 8.673486805300263e-06,
|
| 8898 |
+
"loss": 0.0,
|
| 8899 |
+
"step": 1270
|
| 8900 |
+
},
|
| 8901 |
+
{
|
| 8902 |
+
"epoch": 0.8226537216828479,
|
| 8903 |
+
"grad_norm": 3.189206836395897e-05,
|
| 8904 |
+
"learning_rate": 8.61239496456973e-06,
|
| 8905 |
+
"loss": 0.0,
|
| 8906 |
+
"step": 1271
|
| 8907 |
+
},
|
| 8908 |
+
{
|
| 8909 |
+
"epoch": 0.8233009708737864,
|
| 8910 |
+
"grad_norm": 3.237631972297095e-05,
|
| 8911 |
+
"learning_rate": 8.551498753263133e-06,
|
| 8912 |
+
"loss": 0.0,
|
| 8913 |
+
"step": 1272
|
| 8914 |
+
},
|
| 8915 |
+
{
|
| 8916 |
+
"epoch": 0.8239482200647249,
|
| 8917 |
+
"grad_norm": 3.1884130294201896e-05,
|
| 8918 |
+
"learning_rate": 8.490798459222476e-06,
|
| 8919 |
+
"loss": 0.0,
|
| 8920 |
+
"step": 1273
|
| 8921 |
+
},
|
| 8922 |
+
{
|
| 8923 |
+
"epoch": 0.8245954692556634,
|
| 8924 |
+
"grad_norm": 3.139725959044881e-05,
|
| 8925 |
+
"learning_rate": 8.430294369363667e-06,
|
| 8926 |
+
"loss": 0.0,
|
| 8927 |
+
"step": 1274
|
| 8928 |
+
},
|
| 8929 |
+
{
|
| 8930 |
+
"epoch": 0.8252427184466019,
|
| 8931 |
+
"grad_norm": 3.044573895749636e-05,
|
| 8932 |
+
"learning_rate": 8.369986769675269e-06,
|
| 8933 |
+
"loss": 0.0,
|
| 8934 |
+
"step": 1275
|
| 8935 |
+
},
|
| 8936 |
+
{
|
| 8937 |
+
"epoch": 0.8258899676375404,
|
| 8938 |
+
"grad_norm": 3.236419433960691e-05,
|
| 8939 |
+
"learning_rate": 8.309875945217022e-06,
|
| 8940 |
+
"loss": 0.0,
|
| 8941 |
+
"step": 1276
|
| 8942 |
+
},
|
| 8943 |
+
{
|
| 8944 |
+
"epoch": 0.8265372168284789,
|
| 8945 |
+
"grad_norm": 3.04457080346765e-05,
|
| 8946 |
+
"learning_rate": 8.249962180118581e-06,
|
| 8947 |
+
"loss": 0.0,
|
| 8948 |
+
"step": 1277
|
| 8949 |
+
},
|
| 8950 |
+
{
|
| 8951 |
+
"epoch": 0.8271844660194175,
|
| 8952 |
+
"grad_norm": 3.3080803405027837e-05,
|
| 8953 |
+
"learning_rate": 8.190245757578175e-06,
|
| 8954 |
+
"loss": 0.0,
|
| 8955 |
+
"step": 1278
|
| 8956 |
+
},
|
| 8957 |
+
{
|
| 8958 |
+
"epoch": 0.827831715210356,
|
| 8959 |
+
"grad_norm": 3.0458842957159504e-05,
|
| 8960 |
+
"learning_rate": 8.130726959861201e-06,
|
| 8961 |
+
"loss": 0.0,
|
| 8962 |
+
"step": 1279
|
| 8963 |
+
},
|
| 8964 |
+
{
|
| 8965 |
+
"epoch": 0.8284789644012945,
|
| 8966 |
+
"grad_norm": 2.9737420845776796e-05,
|
| 8967 |
+
"learning_rate": 8.071406068298926e-06,
|
| 8968 |
+
"loss": 0.0,
|
| 8969 |
+
"step": 1280
|
| 8970 |
+
},
|
| 8971 |
+
{
|
| 8972 |
+
"epoch": 0.829126213592233,
|
| 8973 |
+
"grad_norm": 2.828111610142514e-05,
|
| 8974 |
+
"learning_rate": 8.012283363287205e-06,
|
| 8975 |
+
"loss": 0.0,
|
| 8976 |
+
"step": 1281
|
| 8977 |
+
},
|
| 8978 |
+
{
|
| 8979 |
+
"epoch": 0.8297734627831715,
|
| 8980 |
+
"grad_norm": 2.9597998945973814e-05,
|
| 8981 |
+
"learning_rate": 7.95335912428506e-06,
|
| 8982 |
+
"loss": 0.0,
|
| 8983 |
+
"step": 1282
|
| 8984 |
+
},
|
| 8985 |
+
{
|
| 8986 |
+
"epoch": 0.83042071197411,
|
| 8987 |
+
"grad_norm": 2.8280543119763024e-05,
|
| 8988 |
+
"learning_rate": 7.894633629813458e-06,
|
| 8989 |
+
"loss": 0.0,
|
| 8990 |
+
"step": 1283
|
| 8991 |
+
},
|
| 8992 |
+
{
|
| 8993 |
+
"epoch": 0.8310679611650486,
|
| 8994 |
+
"grad_norm": 2.780561408144422e-05,
|
| 8995 |
+
"learning_rate": 7.836107157453937e-06,
|
| 8996 |
+
"loss": 0.0,
|
| 8997 |
+
"step": 1284
|
| 8998 |
+
},
|
| 8999 |
+
{
|
| 9000 |
+
"epoch": 0.8317152103559871,
|
| 9001 |
+
"grad_norm": 3.162684879498556e-05,
|
| 9002 |
+
"learning_rate": 7.77777998384726e-06,
|
| 9003 |
+
"loss": 0.0,
|
| 9004 |
+
"step": 1285
|
| 9005 |
+
},
|
| 9006 |
+
{
|
| 9007 |
+
"epoch": 0.8323624595469256,
|
| 9008 |
+
"grad_norm": 2.8393124011927284e-05,
|
| 9009 |
+
"learning_rate": 7.719652384692216e-06,
|
| 9010 |
+
"loss": 0.0,
|
| 9011 |
+
"step": 1286
|
| 9012 |
+
},
|
| 9013 |
+
{
|
| 9014 |
+
"epoch": 0.8330097087378641,
|
| 9015 |
+
"grad_norm": 2.7938367566093802e-05,
|
| 9016 |
+
"learning_rate": 7.661724634744221e-06,
|
| 9017 |
+
"loss": 0.0,
|
| 9018 |
+
"step": 1287
|
| 9019 |
+
},
|
| 9020 |
+
{
|
| 9021 |
+
"epoch": 0.8336569579288026,
|
| 9022 |
+
"grad_norm": 2.7803946068161167e-05,
|
| 9023 |
+
"learning_rate": 7.6039970078140345e-06,
|
| 9024 |
+
"loss": 0.0,
|
| 9025 |
+
"step": 1288
|
| 9026 |
+
},
|
| 9027 |
+
{
|
| 9028 |
+
"epoch": 0.8343042071197411,
|
| 9029 |
+
"grad_norm": 2.935515840363223e-05,
|
| 9030 |
+
"learning_rate": 7.5464697767665234e-06,
|
| 9031 |
+
"loss": 0.0,
|
| 9032 |
+
"step": 1289
|
| 9033 |
+
},
|
| 9034 |
+
{
|
| 9035 |
+
"epoch": 0.8349514563106796,
|
| 9036 |
+
"grad_norm": 2.8421714887372218e-05,
|
| 9037 |
+
"learning_rate": 7.489143213519301e-06,
|
| 9038 |
+
"loss": 0.0,
|
| 9039 |
+
"step": 1290
|
| 9040 |
+
},
|
| 9041 |
+
{
|
| 9042 |
+
"epoch": 0.8355987055016181,
|
| 9043 |
+
