Training in progress, step 6214, 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 1140880624
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:294d261cda5ced74edd1d4c98c51cee9a61117ef6dd5f02b8134b6528effeb90
|
| 3 |
size 1140880624
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2281891834
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5ff5ee676f925a2181707ff1fd85a4b1de7eef1a04ca23f5ffa62cf1a93e7d67
|
| 3 |
size 2281891834
|
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 sha256:4853c64c7ec9ed6b1547fa5a6e7d3e434268dca3c243cd388dc441364d2a3694
|
| 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 sha256:333a9a1890a497931376258e58c4cbb393bb12fc883542acc2d5b199df5d4780
|
| 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,
|
|
@@ -41832,6 +41832,1679 @@
|
|
| 41832 |
"learning_rate": 8.6607118570576e-05,
|
| 41833 |
"loss": 0.1316,
|
| 41834 |
"step": 5975
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41835 |
}
|
| 41836 |
],
|
| 41837 |
"logging_steps": 1,
|
|
@@ -41851,7 +43524,7 @@
|
|
| 41851 |
"attributes": {}
|
| 41852 |
}
|
| 41853 |
},
|
| 41854 |
-
"total_flos": 3.
|
| 41855 |
"train_batch_size": 4,
|
| 41856 |
"trial_name": null,
|
| 41857 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.25125597662923166,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 6214,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 41832 |
"learning_rate": 8.6607118570576e-05,
|
| 41833 |
"loss": 0.1316,
|
| 41834 |
"step": 5975
|
| 41835 |
+
},
|
| 41836 |
+
{
|
| 41837 |
+
"epoch": 0.24163271907568207,
|
| 41838 |
+
"grad_norm": 2.853416919708252,
|
| 41839 |
+
"learning_rate": 8.660277435465594e-05,
|
| 41840 |
+
"loss": 0.1694,
|
| 41841 |
+
"step": 5976
|
| 41842 |
+
},
|
| 41843 |
+
{
|
| 41844 |
+
"epoch": 0.24167315293094907,
|
| 41845 |
+
"grad_norm": 5.313234329223633,
|
| 41846 |
+
"learning_rate": 8.659842954327992e-05,
|
| 41847 |
+
"loss": 0.2186,
|
| 41848 |
+
"step": 5977
|
| 41849 |
+
},
|
| 41850 |
+
{
|
| 41851 |
+
"epoch": 0.2417135867862161,
|
| 41852 |
+
"grad_norm": 2.550760507583618,
|
| 41853 |
+
"learning_rate": 8.659408413651861e-05,
|
| 41854 |
+
"loss": 0.094,
|
| 41855 |
+
"step": 5978
|
| 41856 |
+
},
|
| 41857 |
+
{
|
| 41858 |
+
"epoch": 0.2417540206414831,
|
| 41859 |
+
"grad_norm": 3.766831874847412,
|
| 41860 |
+
"learning_rate": 8.65897381344427e-05,
|
| 41861 |
+
"loss": 0.2396,
|
| 41862 |
+
"step": 5979
|
| 41863 |
+
},
|
| 41864 |
+
{
|
| 41865 |
+
"epoch": 0.24179445449675013,
|
| 41866 |
+
"grad_norm": 7.09255313873291,
|
| 41867 |
+
"learning_rate": 8.658539153712286e-05,
|
| 41868 |
+
"loss": 0.2572,
|
| 41869 |
+
"step": 5980
|
| 41870 |
+
},
|
| 41871 |
+
{
|
| 41872 |
+
"epoch": 0.24183488835201714,
|
| 41873 |
+
"grad_norm": 6.072479724884033,
|
| 41874 |
+
"learning_rate": 8.658104434462982e-05,
|
| 41875 |
+
"loss": 0.1831,
|
| 41876 |
+
"step": 5981
|
| 41877 |
+
},
|
| 41878 |
+
{
|
| 41879 |
+
"epoch": 0.24187532220728417,
|
| 41880 |
+
"grad_norm": 4.733118057250977,
|
| 41881 |
+
"learning_rate": 8.657669655703433e-05,
|
| 41882 |
+
"loss": 0.1759,
|
| 41883 |
+
"step": 5982
|
| 41884 |
+
},
|
| 41885 |
+
{
|
| 41886 |
+
"epoch": 0.24191575606255117,
|
| 41887 |
+
"grad_norm": 5.013708114624023,
|
| 41888 |
+
"learning_rate": 8.657234817440708e-05,
|
| 41889 |
+
"loss": 0.2601,
|
| 41890 |
+
"step": 5983
|
| 41891 |
+
},
|
| 41892 |
+
{
|
| 41893 |
+
"epoch": 0.2419561899178182,
|
| 41894 |
+
"grad_norm": 4.62626314163208,
|
| 41895 |
+
"learning_rate": 8.656799919681883e-05,
|
| 41896 |
+
"loss": 0.2134,
|
| 41897 |
+
"step": 5984
|
| 41898 |
+
},
|
| 41899 |
+
{
|
| 41900 |
+
"epoch": 0.2419966237730852,
|
| 41901 |
+
"grad_norm": 3.8375024795532227,
|
| 41902 |
+
"learning_rate": 8.656364962434033e-05,
|
| 41903 |
+
"loss": 0.301,
|
| 41904 |
+
"step": 5985
|
| 41905 |
+
},
|
| 41906 |
+
{
|
| 41907 |
+
"epoch": 0.24203705762835223,
|
| 41908 |
+
"grad_norm": 2.5020620822906494,
|
| 41909 |
+
"learning_rate": 8.655929945704233e-05,
|
| 41910 |
+
"loss": 0.1129,
|
| 41911 |
+
"step": 5986
|
| 41912 |
+
},
|
| 41913 |
+
{
|
| 41914 |
+
"epoch": 0.24207749148361923,
|
| 41915 |
+
"grad_norm": 3.215634822845459,
|
| 41916 |
+
"learning_rate": 8.655494869499558e-05,
|
| 41917 |
+
"loss": 0.2104,
|
| 41918 |
+
"step": 5987
|
| 41919 |
+
},
|
| 41920 |
+
{
|
| 41921 |
+
"epoch": 0.24211792533888624,
|
| 41922 |
+
"grad_norm": 3.1493115425109863,
|
| 41923 |
+
"learning_rate": 8.65505973382709e-05,
|
| 41924 |
+
"loss": 0.1872,
|
| 41925 |
+
"step": 5988
|
| 41926 |
+
},
|
| 41927 |
+
{
|
| 41928 |
+
"epoch": 0.24215835919415327,
|
| 41929 |
+
"grad_norm": 3.9308154582977295,
|
| 41930 |
+
"learning_rate": 8.654624538693907e-05,
|
| 41931 |
+