"grad_norm": 2.9472732421709225e-05,
|
| 9044 |
+
"learning_rate": 7.432017589041463e-06,
|
| 9045 |
+
"loss": 0.0,
|
| 9046 |
+
"step": 1291
|
| 9047 |
+
},
|
| 9048 |
+
{
|
| 9049 |
+
"epoch": 0.8362459546925566,
|
| 9050 |
+
"grad_norm": 3.0081349905231036e-05,
|
| 9051 |
+
"learning_rate": 7.375093173352348e-06,
|
| 9052 |
+
"loss": 0.0,
|
| 9053 |
+
"step": 1292
|
| 9054 |
+
},
|
| 9055 |
+
{
|
| 9056 |
+
"epoch": 0.8368932038834952,
|
| 9057 |
+
"grad_norm": 2.7322215828462504e-05,
|
| 9058 |
+
"learning_rate": 7.3183702355202e-06,
|
| 9059 |
+
"loss": 0.0,
|
| 9060 |
+
"step": 1293
|
| 9061 |
+
},
|
| 9062 |
+
{
|
| 9063 |
+
"epoch": 0.8375404530744337,
|
| 9064 |
+
"grad_norm": 2.9117414669599384e-05,
|
| 9065 |
+
"learning_rate": 7.26184904366094e-06,
|
| 9066 |
+
"loss": 0.0,
|
| 9067 |
+
"step": 1294
|
| 9068 |
+
},
|
| 9069 |
+
{
|
| 9070 |
+
"epoch": 0.8381877022653722,
|
| 9071 |
+
"grad_norm": 2.838773616531398e-05,
|
| 9072 |
+
"learning_rate": 7.205529864936883e-06,
|
| 9073 |
+
"loss": 0.0,
|
| 9074 |
+
"step": 1295
|
| 9075 |
+
},
|
| 9076 |
+
{
|
| 9077 |
+
"epoch": 0.8388349514563107,
|
| 9078 |
+
"grad_norm": 2.8870774258393794e-05,
|
| 9079 |
+
"learning_rate": 7.149412965555463e-06,
|
| 9080 |
+
"loss": 0.0,
|
| 9081 |
+
"step": 1296
|
| 9082 |
+
},
|
| 9083 |
+
{
|
| 9084 |
+
"epoch": 0.8394822006472492,
|
| 9085 |
+
"grad_norm": 2.9239281502668746e-05,
|
| 9086 |
+
"learning_rate": 7.093498610768001e-06,
|
| 9087 |
+
"loss": 0.0,
|
| 9088 |
+
"step": 1297
|
| 9089 |
+
},
|
| 9090 |
+
{
|
| 9091 |
+
"epoch": 0.8401294498381877,
|
| 9092 |
+
"grad_norm": 2.780013528536074e-05,
|
| 9093 |
+
"learning_rate": 7.0377870648684394e-06,
|
| 9094 |
+
"loss": 0.0,
|
| 9095 |
+
"step": 1298
|
| 9096 |
+
},
|
| 9097 |
+
{
|
| 9098 |
+
"epoch": 0.8407766990291262,
|
| 9099 |
+
"grad_norm": 2.7307174605084583e-05,
|
| 9100 |
+
"learning_rate": 6.9822785911920766e-06,
|
| 9101 |
+
"loss": 0.0,
|
| 9102 |
+
"step": 1299
|
| 9103 |
+
},
|
| 9104 |
+
{
|
| 9105 |
+
"epoch": 0.8414239482200647,
|
| 9106 |
+
"grad_norm": 2.685461367946118e-05,
|
| 9107 |
+
"learning_rate": 6.926973452114338e-06,
|
| 9108 |
+
"loss": 0.0,
|
| 9109 |
+
"step": 1300
|
| 9110 |
+
},
|
| 9111 |
+
{
|
| 9112 |
+
"epoch": 0.8420711974110032,
|
| 9113 |
+
"grad_norm": 3.219491190975532e-05,
|
| 9114 |
+
"learning_rate": 6.871871909049554e-06,
|
| 9115 |
+
"loss": 0.0,
|
| 9116 |
+
"step": 1301
|
| 9117 |
+
},
|
| 9118 |
+
{
|
| 9119 |
+
"epoch": 0.8427184466019417,
|
| 9120 |
+
"grad_norm": 3.353790452820249e-05,
|
| 9121 |
+
"learning_rate": 6.816974222449674e-06,
|
| 9122 |
+
"loss": 0.0,
|
| 9123 |
+
"step": 1302
|
| 9124 |
+
},
|
| 9125 |
+
{
|
| 9126 |
+
"epoch": 0.8433656957928802,
|
| 9127 |
+
"grad_norm": 3.2828796975081787e-05,
|
| 9128 |
+
"learning_rate": 6.762280651803105e-06,
|
| 9129 |
+
"loss": 0.0,
|
| 9130 |
+
"step": 1303
|
| 9131 |
+
},
|
| 9132 |
+
{
|
| 9133 |
+
"epoch": 0.8440129449838187,
|
| 9134 |
+
"grad_norm": 3.161135100526735e-05,
|
| 9135 |
+
"learning_rate": 6.7077914556334e-06,
|
| 9136 |
+
"loss": 0.0,
|
| 9137 |
+
"step": 1304
|
| 9138 |
+
},
|
| 9139 |
+
{
|
| 9140 |
+
"epoch": 0.8446601941747572,
|
| 9141 |
+
"grad_norm": 3.0181634429027326e-05,
|
| 9142 |
+
"learning_rate": 6.653506891498118e-06,
|
| 9143 |
+
"loss": 0.0,
|
| 9144 |
+
"step": 1305
|
| 9145 |
+
},
|
| 9146 |
+
{
|
| 9147 |
+
"epoch": 0.8453074433656957,
|
| 9148 |
+
"grad_norm": 3.0183506169123575e-05,
|
| 9149 |
+
"learning_rate": 6.599427215987574e-06,
|
| 9150 |
+
"loss": 0.0,
|
| 9151 |
+
"step": 1306
|
| 9152 |
+
},
|
| 9153 |
+
{
|
| 9154 |
+
"epoch": 0.8459546925566344,
|
| 9155 |
+
"grad_norm": 2.7570862584980205e-05,
|
| 9156 |
+
"learning_rate": 6.5455526847235825e-06,
|
| 9157 |
+
"loss": 0.0,
|
| 9158 |
+
"step": 1307
|
| 9159 |
+
},
|
| 9160 |
+
{
|
| 9161 |
+
"epoch": 0.8466019417475729,
|
| 9162 |
+
"grad_norm": 2.912186027970165e-05,
|
| 9163 |
+
"learning_rate": 6.49188355235833e-06,
|
| 9164 |
+
"loss": 0.0,
|
| 9165 |
+
"step": 1308
|
| 9166 |
+
},
|
| 9167 |
+
{
|
| 9168 |
+
"epoch": 0.8472491909385114,
|
| 9169 |
+
"grad_norm": 5.2388324547791854e-05,
|
| 9170 |
+
"learning_rate": 6.438420072573126e-06,
|
| 9171 |
+
"loss": 0.0,
|
| 9172 |
+
"step": 1309
|
| 9173 |
+
},
|
| 9174 |
+
{
|
| 9175 |
+
"epoch": 0.8478964401294499,
|
| 9176 |
+
"grad_norm": 2.9694594559259713e-05,
|
| 9177 |
+
"learning_rate": 6.3851624980771905e-06,
|
| 9178 |
+
"loss": 0.0,
|
| 9179 |
+
"step": 1310
|
| 9180 |
+
},
|
| 9181 |
+
{
|
| 9182 |
+
"epoch": 0.8485436893203884,
|
| 9183 |
+
"grad_norm": 3.2102074328577146e-05,
|
| 9184 |
+
"learning_rate": 6.332111080606467e-06,
|
| 9185 |
+
"loss": 0.0,
|
| 9186 |
+
"step": 1311
|
| 9187 |
+
},
|
| 9188 |
+
{
|
| 9189 |
+
"epoch": 0.8491909385113269,
|
| 9190 |
+
"grad_norm": 3.619419294409454e-05,
|
| 9191 |
+
"learning_rate": 6.279266070922496e-06,
|
| 9192 |
+
"loss": 0.0,
|
| 9193 |
+
"step": 1312
|
| 9194 |
+
},
|
| 9195 |
+
{
|
| 9196 |
+
"epoch": 0.8498381877022654,
|
| 9197 |
+
"grad_norm": 3.186412141076289e-05,
|
| 9198 |
+
"learning_rate": 6.226627718811118e-06,
|
| 9199 |
+
"loss": 0.0,
|
| 9200 |
+
"step": 1313
|
| 9201 |
+
},
|
| 9202 |
+
{
|
| 9203 |
+
"epoch": 0.8504854368932039,
|
| 9204 |
+
"grad_norm": 3.2827385439304635e-05,
|
| 9205 |
+
"learning_rate": 6.1741962730814e-06,
|
| 9206 |
+
"loss": 0.0,
|
| 9207 |
+
"step": 1314
|
| 9208 |
+
},
|
| 9209 |
+
{
|
| 9210 |
+
"epoch": 0.8511326860841424,
|
| 9211 |
+
"grad_norm": 2.8862363251391798e-05,
|
| 9212 |
+
"learning_rate": 6.121971981564367e-06,
|
| 9213 |
+
"loss": 0.0,
|
| 9214 |
+
"step": 1315
|
| 9215 |
+
},
|
| 9216 |
+
{
|
| 9217 |
+
"epoch": 0.8517799352750809,
|
| 9218 |
+
"grad_norm": 2.9111837648088112e-05,
|
| 9219 |
+
"learning_rate": 6.069955091111912e-06,
|
| 9220 |
+
"loss": 0.0,
|
| 9221 |
+
"step": 1316
|
| 9222 |
+
},
|
| 9223 |
+
{
|
| 9224 |
+
"epoch": 0.8524271844660194,
|
| 9225 |
+
"grad_norm": 2.9940256354166195e-05,
|
| 9226 |
+
"learning_rate": 6.018145847595585e-06,
|
| 9227 |
+
"loss": 0.0,
|
| 9228 |
+
"step": 1317
|
| 9229 |
+
},
|
| 9230 |
+
{
|
| 9231 |
+
"epoch": 0.8530744336569579,
|
| 9232 |
+
"grad_norm": 3.0056986361159943e-05,
|
| 9233 |
+
"learning_rate": 5.966544495905408e-06,
|
| 9234 |
+
"loss": 0.0,
|
| 9235 |
+
"step": 1318
|
| 9236 |
+
},
|
| 9237 |
+
{
|
| 9238 |
+
"epoch": 0.8537216828478964,
|
| 9239 |
+
"grad_norm": 3.307464066892862e-05,
|
| 9240 |
+
"learning_rate": 5.915151279948788e-06,
|
| 9241 |
+
"loss": 0.0,
|
| 9242 |
+
"step": 1319
|
| 9243 |
+
},
|
| 9244 |
+
{
|
| 9245 |
+
"epoch": 0.8543689320388349,
|
| 9246 |
+
"grad_norm": 3.189213020959869e-05,
|
| 9247 |
+
"learning_rate": 5.863966442649327e-06,
|
| 9248 |
+
"loss": 0.0,
|
| 9249 |
+
"step": 1320
|
| 9250 |
+
},
|
| 9251 |
+
{
|
| 9252 |
+
"epoch": 0.8550161812297734,
|
| 9253 |
+
"grad_norm": 2.996783223352395e-05,
|
| 9254 |
+
"learning_rate": 5.812990225945603e-06,
|
| 9255 |
+
"loss": 0.0,
|
| 9256 |
+
"step": 1321
|
| 9257 |
+
},
|
| 9258 |
+
{
|
| 9259 |
+
"epoch": 0.855663430420712,
|
| 9260 |
+
"grad_norm": 3.2352731068385765e-05,
|
| 9261 |
+
"learning_rate": 5.762222870790163e-06,
|
| 9262 |
+
"loss": 0.0,
|
| 9263 |
+
"step": 1322
|
| 9264 |
+
},
|
| 9265 |
+
{
|
| 9266 |
+
"epoch": 0.8563106796116505,
|
| 9267 |
+
"grad_norm": 3.427658521104604e-05,
|
| 9268 |
+
"learning_rate": 5.711664617148299e-06,
|
| 9269 |
+
"loss": 0.0,
|
| 9270 |
+
"step": 1323
|
| 9271 |
+
},
|
| 9272 |
+
{
|
| 9273 |
+
"epoch": 0.856957928802589,
|
| 9274 |
+
"grad_norm": 3.427062256378122e-05,
|
| 9275 |
+
"learning_rate": 5.6613157039969055e-06,
|
| 9276 |
+
"loss": 0.0,
|
| 9277 |
+
"step": 1324
|
| 9278 |
+
},
|
| 9279 |
+
{
|
| 9280 |
+
"epoch": 0.8576051779935275,
|
| 9281 |
+
"grad_norm": 3.057840876863338e-05,
|
| 9282 |
+
"learning_rate": 5.611176369323412e-06,
|
| 9283 |
+
"loss": 0.0,
|
| 9284 |
+
"step": 1325
|
| 9285 |
+
},
|
| 9286 |
+
{
|
| 9287 |
+
"epoch": 0.858252427184466,
|
| 9288 |
+
"grad_norm": 3.3830769098130986e-05,
|
| 9289 |
+
"learning_rate": 5.56124685012458e-06,
|
| 9290 |
+
"loss": 0.0,
|
| 9291 |
+
"step": 1326
|
| 9292 |
+
},
|
| 9293 |
+
{
|
| 9294 |
+
"epoch": 0.8588996763754045,
|
| 9295 |
+
"grad_norm": 3.0684615921927616e-05,
|
| 9296 |
+
"learning_rate": 5.511527382405451e-06,
|
| 9297 |
+
"loss": 0.0,
|
| 9298 |
+
"step": 1327
|
| 9299 |
+
},
|
| 9300 |
+
{
|
| 9301 |
+
"epoch": 0.859546925566343,
|
| 9302 |
+
"grad_norm": 2.877054976124782e-05,
|
| 9303 |
+
"learning_rate": 5.462018201178204e-06,
|
| 9304 |
+
"loss": 0.0,
|
| 9305 |
+
"step": 1328
|
| 9306 |
+
},
|
| 9307 |
+
{
|
| 9308 |
+
"epoch": 0.8601941747572815,
|
| 9309 |
+
"grad_norm": 2.8292075512581505e-05,
|
| 9310 |
+
"learning_rate": 5.412719540461015e-06,
|
| 9311 |
+
"loss": 0.0,
|
| 9312 |
+
"step": 1329
|
| 9313 |
+
},
|
| 9314 |
+
{
|
| 9315 |
+
"epoch": 0.86084142394822,
|
| 9316 |
+
"grad_norm": 4.5770942961098626e-05,
|
| 9317 |
+
"learning_rate": 5.363631633277006e-06,
|
| 9318 |
+
"loss": 0.0,
|
| 9319 |
+
"step": 1330
|
| 9320 |
+
},
|
| 9321 |
+
{
|
| 9322 |
+
"epoch": 0.8614886731391586,
|
| 9323 |
+
"grad_norm": 2.9364460715441965e-05,
|
| 9324 |
+
"learning_rate": 5.314754711653125e-06,
|
| 9325 |
+
"loss": 0.0,
|
| 9326 |
+
"step": 1331
|
| 9327 |
+
},
|
| 9328 |
+
{
|
| 9329 |
+
"epoch": 0.8621359223300971,
|
| 9330 |
+
"grad_norm": 2.9971184630994685e-05,
|
| 9331 |
+
"learning_rate": 5.266089006618991e-06,
|
| 9332 |
+
"loss": 0.0,
|