"loss": 0.2378,
|
| 41932 |
+
"step": 5989
|
| 41933 |
+
},
|
| 41934 |
+
{
|
| 41935 |
+
"epoch": 0.24219879304942027,
|
| 41936 |
+
"grad_norm": 2.025744915008545,
|
| 41937 |
+
"learning_rate": 8.654189284107088e-05,
|
| 41938 |
+
"loss": 0.1456,
|
| 41939 |
+
"step": 5990
|
| 41940 |
+
},
|
| 41941 |
+
{
|
| 41942 |
+
"epoch": 0.2422392269046873,
|
| 41943 |
+
"grad_norm": 3.391645908355713,
|
| 41944 |
+
"learning_rate": 8.65375397007371e-05,
|
| 41945 |
+
"loss": 0.1188,
|
| 41946 |
+
"step": 5991
|
| 41947 |
+
},
|
| 41948 |
+
{
|
| 41949 |
+
"epoch": 0.2422796607599543,
|
| 41950 |
+
"grad_norm": 7.289383411407471,
|
| 41951 |
+
"learning_rate": 8.653318596600862e-05,
|
| 41952 |
+
"loss": 0.208,
|
| 41953 |
+
"step": 5992
|
| 41954 |
+
},
|
| 41955 |
+
{
|
| 41956 |
+
"epoch": 0.24232009461522133,
|
| 41957 |
+
"grad_norm": 3.654409408569336,
|
| 41958 |
+
"learning_rate": 8.65288316369562e-05,
|
| 41959 |
+
"loss": 0.1813,
|
| 41960 |
+
"step": 5993
|
| 41961 |
+
},
|
| 41962 |
+
{
|
| 41963 |
+
"epoch": 0.24236052847048833,
|
| 41964 |
+
"grad_norm": 4.555336952209473,
|
| 41965 |
+
"learning_rate": 8.652447671365072e-05,
|
| 41966 |
+
"loss": 0.1197,
|
| 41967 |
+
"step": 5994
|
| 41968 |
+
},
|
| 41969 |
+
{
|
| 41970 |
+
"epoch": 0.24240096232575536,
|
| 41971 |
+
"grad_norm": 4.378109931945801,
|
| 41972 |
+
"learning_rate": 8.6520121196163e-05,
|
| 41973 |
+
"loss": 0.1918,
|
| 41974 |
+
"step": 5995
|
| 41975 |
+
},
|
| 41976 |
+
{
|
| 41977 |
+
"epoch": 0.24244139618102237,
|
| 41978 |
+
"grad_norm": 1.2791835069656372,
|
| 41979 |
+
"learning_rate": 8.651576508456392e-05,
|
| 41980 |
+
"loss": 0.0739,
|
| 41981 |
+
"step": 5996
|
| 41982 |
+
},
|
| 41983 |
+
{
|
| 41984 |
+
"epoch": 0.2424818300362894,
|
| 41985 |
+
"grad_norm": 3.9173526763916016,
|
| 41986 |
+
"learning_rate": 8.651140837892432e-05,
|
| 41987 |
+
"loss": 0.1931,
|
| 41988 |
+
"step": 5997
|
| 41989 |
+
},
|
| 41990 |
+
{
|
| 41991 |
+
"epoch": 0.2425222638915564,
|
| 41992 |
+
"grad_norm": 4.513161659240723,
|
| 41993 |
+
"learning_rate": 8.65070510793151e-05,
|
| 41994 |
+
"loss": 0.0788,
|
| 41995 |
+
"step": 5998
|
| 41996 |
+
},
|
| 41997 |
+
{
|
| 41998 |
+
"epoch": 0.2425626977468234,
|
| 41999 |
+
"grad_norm": 7.073320388793945,
|
| 42000 |
+
"learning_rate": 8.65026931858071e-05,
|
| 42001 |
+
"loss": 0.2172,
|
| 42002 |
+
"step": 5999
|
| 42003 |
+
},
|
| 42004 |
+
{
|
| 42005 |
+
"epoch": 0.24260313160209043,
|
| 42006 |
+
"grad_norm": 3.8800010681152344,
|
| 42007 |
+
"learning_rate": 8.649833469847129e-05,
|
| 42008 |
+
"loss": 0.1563,
|
| 42009 |
+
"step": 6000
|
| 42010 |
+
},
|
| 42011 |
+
{
|
| 42012 |
+
"epoch": 0.24264356545735744,
|
| 42013 |
+
"grad_norm": 2.8107495307922363,
|
| 42014 |
+
"learning_rate": 8.649397561737851e-05,
|
| 42015 |
+
"loss": 0.2392,
|
| 42016 |
+
"step": 6001
|
| 42017 |
+
},
|
| 42018 |
+
{
|
| 42019 |
+
"epoch": 0.24268399931262447,
|
| 42020 |
+
"grad_norm": 8.04423713684082,
|
| 42021 |
+
"learning_rate": 8.648961594259968e-05,
|
| 42022 |
+
"loss": 0.2228,
|
| 42023 |
+
"step": 6002
|
| 42024 |
+
},
|
| 42025 |
+
{
|
| 42026 |
+
"epoch": 0.24272443316789147,
|
| 42027 |
+
"grad_norm": 6.598309516906738,
|
| 42028 |
+
"learning_rate": 8.648525567420576e-05,
|
| 42029 |
+
"loss": 0.1908,
|
| 42030 |
+
"step": 6003
|
| 42031 |
+
},
|
| 42032 |
+
{
|
| 42033 |
+
"epoch": 0.2427648670231585,
|
| 42034 |
+
"grad_norm": 3.7928972244262695,
|
| 42035 |
+
"learning_rate": 8.648089481226764e-05,
|
| 42036 |
+
"loss": 0.2214,
|
| 42037 |
+
"step": 6004
|
| 42038 |
+
},
|
| 42039 |
+
{
|
| 42040 |
+
"epoch": 0.2428053008784255,
|
| 42041 |
+
"grad_norm": 1.895093321800232,
|
| 42042 |
+
"learning_rate": 8.647653335685631e-05,
|
| 42043 |
+
"loss": 0.1524,
|
| 42044 |
+
"step": 6005
|
| 42045 |
+
},
|
| 42046 |
+
{
|
| 42047 |
+
"epoch": 0.24284573473369253,
|
| 42048 |
+
"grad_norm": 4.698232650756836,
|
| 42049 |
+
"learning_rate": 8.647217130804268e-05,
|
| 42050 |
+
"loss": 0.1546,
|
| 42051 |
+
"step": 6006
|
| 42052 |
+
},
|
| 42053 |
+
{
|
| 42054 |
+
"epoch": 0.24288616858895953,
|
| 42055 |
+
"grad_norm": 3.976858615875244,
|
| 42056 |
+
"learning_rate": 8.646780866589772e-05,
|
| 42057 |
+
"loss": 0.2591,
|
| 42058 |
+
"step": 6007
|
| 42059 |
+
},
|
| 42060 |
+
{
|
| 42061 |
+
"epoch": 0.24292660244422656,
|
| 42062 |
+
"grad_norm": 5.355378150939941,
|
| 42063 |
+
"learning_rate": 8.646344543049243e-05,
|
| 42064 |
+
"loss": 0.2391,
|
| 42065 |
+
"step": 6008
|
| 42066 |
+
},
|
| 42067 |
+
{
|
| 42068 |
+
"epoch": 0.24296703629949357,
|
| 42069 |
+
"grad_norm": 3.624940872192383,
|
| 42070 |
+
"learning_rate": 8.645908160189774e-05,
|
| 42071 |
+
"loss": 0.1035,
|
| 42072 |
+
"step": 6009
|
| 42073 |
+
},
|
| 42074 |
+
{
|
| 42075 |
+
"epoch": 0.24300747015476057,
|
| 42076 |
+
"grad_norm": 7.696066379547119,
|
| 42077 |
+
"learning_rate": 8.64547171801847e-05,
|
| 42078 |
+
"loss": 0.1876,
|
| 42079 |
+
"step": 6010
|
| 42080 |
+
},
|
| 42081 |
+
{
|
| 42082 |
+
"epoch": 0.2430479040100276,
|
| 42083 |
+
"grad_norm": 4.4118452072143555,
|
| 42084 |
+
"learning_rate": 8.645035216542428e-05,
|
| 42085 |
+
"loss": 0.1505,
|
| 42086 |
+
"step": 6011
|
| 42087 |
+
},
|
| 42088 |
+
{
|
| 42089 |
+
"epoch": 0.2430883378652946,
|
| 42090 |
+
"grad_norm": 9.357704162597656,
|
| 42091 |
+
"learning_rate": 8.64459865576875e-05,
|
| 42092 |
+
"loss": 0.2712,
|
| 42093 |
+
"step": 6012
|
| 42094 |
+
},
|
| 42095 |
+
{
|
| 42096 |
+
"epoch": 0.24312877172056163,
|
| 42097 |
+
"grad_norm": 2.713459014892578,
|
| 42098 |
+
"learning_rate": 8.644162035704535e-05,
|
| 42099 |
+
"loss": 0.1985,
|
| 42100 |
+
"step": 6013
|
| 42101 |
+
},
|
| 42102 |
+
{
|
| 42103 |
+
"epoch": 0.24316920557582863,
|
| 42104 |
+
"grad_norm": 5.679785251617432,
|
| 42105 |
+
"learning_rate": 8.643725356356889e-05,
|
| 42106 |
+
"loss": 0.1645,
|
| 42107 |
+
"step": 6014
|
| 42108 |
+
},
|
| 42109 |
+
{
|
| 42110 |
+
"epoch": 0.24320963943109566,
|
| 42111 |
+
"grad_norm": 7.06746244430542,
|
| 42112 |
+
"learning_rate": 8.643288617732917e-05,
|
| 42113 |
+
"loss": 0.2948,
|
| 42114 |
+
"step": 6015
|
| 42115 |
+
},
|
| 42116 |
+
{
|
| 42117 |
+
"epoch": 0.24325007328636267,
|
| 42118 |
+
"grad_norm": 3.614020347595215,
|
| 42119 |
+
"learning_rate": 8.642851819839719e-05,
|
| 42120 |
+
"loss": 0.2815,
|
| 42121 |
+
"step": 6016
|
| 42122 |
+
},
|
| 42123 |
+
{
|
| 42124 |
+
"epoch": 0.2432905071416297,
|
| 42125 |
+
"grad_norm": 8.398319244384766,
|
| 42126 |
+
"learning_rate": 8.642414962684406e-05,
|
| 42127 |
+
"loss": 0.1573,
|
| 42128 |
+
"step": 6017
|
| 42129 |
+
},
|
| 42130 |
+
{
|
| 42131 |
+
"epoch": 0.2433309409968967,
|
| 42132 |
+
"grad_norm": 5.846600532531738,
|
| 42133 |
+
"learning_rate": 8.641978046274082e-05,
|
| 42134 |
+
"loss": 0.1309,
|
| 42135 |
+
"step": 6018
|
| 42136 |
+
},
|
| 42137 |
+
{
|
| 42138 |
+
"epoch": 0.24337137485216373,
|
| 42139 |
+
"grad_norm": 2.964691400527954,
|
| 42140 |
+
"learning_rate": 8.641541070615855e-05,
|
| 42141 |
+
"loss": 0.0609,
|
| 42142 |
+
"step": 6019
|
| 42143 |
+
},
|
| 42144 |
+
{
|
| 42145 |
+
"epoch": 0.24341180870743073,
|
| 42146 |
+
"grad_norm": 4.7332444190979,
|
| 42147 |
+
"learning_rate": 8.641104035716836e-05,
|
| 42148 |
+
"loss": 0.2204,
|
| 42149 |
+
"step": 6020
|
| 42150 |
+
},
|
| 42151 |
+
{
|
| 42152 |
+
"epoch": 0.24345224256269773,
|
| 42153 |
+
"grad_norm": 3.9071009159088135,
|
| 42154 |
+
"learning_rate": 8.640666941584132e-05,
|
| 42155 |
+
"loss": 0.1772,
|
| 42156 |
+
"step": 6021
|
| 42157 |
+
},
|
| 42158 |
+
{
|
| 42159 |
+
"epoch": 0.24349267641796477,
|
| 42160 |
+
"grad_norm": 4.52919864654541,
|
| 42161 |
+
"learning_rate": 8.640229788224853e-05,
|
| 42162 |
+
"loss": 0.1845,
|
| 42163 |
+
"step": 6022
|
| 42164 |
+
},
|
| 42165 |
+
{
|
| 42166 |
+
"epoch": 0.24353311027323177,
|
| 42167 |
+
"grad_norm": 6.53327751159668,
|
| 42168 |
+
"learning_rate": 8.639792575646115e-05,
|
| 42169 |
+
"loss": 0.1407,
|
| 42170 |
+
"step": 6023
|
| 42171 |
+
},
|
| 42172 |
+
{
|
| 42173 |
+
"epoch": 0.2435735441284988,
|
| 42174 |
+
"grad_norm": 4.435870170593262,
|
| 42175 |
+
"learning_rate": 8.639355303855028e-05,
|
| 42176 |
+
"loss": 0.0937,
|
| 42177 |
+
"step": 6024
|
| 42178 |
+
},
|
| 42179 |
+
{
|
| 42180 |
+
"epoch": 0.2436139779837658,
|
| 42181 |
+
"grad_norm": 6.9111480712890625,
|
| 42182 |
+
"learning_rate": 8.638917972858704e-05,
|
| 42183 |
+
"loss": 0.2429,
|
| 42184 |
+
"step": 6025
|
| 42185 |
+
},
|
| 42186 |
+
{
|
| 42187 |
+
"epoch": 0.24365441183903283,
|
| 42188 |
+
"grad_norm": 4.262193202972412,
|
| 42189 |
+
"learning_rate": 8.638480582664259e-05,
|
| 42190 |
+
"loss": 0.22,
|
| 42191 |
+
"step": 6026
|
| 42192 |
+
},
|
| 42193 |
+
{
|
| 42194 |
+
"epoch": 0.24369484569429983,
|
| 42195 |
+
"grad_norm": 3.620594024658203,
|
| 42196 |
+
"learning_rate": 8.638043133278809e-05,
|
| 42197 |
+
"loss": 0.1129,
|
| 42198 |
+
"step": 6027
|
| 42199 |
+
},
|
| 42200 |
+
{
|
| 42201 |
+
"epoch": 0.24373527954956686,
|
| 42202 |
+
"grad_norm": 6.571465969085693,
|
| 42203 |
+
"learning_rate": 8.637605624709468e-05,
|
| 42204 |
+
"loss": 0.1726,
|
| 42205 |
+
"step": 6028
|
| 42206 |
+
},
|
| 42207 |
+
{
|
| 42208 |
+
"epoch": 0.24377571340483387,
|
| 42209 |
+
"grad_norm": 2.8471567630767822,
|
| 42210 |
+
"learning_rate": 8.637168056963357e-05,
|
| 42211 |
+
"loss": 0.0976,
|
| 42212 |
+
"step": 6029
|
| 42213 |
+
},
|
| 42214 |
+
{
|
| 42215 |
+