| 9333 |
+
"step": 1332
|
| 9334 |
+
},
|
| 9335 |
+
{
|
| 9336 |
+
"epoch": 0.8627831715210356,
|
| 9337 |
+
"grad_norm": 3.090947211603634e-05,
|
| 9338 |
+
"learning_rate": 5.217634748205902e-06,
|
| 9339 |
+
"loss": 0.0,
|
| 9340 |
+
"step": 1333
|
| 9341 |
+
},
|
| 9342 |
+
{
|
| 9343 |
+
"epoch": 0.8634304207119741,
|
| 9344 |
+
"grad_norm": 2.972232505271677e-05,
|
| 9345 |
+
"learning_rate": 5.169392165445691e-06,
|
| 9346 |
+
"loss": 0.0,
|
| 9347 |
+
"step": 1334
|
| 9348 |
+
},
|
| 9349 |
+
{
|
| 9350 |
+
"epoch": 0.8640776699029126,
|
| 9351 |
+
"grad_norm": 2.7802563636214472e-05,
|
| 9352 |
+
"learning_rate": 5.121361486369625e-06,
|
| 9353 |
+
"loss": 0.0,
|
| 9354 |
+
"step": 1335
|
| 9355 |
+
},
|
| 9356 |
+
{
|
| 9357 |
+
"epoch": 0.8647249190938512,
|
| 9358 |
+
"grad_norm": 2.7798199880635366e-05,
|
| 9359 |
+
"learning_rate": 5.0735429380073865e-06,
|
| 9360 |
+
"loss": 0.0,
|
| 9361 |
+
"step": 1336
|
| 9362 |
+
},
|
| 9363 |
+
{
|
| 9364 |
+
"epoch": 0.8653721682847897,
|
| 9365 |
+
"grad_norm": 2.84150373772718e-05,
|
| 9366 |
+
"learning_rate": 5.025936746385928e-06,
|
| 9367 |
+
"loss": 0.0,
|
| 9368 |
+
"step": 1337
|
| 9369 |
+
},
|
| 9370 |
+
{
|
| 9371 |
+
"epoch": 0.8660194174757282,
|
| 9372 |
+
"grad_norm": 2.972891707031522e-05,
|
| 9373 |
+
"learning_rate": 4.978543136528474e-06,
|
| 9374 |
+
"loss": 0.0,
|
| 9375 |
+
"step": 1338
|
| 9376 |
+
},
|
| 9377 |
+
{
|
| 9378 |
+
"epoch": 0.8666666666666667,
|
| 9379 |
+
"grad_norm": 2.9013235689490102e-05,
|
| 9380 |
+
"learning_rate": 4.931362332453421e-06,
|
| 9381 |
+
"loss": 0.0,
|
| 9382 |
+
"step": 1339
|
| 9383 |
+
},
|
| 9384 |
+
{
|
| 9385 |
+
"epoch": 0.8673139158576052,
|
| 9386 |
+
"grad_norm": 2.887454320443794e-05,
|
| 9387 |
+
"learning_rate": 4.884394557173249e-06,
|
| 9388 |
+
"loss": 0.0,
|
| 9389 |
+
"step": 1340
|
| 9390 |
+
},
|
| 9391 |
+
{
|
| 9392 |
+
"epoch": 0.8679611650485437,
|
| 9393 |
+
"grad_norm": 3.425277827773243e-05,
|
| 9394 |
+
"learning_rate": 4.837640032693558e-06,
|
| 9395 |
+
"loss": 0.0,
|
| 9396 |
+
"step": 1341
|
| 9397 |
+
},
|
| 9398 |
+
{
|
| 9399 |
+
"epoch": 0.8686084142394822,
|
| 9400 |
+
"grad_norm": 2.8632548492169008e-05,
|
| 9401 |
+
"learning_rate": 4.791098980011921e-06,
|
| 9402 |
+
"loss": 0.0,
|
| 9403 |
+
"step": 1342
|
| 9404 |
+
},
|
| 9405 |
+
{
|
| 9406 |
+
"epoch": 0.8692556634304207,
|
| 9407 |
+
"grad_norm": 2.853093610610813e-05,
|
| 9408 |
+
"learning_rate": 4.744771619116872e-06,
|
| 9409 |
+
"loss": 0.0,
|
| 9410 |
+
"step": 1343
|
| 9411 |
+
},
|
| 9412 |
+
{
|
| 9413 |
+
"epoch": 0.8699029126213592,
|
| 9414 |
+
"grad_norm": 2.9239194191177376e-05,
|
| 9415 |
+
"learning_rate": 4.698658168986908e-06,
|
| 9416 |
+
"loss": 0.0,
|
| 9417 |
+
"step": 1344
|
| 9418 |
+
},
|
| 9419 |
+
{
|
| 9420 |
+
"epoch": 0.8705501618122977,
|
| 9421 |
+
"grad_norm": 2.6730338504421525e-05,
|
| 9422 |
+
"learning_rate": 4.652758847589416e-06,
|
| 9423 |
+
"loss": 0.0,
|
| 9424 |
+
"step": 1345
|
| 9425 |
+
},
|
| 9426 |
+
{
|
| 9427 |
+
"epoch": 0.8711974110032362,
|
| 9428 |
+
"grad_norm": 2.781052353384439e-05,
|
| 9429 |
+
"learning_rate": 4.607073871879613e-06,
|
| 9430 |
+
"loss": 0.0,
|
| 9431 |
+
"step": 1346
|
| 9432 |
+
},
|
| 9433 |
+
{
|
| 9434 |
+
"epoch": 0.8718446601941747,
|
| 9435 |
+
"grad_norm": 2.9233944587758742e-05,
|
| 9436 |
+
"learning_rate": 4.5616034577995955e-06,
|
| 9437 |
+
"loss": 0.0,
|
| 9438 |
+
"step": 1347
|
| 9439 |
+
},
|
| 9440 |
+
{
|
| 9441 |
+
"epoch": 0.8724919093851132,
|
| 9442 |
+
"grad_norm": 2.815485822793562e-05,
|
| 9443 |
+
"learning_rate": 4.5163478202772615e-06,
|
| 9444 |
+
"loss": 0.0,
|
| 9445 |
+
"step": 1348
|
| 9446 |
+
},
|
| 9447 |
+
{
|
| 9448 |
+
"epoch": 0.8731391585760517,
|
| 9449 |
+
"grad_norm": 2.8653046683757566e-05,
|
| 9450 |
+
"learning_rate": 4.471307173225292e-06,
|
| 9451 |
+
"loss": 0.0,
|
| 9452 |
+
"step": 1349
|
| 9453 |
+
},
|
| 9454 |
+
{
|
| 9455 |
+
"epoch": 0.8737864077669902,
|
| 9456 |
+
"grad_norm": 3.0307597626233473e-05,
|
| 9457 |
+
"learning_rate": 4.426481729540205e-06,
|
| 9458 |
+
"loss": 0.0,
|
| 9459 |
+
"step": 1350
|
| 9460 |
+
},
|
| 9461 |
+
{
|
| 9462 |
+
"epoch": 0.8744336569579289,
|
| 9463 |
+
"grad_norm": 3.496236968203448e-05,
|
| 9464 |
+
"learning_rate": 4.381871701101248e-06,
|
| 9465 |
+
"loss": 0.0,
|
| 9466 |
+
"step": 1351
|
| 9467 |
+
},
|
| 9468 |
+
{
|
| 9469 |
+
"epoch": 0.8750809061488674,
|
| 9470 |
+
"grad_norm": 3.211486182408407e-05,
|
| 9471 |
+
"learning_rate": 4.337477298769493e-06,
|
| 9472 |
+
"loss": 0.0,
|
| 9473 |
+
"step": 1352
|
| 9474 |
+
},
|
| 9475 |
+
{
|
| 9476 |
+
"epoch": 0.8757281553398059,
|
| 9477 |
+
"grad_norm": 2.8743703296640888e-05,
|
| 9478 |
+