"epoch": 0.2438161472601009,
|
| 42216 |
+
"grad_norm": 6.479422092437744,
|
| 42217 |
+
"learning_rate": 8.636730430047593e-05,
|
| 42218 |
+
"loss": 0.2516,
|
| 42219 |
+
"step": 6030
|
| 42220 |
+
},
|
| 42221 |
+
{
|
| 42222 |
+
"epoch": 0.2438565811153679,
|
| 42223 |
+
"grad_norm": 2.0430047512054443,
|
| 42224 |
+
"learning_rate": 8.636292743969294e-05,
|
| 42225 |
+
"loss": 0.1564,
|
| 42226 |
+
"step": 6031
|
| 42227 |
+
},
|
| 42228 |
+
{
|
| 42229 |
+
"epoch": 0.2438970149706349,
|
| 42230 |
+
"grad_norm": 3.429826498031616,
|
| 42231 |
+
"learning_rate": 8.635854998735582e-05,
|
| 42232 |
+
"loss": 0.1513,
|
| 42233 |
+
"step": 6032
|
| 42234 |
+
},
|
| 42235 |
+
{
|
| 42236 |
+
"epoch": 0.24393744882590193,
|
| 42237 |
+
"grad_norm": 2.052880048751831,
|
| 42238 |
+
"learning_rate": 8.635417194353578e-05,
|
| 42239 |
+
"loss": 0.0597,
|
| 42240 |
+
"step": 6033
|
| 42241 |
+
},
|
| 42242 |
+
{
|
| 42243 |
+
"epoch": 0.24397788268116893,
|
| 42244 |
+
"grad_norm": 8.362695693969727,
|
| 42245 |
+
"learning_rate": 8.634979330830404e-05,
|
| 42246 |
+
"loss": 0.2508,
|
| 42247 |
+
"step": 6034
|
| 42248 |
+
},
|
| 42249 |
+
{
|
| 42250 |
+
"epoch": 0.24401831653643596,
|
| 42251 |
+
"grad_norm": 2.54459285736084,
|
| 42252 |
+
"learning_rate": 8.634541408173182e-05,
|
| 42253 |
+
"loss": 0.0596,
|
| 42254 |
+
"step": 6035
|
| 42255 |
+
},
|
| 42256 |
+
{
|
| 42257 |
+
"epoch": 0.24405875039170297,
|
| 42258 |
+
"grad_norm": 4.136692523956299,
|
| 42259 |
+
"learning_rate": 8.634103426389037e-05,
|
| 42260 |
+
"loss": 0.1318,
|
| 42261 |
+
"step": 6036
|
| 42262 |
+
},
|
| 42263 |
+
{
|
| 42264 |
+
"epoch": 0.24409918424697,
|
| 42265 |
+
"grad_norm": 10.016950607299805,
|
| 42266 |
+
"learning_rate": 8.633665385485094e-05,
|
| 42267 |
+
"loss": 0.2157,
|
| 42268 |
+
"step": 6037
|
| 42269 |
+
},
|
| 42270 |
+
{
|
| 42271 |
+
"epoch": 0.244139618102237,
|
| 42272 |
+
"grad_norm": 4.006833076477051,
|
| 42273 |
+
"learning_rate": 8.633227285468481e-05,
|
| 42274 |
+
"loss": 0.232,
|
| 42275 |
+
"step": 6038
|
| 42276 |
+
},
|
| 42277 |
+
{
|
| 42278 |
+
"epoch": 0.24418005195750403,
|
| 42279 |
+
"grad_norm": 3.8136184215545654,
|
| 42280 |
+
"learning_rate": 8.632789126346322e-05,
|
| 42281 |
+
"loss": 0.0961,
|
| 42282 |
+
"step": 6039
|
| 42283 |
+
},
|
| 42284 |
+
{
|
| 42285 |
+
"epoch": 0.24422048581277103,
|
| 42286 |
+
"grad_norm": 7.684732913970947,
|
| 42287 |
+
"learning_rate": 8.632350908125748e-05,
|
| 42288 |
+
"loss": 0.1847,
|
| 42289 |
+
"step": 6040
|
| 42290 |
+
},
|
| 42291 |
+
{
|
| 42292 |
+
"epoch": 0.24426091966803803,
|
| 42293 |
+
"grad_norm": 3.978747606277466,
|
| 42294 |
+
"learning_rate": 8.631912630813885e-05,
|
| 42295 |
+
"loss": 0.3372,
|
| 42296 |
+
"step": 6041
|
| 42297 |
+
},
|
| 42298 |
+
{
|
| 42299 |
+
"epoch": 0.24430135352330506,
|
| 42300 |
+
"grad_norm": 5.78717565536499,
|
| 42301 |
+
"learning_rate": 8.631474294417864e-05,
|
| 42302 |
+
"loss": 0.1573,
|
| 42303 |
+
"step": 6042
|
| 42304 |
+
},
|
| 42305 |
+
{
|
| 42306 |
+
"epoch": 0.24434178737857207,
|
| 42307 |
+
"grad_norm": 3.336500406265259,
|
| 42308 |
+
"learning_rate": 8.631035898944817e-05,
|
| 42309 |
+
"loss": 0.1248,
|
| 42310 |
+
"step": 6043
|
| 42311 |
+
},
|
| 42312 |
+
{
|
| 42313 |
+
"epoch": 0.2443822212338391,
|
| 42314 |
+
"grad_norm": 3.878305673599243,
|
| 42315 |
+
"learning_rate": 8.630597444401873e-05,
|
| 42316 |
+
"loss": 0.1764,
|
| 42317 |
+
"step": 6044
|
| 42318 |
+
},
|
| 42319 |
+
{
|
| 42320 |
+
"epoch": 0.2444226550891061,
|
| 42321 |
+
"grad_norm": 3.5070252418518066,
|
| 42322 |
+
"learning_rate": 8.63015893079617e-05,
|
| 42323 |
+
"loss": 0.1417,
|
| 42324 |
+
"step": 6045
|
| 42325 |
+
},
|
| 42326 |
+
{
|
| 42327 |
+
"epoch": 0.24446308894437313,
|
| 42328 |
+
"grad_norm": 9.929906845092773,
|
| 42329 |
+
"learning_rate": 8.629720358134836e-05,
|
| 42330 |
+
"loss": 0.2483,
|
| 42331 |
+
"step": 6046
|
| 42332 |
+
},
|
| 42333 |
+
{
|
| 42334 |
+
"epoch": 0.24450352279964013,
|
| 42335 |
+
"grad_norm": 2.988154888153076,
|
| 42336 |
+
"learning_rate": 8.62928172642501e-05,
|
| 42337 |
+
"loss": 0.2212,
|
| 42338 |
+
"step": 6047
|
| 42339 |
+
},
|
| 42340 |
+
{
|
| 42341 |
+
"epoch": 0.24454395665490716,
|
| 42342 |
+
"grad_norm": 2.689389944076538,
|
| 42343 |
+
"learning_rate": 8.628843035673824e-05,
|
| 42344 |
+
"loss": 0.0792,
|
| 42345 |
+
"step": 6048
|
| 42346 |
+
},
|
| 42347 |
+
{
|
| 42348 |
+
"epoch": 0.24458439051017417,
|
| 42349 |
+
"grad_norm": 12.93316650390625,
|
| 42350 |
+
"learning_rate": 8.628404285888418e-05,
|
| 42351 |
+
"loss": 0.2807,
|
| 42352 |
+
"step": 6049
|
| 42353 |
+
},
|
| 42354 |
+
{
|
| 42355 |
+
"epoch": 0.2446248243654412,
|
| 42356 |
+
"grad_norm": 9.077486038208008,
|
| 42357 |
+
"learning_rate": 8.627965477075928e-05,
|
| 42358 |
+
"loss": 0.3347,
|
| 42359 |
+
"step": 6050
|
| 42360 |
+
},
|
| 42361 |
+
{
|
| 42362 |
+
"epoch": 0.2446652582207082,
|
| 42363 |
+
"grad_norm": 15.951316833496094,
|
| 42364 |
+
"learning_rate": 8.627526609243492e-05,
|
| 42365 |
+
"loss": 0.3689,
|
| 42366 |
+
"step": 6051
|
| 42367 |
+
},
|
| 42368 |
+
{
|
| 42369 |
+
"epoch": 0.2447056920759752,
|
| 42370 |
+
"grad_norm": 2.9297943115234375,
|
| 42371 |
+
"learning_rate": 8.627087682398252e-05,
|
| 42372 |
+
"loss": 0.1827,
|
| 42373 |
+
"step": 6052
|
| 42374 |
+
},
|
| 42375 |
+
{
|
| 42376 |
+
"epoch": 0.24474612593124223,
|
| 42377 |
+
"grad_norm": 3.517852544784546,
|
| 42378 |
+
"learning_rate": 8.626648696547346e-05,
|
| 42379 |
+
"loss": 0.087,
|
| 42380 |
+
"step": 6053
|
| 42381 |
+
},
|
| 42382 |
+
{
|
| 42383 |
+
"epoch": 0.24478655978650923,
|
| 42384 |
+
"grad_norm": 11.33461856842041,
|
| 42385 |
+
"learning_rate": 8.626209651697916e-05,
|
| 42386 |
+
"loss": 0.337,
|
| 42387 |
+
"step": 6054
|
| 42388 |
+
},
|
| 42389 |
+
{
|
| 42390 |
+
"epoch": 0.24482699364177626,
|
| 42391 |
+
"grad_norm": 3.5571320056915283,
|
| 42392 |
+
"learning_rate": 8.625770547857104e-05,
|
| 42393 |
+
"loss": 0.1222,
|
| 42394 |
+
"step": 6055
|
| 42395 |
+
},
|
| 42396 |
+
{
|
| 42397 |
+
"epoch": 0.24486742749704327,
|
| 42398 |
+
"grad_norm": 3.389450788497925,
|
| 42399 |
+
"learning_rate": 8.625331385032056e-05,
|
| 42400 |
+
"loss": 0.1753,
|
| 42401 |
+
"step": 6056
|
| 42402 |
+
},
|
| 42403 |
+
{
|
| 42404 |
+
"epoch": 0.2449078613523103,
|
| 42405 |
+
"grad_norm": 4.702533721923828,
|
| 42406 |
+
"learning_rate": 8.624892163229912e-05,
|
| 42407 |
+
"loss": 0.1709,
|
| 42408 |
+
"step": 6057
|
| 42409 |
+
},
|
| 42410 |
+
{
|
| 42411 |
+
"epoch": 0.2449482952075773,
|
| 42412 |
+
"grad_norm": 6.528069972991943,
|
| 42413 |
+
"learning_rate": 8.624452882457821e-05,
|
| 42414 |
+
"loss": 0.1442,
|
| 42415 |
+
"step": 6058
|
| 42416 |
+
},
|
| 42417 |
+
{
|
| 42418 |
+
"epoch": 0.24498872906284433,
|
| 42419 |
+
"grad_norm": 2.3681833744049072,
|
| 42420 |
+
"learning_rate": 8.624013542722927e-05,
|
| 42421 |
+
"loss": 0.1071,
|
| 42422 |
+
"step": 6059
|
| 42423 |
+
},
|
| 42424 |
+
{
|
| 42425 |
+
"epoch": 0.24502916291811133,
|
| 42426 |
+
"grad_norm": 6.04899263381958,
|
| 42427 |
+
"learning_rate": 8.623574144032379e-05,
|
| 42428 |
+
"loss": 0.2262,
|
| 42429 |
+
"step": 6060
|
| 42430 |
+
},
|
| 42431 |
+
{
|
| 42432 |
+
"epoch": 0.24506959677337836,
|
| 42433 |
+
"grad_norm": 3.0587050914764404,
|
| 42434 |
+
"learning_rate": 8.623134686393324e-05,
|
| 42435 |
+
"loss": 0.1647,
|
| 42436 |
+
"step": 6061
|
| 42437 |
+
},
|
| 42438 |
+
{
|
| 42439 |
+
"epoch": 0.24511003062864536,
|
| 42440 |
+
"grad_norm": 3.833815813064575,
|
| 42441 |
+
"learning_rate": 8.622695169812911e-05,
|
| 42442 |
+
"loss": 0.1766,
|
| 42443 |
+
"step": 6062
|
| 42444 |
+
},
|
| 42445 |
+
{
|
| 42446 |
+
"epoch": 0.24515046448391237,
|
| 42447 |
+
"grad_norm": 5.266330718994141,
|
| 42448 |
+
"learning_rate": 8.622255594298291e-05,
|
| 42449 |
+
"loss": 0.2156,
|
| 42450 |
+
"step": 6063
|
| 42451 |
+
},
|
| 42452 |
+
{
|
| 42453 |
+
"epoch": 0.2451908983391794,
|
| 42454 |
+
"grad_norm": 3.1624538898468018,
|
| 42455 |
+
"learning_rate": 8.621815959856614e-05,
|
| 42456 |
+
"loss": 0.1147,
|
| 42457 |
+
"step": 6064
|
| 42458 |
+
},
|
| 42459 |
+
{
|
| 42460 |
+
"epoch": 0.2452313321944464,
|
| 42461 |
+
"grad_norm": 7.124099254608154,
|
| 42462 |
+
"learning_rate": 8.621376266495031e-05,
|
| 42463 |
+
"loss": 0.274,
|
| 42464 |
+
"step": 6065
|
| 42465 |
+
},
|
| 42466 |
+
{
|
| 42467 |
+
"epoch": 0.24527176604971343,
|
| 42468 |
+
"grad_norm": 2.109743118286133,
|
| 42469 |
+
"learning_rate": 8.620936514220699e-05,
|
| 42470 |
+
"loss": 0.0597,
|
| 42471 |
+
"step": 6066
|
| 42472 |
+
},
|
| 42473 |
+
{
|
| 42474 |
+
"epoch": 0.24531219990498043,
|
| 42475 |
+
"grad_norm": 5.317684650421143,
|
| 42476 |
+
"learning_rate": 8.620496703040769e-05,
|
| 42477 |
+
"loss": 0.2001,
|
| 42478 |
+
"step": 6067
|
| 42479 |
+
},
|
| 42480 |
+
{
|
| 42481 |
+
"epoch": 0.24535263376024746,
|
| 42482 |
+
"grad_norm": 4.574336528778076,
|
| 42483 |
+
"learning_rate": 8.620056832962394e-05,
|
| 42484 |
+
"loss": 0.1516,
|
| 42485 |
+
"step": 6068
|
| 42486 |
+
},
|
| 42487 |
+
{
|
| 42488 |
+
"epoch": 0.24539306761551447,
|
| 42489 |
+
"grad_norm": 4.148901462554932,
|
| 42490 |
+
"learning_rate": 8.619616903992733e-05,
|
| 42491 |
+
"loss": 0.0779,
|
| 42492 |
+
"step": 6069
|
| 42493 |
+
},
|
| 42494 |
+
{
|
| 42495 |
+
"epoch": 0.2454335014707815,
|
| 42496 |
+
"grad_norm": 5.344276428222656,
|
| 42497 |
+