"learning_rate": 4.293298732386786e-06,
|
| 9479 |
+
"loss": 0.0,
|
| 9480 |
+
"step": 1353
|
| 9481 |
+
},
|
| 9482 |
+
{
|
| 9483 |
+
"epoch": 0.8763754045307444,
|
| 9484 |
+
"grad_norm": 3.354837099323049e-05,
|
| 9485 |
+
"learning_rate": 4.249336210774746e-06,
|
| 9486 |
+
"loss": 0.0,
|
| 9487 |
+
"step": 1354
|
| 9488 |
+
},
|
| 9489 |
+
{
|
| 9490 |
+
"epoch": 0.8770226537216829,
|
| 9491 |
+
"grad_norm": 3.329403989482671e-05,
|
| 9492 |
+
"learning_rate": 4.205589941733834e-06,
|
| 9493 |
+
"loss": 0.0,
|
| 9494 |
+
"step": 1355
|
| 9495 |
+
},
|
| 9496 |
+
{
|
| 9497 |
+
"epoch": 0.8776699029126214,
|
| 9498 |
+
"grad_norm": 2.9112465199432336e-05,
|
| 9499 |
+
"learning_rate": 4.162060132042333e-06,
|
| 9500 |
+
"loss": 0.0,
|
| 9501 |
+
"step": 1356
|
| 9502 |
+
},
|
| 9503 |
+
{
|
| 9504 |
+
"epoch": 0.8783171521035599,
|
| 9505 |
+
"grad_norm": 3.116092921118252e-05,
|
| 9506 |
+
"learning_rate": 4.118746987455336e-06,
|
| 9507 |
+
"loss": 0.0,
|
| 9508 |
+
"step": 1357
|
| 9509 |
+
},
|
| 9510 |
+
{
|
| 9511 |
+
"epoch": 0.8789644012944984,
|
| 9512 |
+
"grad_norm": 3.0895193049218506e-05,
|
| 9513 |
+
"learning_rate": 4.075650712703849e-06,
|
| 9514 |
+
"loss": 0.0,
|
| 9515 |
+
"step": 1358
|
| 9516 |
+
},
|
| 9517 |
+
{
|
| 9518 |
+
"epoch": 0.8796116504854369,
|
| 9519 |
+
"grad_norm": 3.210348222637549e-05,
|
| 9520 |
+
"learning_rate": 4.032771511493782e-06,
|
| 9521 |
+
"loss": 0.0,
|
| 9522 |
+
"step": 1359
|
| 9523 |
+
},
|
| 9524 |
+
{
|
| 9525 |
+
"epoch": 0.8802588996763754,
|
| 9526 |
+
"grad_norm": 3.0547904316335917e-05,
|
| 9527 |
+
"learning_rate": 3.990109586504965e-06,
|
| 9528 |
+
"loss": 0.0,
|
| 9529 |
+
"step": 1360
|
| 9530 |
+
},
|
| 9531 |
+
{
|
| 9532 |
+
"epoch": 0.8809061488673139,
|
| 9533 |
+
"grad_norm": 2.946665154013317e-05,
|
| 9534 |
+
"learning_rate": 3.9476651393902494e-06,
|
| 9535 |
+
"loss": 0.0,
|
| 9536 |
+
"step": 1361
|
| 9537 |
+
},
|
| 9538 |
+
{
|
| 9539 |
+
"epoch": 0.8815533980582524,
|
| 9540 |
+
"grad_norm": 3.0193479688023217e-05,
|
| 9541 |
+
"learning_rate": 3.905438370774495e-06,
|
| 9542 |
+
"loss": 0.0,
|
| 9543 |
+
"step": 1362
|
| 9544 |
+
},
|
| 9545 |
+
{
|
| 9546 |
+
"epoch": 0.8822006472491909,
|
| 9547 |
+
"grad_norm": 3.0318025892484002e-05,
|
| 9548 |
+
"learning_rate": 3.8634294802536384e-06,
|
| 9549 |
+
"loss": 0.0,
|
| 9550 |
+
"step": 1363
|
| 9551 |
+
},
|
| 9552 |
+
{
|
| 9553 |
+
"epoch": 0.8828478964401294,
|
| 9554 |
+
"grad_norm": 3.162304710713215e-05,
|
| 9555 |
+
"learning_rate": 3.82163866639379e-06,
|
| 9556 |
+
"loss": 0.0,
|
| 9557 |
+
"step": 1364
|
| 9558 |
+
},
|
| 9559 |
+
{
|
| 9560 |
+
"epoch": 0.883495145631068,
|
| 9561 |
+
"grad_norm": 3.113932689302601e-05,
|
| 9562 |
+
"learning_rate": 3.7800661267302417e-06,
|
| 9563 |
+
"loss": 0.0,
|
| 9564 |
+
"step": 1365
|
| 9565 |
+
},
|
| 9566 |
+
{
|
| 9567 |
+
"epoch": 0.8841423948220065,
|
| 9568 |
+
"grad_norm": 3.0186189178493805e-05,
|
| 9569 |
+
"learning_rate": 3.7387120577665524e-06,
|
| 9570 |
+
"loss": 0.0,
|
| 9571 |
+
"step": 1366
|
| 9572 |
+
},
|
| 9573 |
+
{
|
| 9574 |
+
"epoch": 0.884789644012945,
|
| 9575 |
+
"grad_norm": 3.017430026375223e-05,
|
| 9576 |
+
"learning_rate": 3.69757665497365e-06,
|
| 9577 |
+
"loss": 0.0,
|
| 9578 |
+
"step": 1367
|
| 9579 |
+
},
|
| 9580 |
+
{
|
| 9581 |
+
"epoch": 0.8854368932038835,
|
| 9582 |
+
"grad_norm": 3.1866406061453745e-05,
|
| 9583 |
+
"learning_rate": 3.6566601127888344e-06,
|
| 9584 |
+
"loss": 0.0,
|
| 9585 |
+
"step": 1368
|
| 9586 |
+
},
|
| 9587 |
+
{
|
| 9588 |
+
"epoch": 0.886084142394822,
|
| 9589 |
+
"grad_norm": 2.9112245101714507e-05,
|
| 9590 |
+
"learning_rate": 3.615962624614938e-06,
|
| 9591 |
+
"loss": 0.0,
|
| 9592 |
+
"step": 1369
|
| 9593 |
+
},
|
| 9594 |
+
{
|
| 9595 |
+
"epoch": 0.8867313915857605,
|
| 9596 |
+
"grad_norm": 2.9124956199666485e-05,
|
| 9597 |
+
"learning_rate": 3.5754843828193716e-06,
|
| 9598 |
+
"loss": 0.0,
|
| 9599 |
+
"step": 1370
|
| 9600 |
+
},
|
| 9601 |
+
{
|
| 9602 |
+
"epoch": 0.887378640776699,
|
| 9603 |
+
"grad_norm": 3.236894553992897e-05,
|
| 9604 |
+
"learning_rate": 3.535225578733198e-06,
|
| 9605 |
+
"loss": 0.0,
|
| 9606 |
+
"step": 1371
|
| 9607 |
+
},
|
| 9608 |
+
{
|
| 9609 |
+
"epoch": 0.8880258899676375,
|
| 9610 |
+
"grad_norm": 3.188128903275356e-05,
|
| 9611 |
+
"learning_rate": 3.4951864026502854e-06,
|
| 9612 |
+
"loss": 0.0,
|
| 9613 |
+
"step": 1372
|
| 9614 |
+
},
|
| 9615 |
+
{
|
| 9616 |
+
"epoch": 0.888673139158576,
|
| 9617 |
+
"grad_norm": 3.475479024928063e-05,
|
| 9618 |
+
"learning_rate": 3.4553670438263408e-06,
|
| 9619 |
+
"loss": 0.0,
|
| 9620 |
+
"step": 1373
|
| 9621 |
+
},
|
| 9622 |
+
{
|
| 9623 |
+
"epoch": 0.8893203883495145,
|
| 9624 |