"learning_rate": 8.619176916138944e-05,
|
| 42498 |
+
"loss": 0.1638,
|
| 42499 |
+
"step": 6070
|
| 42500 |
+
},
|
| 42501 |
+
{
|
| 42502 |
+
"epoch": 0.2454739353260485,
|
| 42503 |
+
"grad_norm": 4.519772529602051,
|
| 42504 |
+
"learning_rate": 8.61873686940818e-05,
|
| 42505 |
+
"loss": 0.2012,
|
| 42506 |
+
"step": 6071
|
| 42507 |
+
},
|
| 42508 |
+
{
|
| 42509 |
+
"epoch": 0.24551436918131553,
|
| 42510 |
+
"grad_norm": 4.1637115478515625,
|
| 42511 |
+
"learning_rate": 8.618296763807603e-05,
|
| 42512 |
+
"loss": 0.1288,
|
| 42513 |
+
"step": 6072
|
| 42514 |
+
},
|
| 42515 |
+
{
|
| 42516 |
+
"epoch": 0.24555480303658253,
|
| 42517 |
+
"grad_norm": 8.29826545715332,
|
| 42518 |
+
"learning_rate": 8.617856599344372e-05,
|
| 42519 |
+
"loss": 0.2354,
|
| 42520 |
+
"step": 6073
|
| 42521 |
+
},
|
| 42522 |
+
{
|
| 42523 |
+
"epoch": 0.24559523689184953,
|
| 42524 |
+
"grad_norm": 6.626035213470459,
|
| 42525 |
+
"learning_rate": 8.617416376025649e-05,
|
| 42526 |
+
"loss": 0.1596,
|
| 42527 |
+
"step": 6074
|
| 42528 |
+
},
|
| 42529 |
+
{
|
| 42530 |
+
"epoch": 0.24563567074711656,
|
| 42531 |
+
"grad_norm": 5.0871477127075195,
|
| 42532 |
+
"learning_rate": 8.616976093858591e-05,
|
| 42533 |
+
"loss": 0.1753,
|
| 42534 |
+
"step": 6075
|
| 42535 |
+
},
|
| 42536 |
+
{
|
| 42537 |
+
"epoch": 0.24567610460238357,
|
| 42538 |
+
"grad_norm": 4.304733753204346,
|
| 42539 |
+
"learning_rate": 8.616535752850366e-05,
|
| 42540 |
+
"loss": 0.2885,
|
| 42541 |
+
"step": 6076
|
| 42542 |
+
},
|
| 42543 |
+
{
|
| 42544 |
+
"epoch": 0.2457165384576506,
|
| 42545 |
+
"grad_norm": 3.2352781295776367,
|
| 42546 |
+
"learning_rate": 8.616095353008135e-05,
|
| 42547 |
+
"loss": 0.1896,
|
| 42548 |
+
"step": 6077
|
| 42549 |
+
},
|
| 42550 |
+
{
|
| 42551 |
+
"epoch": 0.2457569723129176,
|
| 42552 |
+
"grad_norm": 2.7523484230041504,
|
| 42553 |
+
"learning_rate": 8.615654894339063e-05,
|
| 42554 |
+
"loss": 0.1781,
|
| 42555 |
+
"step": 6078
|
| 42556 |
+
},
|
| 42557 |
+
{
|
| 42558 |
+
"epoch": 0.24579740616818463,
|
| 42559 |
+
"grad_norm": 3.727778673171997,
|
| 42560 |
+
"learning_rate": 8.615214376850314e-05,
|
| 42561 |
+
"loss": 0.1103,
|
| 42562 |
+
"step": 6079
|
| 42563 |
+
},
|
| 42564 |
+
{
|
| 42565 |
+
"epoch": 0.24583784002345163,
|
| 42566 |
+
"grad_norm": 3.9669647216796875,
|
| 42567 |
+
"learning_rate": 8.614773800549055e-05,
|
| 42568 |
+
"loss": 0.0879,
|
| 42569 |
+
"step": 6080
|
| 42570 |
+
},
|
| 42571 |
+
{
|
| 42572 |
+
"epoch": 0.24587827387871866,
|
| 42573 |
+
"grad_norm": 6.558189392089844,
|
| 42574 |
+
"learning_rate": 8.614333165442454e-05,
|
| 42575 |
+
"loss": 0.1847,
|
| 42576 |
+
"step": 6081
|
| 42577 |
+
},
|
| 42578 |
+
{
|
| 42579 |
+
"epoch": 0.24591870773398566,
|
| 42580 |
+
"grad_norm": 6.306224346160889,
|
| 42581 |
+
"learning_rate": 8.61389247153768e-05,
|
| 42582 |
+
"loss": 0.1495,
|
| 42583 |
+
"step": 6082
|
| 42584 |
+
},
|
| 42585 |
+
{
|
| 42586 |
+
"epoch": 0.2459591415892527,
|
| 42587 |
+
"grad_norm": 3.4565958976745605,
|
| 42588 |
+
"learning_rate": 8.6134517188419e-05,
|
| 42589 |
+
"loss": 0.0916,
|
| 42590 |
+
"step": 6083
|
| 42591 |
+
},
|
| 42592 |
+
{
|
| 42593 |
+
"epoch": 0.2459995754445197,
|
| 42594 |
+
"grad_norm": 4.450873851776123,
|
| 42595 |
+
"learning_rate": 8.613010907362287e-05,
|
| 42596 |
+
"loss": 0.1449,
|
| 42597 |
+
"step": 6084
|
| 42598 |
+
},
|
| 42599 |
+
{
|
| 42600 |
+
"epoch": 0.2460400092997867,
|
| 42601 |
+
"grad_norm": 6.050655364990234,
|
| 42602 |
+
"learning_rate": 8.612570037106009e-05,
|
| 42603 |
+
"loss": 0.2411,
|
| 42604 |
+
"step": 6085
|
| 42605 |
+
},
|
| 42606 |
+
{
|
| 42607 |
+
"epoch": 0.24608044315505373,
|
| 42608 |
+
"grad_norm": 4.706817150115967,
|
| 42609 |
+
"learning_rate": 8.61212910808024e-05,
|
| 42610 |
+
"loss": 0.0666,
|
| 42611 |
+
"step": 6086
|
| 42612 |
+
},
|
| 42613 |
+
{
|
| 42614 |
+
"epoch": 0.24612087701032073,
|
| 42615 |
+
"grad_norm": 2.352297067642212,
|
| 42616 |
+
"learning_rate": 8.611688120292155e-05,
|
| 42617 |
+
"loss": 0.1034,
|
| 42618 |
+
"step": 6087
|
| 42619 |
+
},
|
| 42620 |
+
{
|
| 42621 |
+
"epoch": 0.24616131086558776,
|
| 42622 |
+
"grad_norm": 4.81610107421875,
|
| 42623 |
+
"learning_rate": 8.611247073748922e-05,
|
| 42624 |
+
"loss": 0.0981,
|
| 42625 |
+
"step": 6088
|
| 42626 |
+
},
|
| 42627 |
+
{
|
| 42628 |
+
"epoch": 0.24620174472085476,
|
| 42629 |
+
"grad_norm": 8.78715991973877,
|
| 42630 |
+
"learning_rate": 8.610805968457721e-05,
|
| 42631 |
+
"loss": 0.211,
|
| 42632 |
+
"step": 6089
|
| 42633 |
+
},
|
| 42634 |
+
{
|
| 42635 |
+
"epoch": 0.2462421785761218,
|
| 42636 |
+
"grad_norm": 6.008256912231445,
|
| 42637 |
+
"learning_rate": 8.610364804425728e-05,
|
| 42638 |
+
"loss": 0.0955,
|
| 42639 |
+
"step": 6090
|
| 42640 |
+
},
|
| 42641 |
+
{
|
| 42642 |
+
"epoch": 0.2462826124313888,
|
| 42643 |
+
"grad_norm": 4.646786212921143,
|
| 42644 |
+
"learning_rate": 8.609923581660118e-05,
|
| 42645 |
+
"loss": 0.2153,
|
| 42646 |
+
"step": 6091
|
| 42647 |
+
},
|
| 42648 |
+
{
|
| 42649 |
+
"epoch": 0.24632304628665583,
|
| 42650 |
+
"grad_norm": 1.5845496654510498,
|
| 42651 |
+
"learning_rate": 8.609482300168067e-05,
|
| 42652 |
+
"loss": 0.0434,
|
| 42653 |
+
"step": 6092
|
| 42654 |
+
},
|
| 42655 |
+
{
|
| 42656 |
+
"epoch": 0.24636348014192283,
|
| 42657 |
+
"grad_norm": 2.4028897285461426,
|
| 42658 |
+
"learning_rate": 8.609040959956759e-05,
|
| 42659 |
+
"loss": 0.0609,
|
| 42660 |
+
"step": 6093
|
| 42661 |
+
},
|
| 42662 |
+
{
|
| 42663 |
+
"epoch": 0.24640391399718986,
|
| 42664 |
+
"grad_norm": 36.975887298583984,
|
| 42665 |
+
"learning_rate": 8.608599561033369e-05,
|
| 42666 |
+
"loss": 0.1087,
|
| 42667 |
+
"step": 6094
|
| 42668 |
+
},
|
| 42669 |
+
{
|
| 42670 |
+
"epoch": 0.24644434785245686,
|
| 42671 |
+
"grad_norm": 7.599268913269043,
|
| 42672 |
+
"learning_rate": 8.60815810340508e-05,
|
| 42673 |
+
"loss": 0.2089,
|
| 42674 |
+
"step": 6095
|
| 42675 |
+
},
|
| 42676 |
+
{
|
| 42677 |
+
"epoch": 0.24648478170772387,
|
| 42678 |
+
"grad_norm": 2.8600575923919678,
|
| 42679 |
+
"learning_rate": 8.607716587079074e-05,
|
| 42680 |
+
"loss": 0.0742,
|
| 42681 |
+
"step": 6096
|
| 42682 |
+
},
|
| 42683 |
+
{
|
| 42684 |
+
"epoch": 0.2465252155629909,
|
| 42685 |
+
"grad_norm": 2.9021806716918945,
|
| 42686 |
+
"learning_rate": 8.607275012062532e-05,
|
| 42687 |
+
"loss": 0.0799,
|
| 42688 |
+
"step": 6097
|
| 42689 |
+
},
|
| 42690 |
+
{
|
| 42691 |
+
"epoch": 0.2465656494182579,
|
| 42692 |
+
"grad_norm": 4.6386332511901855,
|
| 42693 |
+
"learning_rate": 8.606833378362639e-05,
|
| 42694 |
+
"loss": 0.1038,
|
| 42695 |
+
"step": 6098
|
| 42696 |
+
},
|
| 42697 |
+
{
|
| 42698 |
+
"epoch": 0.24660608327352493,
|
| 42699 |
+
"grad_norm": 5.46524715423584,
|
| 42700 |
+
"learning_rate": 8.606391685986579e-05,
|
| 42701 |
+
"loss": 0.1079,
|
| 42702 |
+
"step": 6099
|
| 42703 |
+
},
|
| 42704 |
+
{
|
| 42705 |
+
"epoch": 0.24664651712879193,
|
| 42706 |
+
"grad_norm": 4.315202236175537,
|
| 42707 |
+
"learning_rate": 8.605949934941537e-05,
|
| 42708 |
+
"loss": 0.1017,
|
| 42709 |
+
"step": 6100
|
| 42710 |
+
},
|
| 42711 |
+
{
|
| 42712 |
+
"epoch": 0.24668695098405896,
|
| 42713 |
+
"grad_norm": 4.34230375289917,
|
| 42714 |
+
"learning_rate": 8.6055081252347e-05,
|
| 42715 |
+
"loss": 0.135,
|
| 42716 |
+
"step": 6101
|
| 42717 |
+
},
|
| 42718 |
+
{
|
| 42719 |
+
"epoch": 0.24672738483932596,
|
| 42720 |
+
"grad_norm": 10.488526344299316,
|
| 42721 |
+
"learning_rate": 8.605066256873256e-05,
|
| 42722 |
+
"loss": 0.2443,
|
| 42723 |
+
"step": 6102
|
| 42724 |
+
},
|
| 42725 |
+
{
|
| 42726 |
+
"epoch": 0.246767818694593,
|
| 42727 |
+
"grad_norm": 4.642398357391357,
|
| 42728 |
+
"learning_rate": 8.604624329864392e-05,
|
| 42729 |
+
"loss": 0.1518,
|
| 42730 |
+
"step": 6103
|
| 42731 |
+
},
|
| 42732 |
+
{
|
| 42733 |
+
"epoch": 0.24680825254986,
|
| 42734 |
+
"grad_norm": 3.78501296043396,
|
| 42735 |
+
"learning_rate": 8.604182344215298e-05,
|
| 42736 |
+
"loss": 0.0736,
|
| 42737 |
+
"step": 6104
|
| 42738 |
+
},
|
| 42739 |
+
{
|
| 42740 |
+
"epoch": 0.246848686405127,
|
| 42741 |
+
"grad_norm": 9.167596817016602,
|
| 42742 |
+
"learning_rate": 8.603740299933164e-05,
|
| 42743 |
+
"loss": 0.3418,
|
| 42744 |
+
"step": 6105
|
| 42745 |
+
},
|
| 42746 |
+
{
|
| 42747 |
+
"epoch": 0.24688912026039403,
|
| 42748 |
+
"grad_norm": 7.805789947509766,
|
| 42749 |
+
"learning_rate": 8.603298197025181e-05,
|
| 42750 |
+
"loss": 0.4001,
|
| 42751 |
+
"step": 6106
|
| 42752 |
+
},
|
| 42753 |
+
{
|
| 42754 |
+
"epoch": 0.24692955411566103,
|
| 42755 |
+
"grad_norm": 6.4627275466918945,
|
| 42756 |
+
"learning_rate": 8.60285603549854e-05,
|
| 42757 |
+
"loss": 0.2139,
|
| 42758 |
+
"step": 6107
|
| 42759 |
+
},
|
| 42760 |
+
{
|
| 42761 |
+
"epoch": 0.24696998797092806,
|
| 42762 |
+
"grad_norm": 2.7034740447998047,
|
| 42763 |
+
"learning_rate": 8.602413815360439e-05,
|
| 42764 |
+
"loss": 0.0862,
|
| 42765 |
+
"step": 6108
|
| 42766 |
+
},
|
| 42767 |
+
{
|
| 42768 |
+
"epoch": 0.24701042182619506,
|
| 42769 |
+
"grad_norm": 3.240323066711426,
|
| 42770 |
+
"learning_rate": 8.601971536618067e-05,
|
| 42771 |
+
"loss": 0.1273,
|
| 42772 |
+
"step": 6109
|
| 42773 |
+
},
|
| 42774 |
+
{
|
| 42775 |
+
"epoch": 0.2470508556814621,
|
| 42776 |
+
"grad_norm": 4.7289886474609375,
|
| 42777 |
+
"learning_rate": 8.60152919927862e-05,
|
| 42778 |
+
"loss": 0.2225,
|
| 42779 |
+
"step": 6110
|
| 42780 |
+