+
"grad_norm": 3.189170456607826e-05,
|
| 9625 |
+
"learning_rate": 3.41576769047805e-06,
|
| 9626 |
+
"loss": 0.0,
|
| 9627 |
+
"step": 1374
|
| 9628 |
+
},
|
| 9629 |
+
{
|
| 9630 |
+
"epoch": 0.889967637540453,
|
| 9631 |
+
"grad_norm": 3.0447605240624398e-05,
|
| 9632 |
+
"learning_rate": 3.376388529782215e-06,
|
| 9633 |
+
"loss": 0.0,
|
| 9634 |
+
"step": 1375
|
| 9635 |
+
},
|
| 9636 |
+
{
|
| 9637 |
+
"epoch": 0.8906148867313916,
|
| 9638 |
+
"grad_norm": 3.116185325779952e-05,
|
| 9639 |
+
"learning_rate": 3.3372297478748038e-06,
|
| 9640 |
+
"loss": 0.0,
|
| 9641 |
+
"step": 1376
|
| 9642 |
+
},
|
| 9643 |
+
{
|
| 9644 |
+
"epoch": 0.8912621359223301,
|
| 9645 |
+
"grad_norm": 2.9720120437559672e-05,
|
| 9646 |
+
"learning_rate": 3.2982915298501173e-06,
|
| 9647 |
+
"loss": 0.0,
|
| 9648 |
+
"step": 1377
|
| 9649 |
+
},
|
| 9650 |
+
{
|
| 9651 |
+
"epoch": 0.8919093851132686,
|
| 9652 |
+
"grad_norm": 3.618436312535778e-05,
|
| 9653 |
+
"learning_rate": 3.2595740597599234e-06,
|
| 9654 |
+
"loss": 0.0,
|
| 9655 |
+
"step": 1378
|
| 9656 |
+
},
|
| 9657 |
+
{
|
| 9658 |
+
"epoch": 0.8925566343042072,
|
| 9659 |
+
"grad_norm": 2.9376269594649784e-05,
|
| 9660 |
+
"learning_rate": 3.221077520612531e-06,
|
| 9661 |
+
"loss": 0.0,
|
| 9662 |
+
"step": 1379
|
| 9663 |
+
},
|
| 9664 |
+
{
|
| 9665 |
+
"epoch": 0.8932038834951457,
|
| 9666 |
+
"grad_norm": 3.2120238756760955e-05,
|
| 9667 |
+
"learning_rate": 3.1828020943719894e-06,
|
| 9668 |
+
"loss": 0.0,
|
| 9669 |
+
"step": 1380
|
| 9670 |
+
},
|
| 9671 |
+
{
|
| 9672 |
+
"epoch": 0.8938511326860842,
|
| 9673 |
+
"grad_norm": 3.238513454562053e-05,
|
| 9674 |
+
"learning_rate": 3.144747961957195e-06,
|
| 9675 |
+
"loss": 0.0,
|
| 9676 |
+
"step": 1381
|
| 9677 |
+
},
|
| 9678 |
+
{
|
| 9679 |
+
"epoch": 0.8944983818770227,
|
| 9680 |
+
"grad_norm": 3.115571234957315e-05,
|
| 9681 |
+
"learning_rate": 3.1069153032410147e-06,
|
| 9682 |
+
"loss": 0.0,
|
| 9683 |
+
"step": 1382
|
| 9684 |
+
},
|
| 9685 |
+
{
|
| 9686 |
+
"epoch": 0.8951456310679612,
|
| 9687 |
+
"grad_norm": 2.8277658202569e-05,
|
| 9688 |
+
"learning_rate": 3.069304297049508e-06,
|
| 9689 |
+
"loss": 0.0,
|
| 9690 |
+
"step": 1383
|
| 9691 |
+
},
|
| 9692 |
+
{
|
| 9693 |
+
"epoch": 0.8957928802588997,
|
| 9694 |
+
"grad_norm": 2.8890643079648726e-05,
|
| 9695 |
+
"learning_rate": 3.0319151211609886e-06,
|
| 9696 |
+
"loss": 0.0,
|
| 9697 |
+
"step": 1384
|
| 9698 |
+
},
|
| 9699 |
+
{
|
| 9700 |
+
"epoch": 0.8964401294498382,
|
| 9701 |
+
"grad_norm": 3.032093809451908e-05,
|
| 9702 |
+
"learning_rate": 2.9947479523052548e-06,
|
| 9703 |
+
"loss": 0.0,
|
| 9704 |
+
"step": 1385
|
| 9705 |
+
},
|
| 9706 |
+
{
|
| 9707 |
+
"epoch": 0.8970873786407767,
|
| 9708 |
+
"grad_norm": 3.0906725442036986e-05,
|
| 9709 |
+
"learning_rate": 2.9578029661627314e-06,
|
| 9710 |
+
"loss": 0.0,
|
| 9711 |
+
"step": 1386
|
| 9712 |
+
},
|
| 9713 |
+
{
|
| 9714 |
+
"epoch": 0.8977346278317152,
|
| 9715 |
+
"grad_norm": 3.1875537388259545e-05,
|
| 9716 |
+
"learning_rate": 2.921080337363624e-06,
|
| 9717 |
+
"loss": 0.0,
|
| 9718 |
+
"step": 1387
|
| 9719 |
+
},
|
| 9720 |
+
{
|
| 9721 |
+
"epoch": 0.8983818770226537,
|
| 9722 |
+
"grad_norm": 2.8284761356189847e-05,
|
| 9723 |
+
"learning_rate": 2.884580239487128e-06,
|
| 9724 |
+
"loss": 0.0,
|
| 9725 |
+
"step": 1388
|
| 9726 |
+
},
|
| 9727 |
+
{
|
| 9728 |
+
"epoch": 0.8990291262135922,
|
| 9729 |
+
"grad_norm": 2.9141221602912992e-05,
|
| 9730 |
+
"learning_rate": 2.8483028450605742e-06,
|
| 9731 |
+
"loss": 0.0,
|
| 9732 |
+
"step": 1389
|
| 9733 |
+
},
|
| 9734 |
+
{
|
| 9735 |
+
"epoch": 0.8996763754045307,
|
| 9736 |
+
"grad_norm": 2.780696377158165e-05,
|
| 9737 |
+
"learning_rate": 2.8122483255586252e-06,
|
| 9738 |
+
"loss": 0.0,
|
| 9739 |
+
"step": 1390
|
| 9740 |
+
},
|
| 9741 |
+
{
|
| 9742 |
+
"epoch": 0.9003236245954692,
|
| 9743 |
+
"grad_norm": 3.0303663152153604e-05,
|
| 9744 |
+
"learning_rate": 2.776416851402469e-06,
|
| 9745 |
+
"loss": 0.0,
|
| 9746 |
+
"step": 1391
|
| 9747 |
+
},
|
| 9748 |
+
{
|
| 9749 |
+
"epoch": 0.9009708737864077,
|
| 9750 |
+
"grad_norm": 2.9251741580083035e-05,
|
| 9751 |
+
"learning_rate": 2.7408085919590264e-06,
|
| 9752 |
+
"loss": 0.0,
|
| 9753 |
+
"step": 1392
|
| 9754 |
+
},
|
| 9755 |
+
{
|
| 9756 |
+
"epoch": 0.9016181229773462,
|
| 9757 |
+
"grad_norm": 2.935952397820074e-05,
|
| 9758 |
+
"learning_rate": 2.705423715540101e-06,
|
| 9759 |
+
"loss": 0.0,
|
| 9760 |
+
"step": 1393
|
| 9761 |
+
},
|
| 9762 |
+
{
|
| 9763 |
+
"epoch": 0.9022653721682848,
|
| 9764 |
+
"grad_norm": 3.0076322218519635e-05,
|
| 9765 |
+
"learning_rate": 2.670262389401651e-06,
|
| 9766 |
+
"loss": 0.0,
|
| 9767 |
+
"step": 1394
|
| 9768 |
+
},
|
| 9769 |
+
{
|
| 9770 |
+
"epoch": 0.9029126213592233,
|
| 9771 |
+
"grad_norm": 2.7805677746073343e-05,
|
| 9772 |
+
"learning_rate": 2.6353247797429535e-06,
|
| 9773 |
+
"loss": 0.0,
|
| 9774 |
+
"step": 1395
|
| 9775 |
+
},
|
| 9776 |
+
{
|
| 9777 |
+
"epoch": 0.9035598705501618,
|
| 9778 |
+
"grad_norm": 2.8380973162711598e-05,
|
| 9779 |
+
"learning_rate": 2.6006110517058144e-06,
|
| 9780 |
+
"loss": 0.0,
|
| 9781 |
+
"step": 1396
|
| 9782 |
+
},
|
| 9783 |
+
{
|
| 9784 |
+
"epoch": 0.9042071197411004,
|
| 9785 |
+
"grad_norm": 2.719420808716677e-05,
|
| 9786 |
+
"learning_rate": 2.566121369373836e-06,
|
| 9787 |
+
"loss": 0.0,
|
| 9788 |
+
"step": 1397
|
| 9789 |
+
},
|
| 9790 |
+
{
|
| 9791 |
+
"epoch": 0.9048543689320389,
|
| 9792 |
+
"grad_norm": 2.673549715836998e-05,
|
| 9793 |
+
"learning_rate": 2.531855895771579e-06,
|
| 9794 |
+
"loss": 0.0,
|
| 9795 |
+
"step": 1398
|
| 9796 |
+
},
|
| 9797 |
+
{
|
| 9798 |
+
"epoch": 0.9055016181229774,
|
| 9799 |
+
"grad_norm": 3.116455263807438e-05,
|
| 9800 |
+
"learning_rate": 2.4978147928638397e-06,
|
| 9801 |
+
"loss": 0.0,
|
| 9802 |
+
"step": 1399
|
| 9803 |
+
},
|
| 9804 |
+
{
|
| 9805 |
+
"epoch": 0.9061488673139159,
|
| 9806 |
+
"grad_norm": 2.9239681680337526e-05,
|
| 9807 |
+
"learning_rate": 2.463998221554875e-06,
|
| 9808 |
+
"loss": 0.0,
|
| 9809 |
+
"step": 1400
|
| 9810 |
+
},
|
| 9811 |
+
{
|
| 9812 |
+
"epoch": 0.9067961165048544,
|
| 9813 |
+
"grad_norm": 3.38866411766503e-05,
|
| 9814 |
+
"learning_rate": 2.430406341687608e-06,
|
| 9815 |
+
"loss": 0.0,
|
| 9816 |
+
"step": 1401
|
| 9817 |
+
},
|
| 9818 |
+
{
|
| 9819 |
+
"epoch": 0.9074433656957929,
|
| 9820 |
+
"grad_norm": 3.031195592484437e-05,
|
| 9821 |
+
"learning_rate": 2.3970393120429145e-06,
|
| 9822 |
+
"loss": 0.0,
|
| 9823 |
+
"step": 1402
|
| 9824 |
+
},
|
| 9825 |
+
{
|
| 9826 |
+
"epoch": 0.9080906148867314,
|
| 9827 |
+
"grad_norm": 3.054170520044863e-05,
|
| 9828 |
+
"learning_rate": 2.363897290338868e-06,
|
| 9829 |
+
"loss": 0.0,
|
| 9830 |
+
"step": 1403
|
| 9831 |
+
},
|
| 9832 |
+
{
|
| 9833 |
+
"epoch": 0.9087378640776699,
|
| 9834 |
+
"grad_norm": 2.8882692276965827e-05,
|
| 9835 |
+
"learning_rate": 2.3309804332299566e-06,
|
| 9836 |
+
"loss": 0.0,
|
| 9837 |
+
"step": 1404
|
| 9838 |
+
},
|
| 9839 |
+
{
|
| 9840 |
+
"epoch": 0.9093851132686084,
|
| 9841 |
+
"grad_norm": 2.8627364372368902e-05,
|
| 9842 |
+
"learning_rate": 2.2982888963063774e-06,
|
| 9843 |
+
"loss": 0.0,
|
| 9844 |
+
"step": 1405
|
| 9845 |
+
},
|
| 9846 |
+
{
|
| 9847 |
+
"epoch": 0.9100323624595469,
|
| 9848 |
+
"grad_norm": 2.7794889319920912e-05,
|
| 9849 |
+
"learning_rate": 2.2658228340933117e-06,
|
| 9850 |
+
"loss": 0.0,
|
| 9851 |
+
"step": 1406
|
| 9852 |
+
},
|
| 9853 |
+
{
|
| 9854 |
+
"epoch": 0.9106796116504854,
|
| 9855 |
+
"grad_norm": 3.1401821615872905e-05,
|
| 9856 |
+
"learning_rate": 2.2335824000501437e-06,
|
| 9857 |
+
"loss": 0.0,
|
| 9858 |
+
"step": 1407
|
| 9859 |
+
},
|
| 9860 |
+
{
|
| 9861 |
+
"epoch": 0.911326860841424,
|
| 9862 |
+
"grad_norm": 2.909961222030688e-05,
|
| 9863 |
+
"learning_rate": 2.201567746569794e-06,
|
| 9864 |
+
"loss": 0.0,
|
| 9865 |
+
"step": 1408
|
| 9866 |
+
},
|
| 9867 |
+
{
|
| 9868 |
+
"epoch": 0.9119741100323625,
|
| 9869 |
+
"grad_norm": 3.055423076148145e-05,
|
| 9870 |
+
"learning_rate": 2.1697790249779636e-06,
|
| 9871 |
+
"loss": 0.0,
|
| 9872 |
+
"step": 1409
|
| 9873 |
+
},
|
| 9874 |
+
{
|
| 9875 |
+
"epoch": 0.912621359223301,
|
| 9876 |
+
"grad_norm": 2.8502765417215414e-05,
|
| 9877 |
+
"learning_rate": 2.13821638553241e-06,
|
| 9878 |
+
"loss": 0.0,
|
| 9879 |
+
"step": 1410
|
| 9880 |
+
},
|
| 9881 |
+
{
|
| 9882 |
+
"epoch": 0.9132686084142395,
|
| 9883 |
+
"grad_norm": 3.2103438570629805e-05,
|
| 9884 |
+
"learning_rate": 2.106879977422277e-06,
|
| 9885 |
+
"loss": 0.0,
|
| 9886 |
+
"step": 1411
|
| 9887 |
+
},
|
| 9888 |
+
{
|
| 9889 |
+
"epoch": 0.913915857605178,
|
| 9890 |
+
"grad_norm": 3.0755229090573266e-05,
|
| 9891 |
+
"learning_rate": 2.0757699487673533e-06,
|
| 9892 |
+
"loss": 0.0,
|
| 9893 |
+
"step": 1412
|
| 9894 |
+
},
|
| 9895 |
+
{
|
| 9896 |
+
"epoch": 0.9145631067961165,
|
| 9897 |
+
"grad_norm": 2.8395124900271185e-05,
|
| 9898 |
+
"learning_rate": 2.044886446617389e-06,
|
| 9899 |
+
"loss": 0.0,
|
| 9900 |
+
"step": 1413
|
| 9901 |
+
},
|
| 9902 |
+
{
|
| 9903 |
+
"epoch": 0.915210355987055,
|
| 9904 |
+
"grad_norm": 3.307309452793561e-05,
|
| 9905 |
+
"learning_rate": 2.0142296169514073e-06,
|
| 9906 |
+
"loss": 0.0,
|
| 9907 |
+
"step": 1414
|
| 9908 |
+
},
|
| 9909 |
+
{
|
| 9910 |
+
"epoch": 0.9158576051779935,
|
| 9911 |
+
"grad_norm": 2.995224713231437e-05,
|
| 9912 |
+
"learning_rate": 1.9837996046769837e-06,
|
| 9913 |
+
"loss": 0.0,
|
| 9914 |
+
"step": 1415
|
| 9915 |
+
},
|
| 9916 |
+
{
|
| 9917 |
+
"epoch": 0.916504854368932,
|
| 9918 |
+
"grad_norm": 3.0005847293068655e-05,
|
| 9919 |
+
"learning_rate": 1.9535965536295885e-06,
|
| 9920 |
+
"loss": 0.0,
|
| 9921 |
+
"step": 1416
|
| 9922 |
+
},
|
| 9923 |
+
{
|
| 9924 |
+
"epoch": 0.9171521035598705,
|
| 9925 |
+
"grad_norm": 3.1141447834670544e-05,
|
| 9926 |
+
"learning_rate": 1.923620606571919e-06,
|
| 9927 |
+
"loss": 0.0,
|
| 9928 |
+
"step": 1417
|
| 9929 |
+
},
|
| 9930 |
+
{
|
| 9931 |
+
"epoch": 0.917799352750809,
|
| 9932 |
+
"grad_norm": 2.9125007131369784e-05,
|
| 9933 |
+
"learning_rate": 1.8938719051931674e-06,
|
| 9934 |
+
"loss": 0.0,
|
| 9935 |
+
"step": 1418
|
| 9936 |
+
},
|
| 9937 |
+
{
|
| 9938 |
+
"epoch": 0.9184466019417475,
|
| 9939 |
+
"grad_norm": 3.1149378628470004e-05,
|
| 9940 |
+
"learning_rate": 1.8643505901084268e-06,
|
| 9941 |
+
"loss": 0.0,
|
| 9942 |
+
"step": 1419
|
| 9943 |
+
},
|
| 9944 |
+
{
|
| 9945 |
+
"epoch": 0.919093851132686,
|
| 9946 |
+
"grad_norm": 3.2131334592122585e-05,
|
| 9947 |
+
"learning_rate": 1.8350568008579705e-06,
|
| 9948 |
+
"loss": 0.0,
|
| 9949 |
+
"step": 1420
|
| 9950 |
+
},
|
| 9951 |
+
{
|
| 9952 |
+
"epoch": 0.9197411003236245,
|
| 9953 |
+
"grad_norm": 3.188562550349161e-05,
|
| 9954 |
+
"learning_rate": 1.8059906759066159e-06,
|
| 9955 |
+
"loss": 0.0,
|
| 9956 |
+
"step": 1421
|
| 9957 |
+
},
|
| 9958 |
+
{
|
| 9959 |
+
"epoch": 0.920388349514563,
|
| 9960 |
+
"grad_norm": 3.0224360671127215e-05,
|
| 9961 |
+
"learning_rate": 1.777152352643069e-06,
|
| 9962 |
+
"loss": 0.0,
|
| 9963 |
+
"step": 1422
|
| 9964 |
+
},
|
| 9965 |
+
{
|
| 9966 |
+
"epoch": 0.9210355987055017,
|
| 9967 |
+
"grad_norm": 3.379662666702643e-05,
|
| 9968 |
+
"learning_rate": 1.7485419673792524e-06,
|
| 9969 |
+
"loss": 0.0,
|
| 9970 |
+
"step": 1423
|
| 9971 |
+
},
|
| 9972 |
+
{
|
| 9973 |
+
"epoch": 0.9216828478964402,
|
| 9974 |
+
"grad_norm": 3.2119089155457914e-05,
|
| 9975 |
+
"learning_rate": 1.7201596553497013e-06,
|
| 9976 |
+
"loss": 0.0,
|
| 9977 |
+
"step": 1424
|
| 9978 |
+
},
|
| 9979 |
+
{
|
| 9980 |
+
"epoch": 0.9223300970873787,
|
| 9981 |
+
"grad_norm": 3.09307397401426e-05,
|
| 9982 |
+
"learning_rate": 1.692005550710901e-06,
|
| 9983 |
+
"loss": 0.0,
|
| 9984 |
+
"step": 1425
|
| 9985 |
+
},
|
| 9986 |
+
{
|
| 9987 |
+
"epoch": 0.9229773462783172,
|
| 9988 |
+
"grad_norm": 3.093123814323917e-05,
|
| 9989 |
+
"learning_rate": 1.6640797865406288e-06,
|
| 9990 |
+
"loss": 0.0,
|
| 9991 |
+
"step": 1426
|
| 9992 |
+
},
|
| 9993 |
+
{
|
| 9994 |
+
"epoch": 0.9236245954692557,
|
| 9995 |
+
"grad_norm": 3.285687125753611e-05,
|
| 9996 |
+
"learning_rate": 1.6363824948373852e-06,
|
| 9997 |
+
"loss": 0.0,
|
| 9998 |
+
"step": 1427
|
| 9999 |
+
},
|
| 10000 |
+
{
|
| 10001 |
+
"epoch": 0.9242718446601942,
|
| 10002 |
+
"grad_norm": 2.959301127702929e-05,
|
| 10003 |
+
"learning_rate": 1.6089138065197185e-06,
|
| 10004 |
+
"loss": 0.0,
|
| 10005 |
+
"step": 1428
|
| 10006 |
+
},
|
| 10007 |
+
{
|
| 10008 |
+
"epoch": 0.9249190938511327,
|
| 10009 |
+
"grad_norm": 3.1159772333921865e-05,
|
| 10010 |
+
"learning_rate": 1.5816738514256135e-06,
|
| 10011 |
+
"loss": 0.0,
|
| 10012 |
+
"step": 1429
|
| 10013 |
+
},
|
| 10014 |
+
{
|
| 10015 |
+
"epoch": 0.9255663430420712,
|
| 10016 |
+
"grad_norm": 2.8291822673054412e-05,
|
| 10017 |
+
"learning_rate": 1.5546627583119088e-06,
|
| 10018 |
+
"loss": 0.0,
|
| 10019 |
+
"step": 1430
|
| 10020 |
+
},
|
| 10021 |
+
{
|
| 10022 |
+
"epoch": 0.9262135922330097,
|
| 10023 |
+
"grad_norm": 2.936238524853252e-05,
|
| 10024 |
+
"learning_rate": 1.5278806548536584e-06,
|
| 10025 |
+
"loss": 0.0,
|
| 10026 |
+
"step": 1431
|
| 10027 |
+
},
|
| 10028 |
+
{
|
| 10029 |
+
"epoch": 0.9268608414239482,
|
| 10030 |
+
"grad_norm": 3.067820944124833e-05,
|
| 10031 |
+
"learning_rate": 1.501327667643515e-06,
|
| 10032 |
+
"loss": 0.0,
|
| 10033 |
+
"step": 1432
|
| 10034 |
+
},
|
| 10035 |
+
{
|
| 10036 |
+
"epoch": 0.9275080906148867,
|
| 10037 |
+
"grad_norm": 2.8284572181291878e-05,
|
| 10038 |
+
"learning_rate": 1.4750039221911926e-06,
|
| 10039 |
+
"loss": 0.0,
|
| 10040 |
+
"step": 1433
|
| 10041 |
+
},
|
| 10042 |
+
{
|
| 10043 |
+
"epoch": 0.9281553398058252,
|
| 10044 |
+
"grad_norm": 2.9370548872975633e-05,
|
| 10045 |
+
"learning_rate": 1.4489095429227995e-06,
|
| 10046 |
+
"loss": 0.0,
|
| 10047 |
+
"step": 1434
|
| 10048 |
}
|
| 10049 |
],
|
| 10050 |
"logging_steps": 1,
|
|
|
|
| 10064 |
"attributes": {}
|
| 10065 |
}
|
| 10066 |
},
|
| 10067 |
+
"total_flos": 4.203003914074718e+18,
|
| 10068 |
"train_batch_size": 4,
|
| 10069 |
"trial_name": null,
|
| 10070 |
"trial_params": null
|