},
|
| 42781 |
+
{
|
| 42782 |
+
"epoch": 0.2470912895367291,
|
| 42783 |
+
"grad_norm": 3.309140920639038,
|
| 42784 |
+
"learning_rate": 8.601086803349296e-05,
|
| 42785 |
+
"loss": 0.0905,
|
| 42786 |
+
"step": 6111
|
| 42787 |
+
},
|
| 42788 |
+
{
|
| 42789 |
+
"epoch": 0.24713172339199613,
|
| 42790 |
+
"grad_norm": 6.798061847686768,
|
| 42791 |
+
"learning_rate": 8.60064434883729e-05,
|
| 42792 |
+
"loss": 0.2208,
|
| 42793 |
+
"step": 6112
|
| 42794 |
+
},
|
| 42795 |
+
{
|
| 42796 |
+
"epoch": 0.24717215724726313,
|
| 42797 |
+
"grad_norm": 3.6816961765289307,
|
| 42798 |
+
"learning_rate": 8.6002018357498e-05,
|
| 42799 |
+
"loss": 0.1211,
|
| 42800 |
+
"step": 6113
|
| 42801 |
+
},
|
| 42802 |
+
{
|
| 42803 |
+
"epoch": 0.24721259110253016,
|
| 42804 |
+
"grad_norm": 3.832526683807373,
|
| 42805 |
+
"learning_rate": 8.599759264094025e-05,
|
| 42806 |
+
"loss": 0.0823,
|
| 42807 |
+
"step": 6114
|
| 42808 |
+
},
|
| 42809 |
+
{
|
| 42810 |
+
"epoch": 0.24725302495779716,
|
| 42811 |
+
"grad_norm": 3.5318474769592285,
|
| 42812 |
+
"learning_rate": 8.599316633877164e-05,
|
| 42813 |
+
"loss": 0.1528,
|
| 42814 |
+
"step": 6115
|
| 42815 |
+
},
|
| 42816 |
+
{
|
| 42817 |
+
"epoch": 0.24729345881306417,
|
| 42818 |
+
"grad_norm": 3.617408514022827,
|
| 42819 |
+
"learning_rate": 8.59887394510642e-05,
|
| 42820 |
+
"loss": 0.0787,
|
| 42821 |
+
"step": 6116
|
| 42822 |
+
},
|
| 42823 |
+
{
|
| 42824 |
+
"epoch": 0.2473338926683312,
|
| 42825 |
+
"grad_norm": 5.636387348175049,
|
| 42826 |
+
"learning_rate": 8.598431197788994e-05,
|
| 42827 |
+
"loss": 0.1539,
|
| 42828 |
+
"step": 6117
|
| 42829 |
+
},
|
| 42830 |
+
{
|
| 42831 |
+
"epoch": 0.2473743265235982,
|
| 42832 |
+
"grad_norm": 6.630018711090088,
|
| 42833 |
+
"learning_rate": 8.597988391932089e-05,
|
| 42834 |
+
"loss": 0.4171,
|
| 42835 |
+
"step": 6118
|
| 42836 |
+
},
|
| 42837 |
+
{
|
| 42838 |
+
"epoch": 0.24741476037886523,
|
| 42839 |
+
"grad_norm": 7.672569751739502,
|
| 42840 |
+
"learning_rate": 8.597545527542905e-05,
|
| 42841 |
+
"loss": 0.3314,
|
| 42842 |
+
"step": 6119
|
| 42843 |
+
},
|
| 42844 |
+
{
|
| 42845 |
+
"epoch": 0.24745519423413223,
|
| 42846 |
+
"grad_norm": 7.134288787841797,
|
| 42847 |
+
"learning_rate": 8.597102604628651e-05,
|
| 42848 |
+
"loss": 0.2681,
|
| 42849 |
+
"step": 6120
|
| 42850 |
+
},
|
| 42851 |
+
{
|
| 42852 |
+
"epoch": 0.24749562808939926,
|
| 42853 |
+
"grad_norm": 7.464917182922363,
|
| 42854 |
+
"learning_rate": 8.59665962319653e-05,
|
| 42855 |
+
"loss": 0.1661,
|
| 42856 |
+
"step": 6121
|
| 42857 |
+
},
|
| 42858 |
+
{
|
| 42859 |
+
"epoch": 0.24753606194466626,
|
| 42860 |
+
"grad_norm": 4.142001152038574,
|
| 42861 |
+
"learning_rate": 8.59621658325375e-05,
|
| 42862 |
+
"loss": 0.2082,
|
| 42863 |
+
"step": 6122
|
| 42864 |
+
},
|
| 42865 |
+
{
|
| 42866 |
+
"epoch": 0.2475764957999333,
|
| 42867 |
+
"grad_norm": 4.803020000457764,
|
| 42868 |
+
"learning_rate": 8.595773484807516e-05,
|
| 42869 |
+
"loss": 0.2261,
|
| 42870 |
+
"step": 6123
|
| 42871 |
+
},
|
| 42872 |
+
{
|
| 42873 |
+
"epoch": 0.2476169296552003,
|
| 42874 |
+
"grad_norm": 3.1889495849609375,
|
| 42875 |
+
"learning_rate": 8.595330327865038e-05,
|
| 42876 |
+
"loss": 0.141,
|
| 42877 |
+
"step": 6124
|
| 42878 |
+
},
|
| 42879 |
+
{
|
| 42880 |
+
"epoch": 0.24765736351046733,
|
| 42881 |
+
"grad_norm": 3.5087060928344727,
|
| 42882 |
+
"learning_rate": 8.594887112433528e-05,
|
| 42883 |
+
"loss": 0.1012,
|
| 42884 |
+
"step": 6125
|
| 42885 |
+
},
|
| 42886 |
+
{
|
| 42887 |
+
"epoch": 0.24769779736573433,
|
| 42888 |
+
"grad_norm": 3.0082900524139404,
|
| 42889 |
+
"learning_rate": 8.594443838520191e-05,
|
| 42890 |
+
"loss": 0.1087,
|
| 42891 |
+
"step": 6126
|
| 42892 |
+
},
|
| 42893 |
+
{
|
| 42894 |
+
"epoch": 0.24773823122100133,
|
| 42895 |
+
"grad_norm": 5.638223648071289,
|
| 42896 |
+
"learning_rate": 8.594000506132241e-05,
|
| 42897 |
+
"loss": 0.1215,
|
| 42898 |
+
"step": 6127
|
| 42899 |
+
},
|
| 42900 |
+
{
|
| 42901 |
+
"epoch": 0.24777866507626836,
|
| 42902 |
+
"grad_norm": 2.4479122161865234,
|
| 42903 |
+
"learning_rate": 8.59355711527689e-05,
|
| 42904 |
+
"loss": 0.1683,
|
| 42905 |
+
"step": 6128
|
| 42906 |
+
},
|
| 42907 |
+
{
|
| 42908 |
+
"epoch": 0.24781909893153536,
|
| 42909 |
+
"grad_norm": 4.197991847991943,
|
| 42910 |
+
"learning_rate": 8.59311366596135e-05,
|
| 42911 |
+
"loss": 0.093,
|
| 42912 |
+
"step": 6129
|
| 42913 |
+
},
|
| 42914 |
+
{
|
| 42915 |
+
"epoch": 0.2478595327868024,
|
| 42916 |
+
"grad_norm": 6.671213626861572,
|
| 42917 |
+
"learning_rate": 8.592670158192838e-05,
|
| 42918 |
+
"loss": 0.2066,
|
| 42919 |
+
"step": 6130
|
| 42920 |
+
},
|
| 42921 |
+
{
|
| 42922 |
+
"epoch": 0.2478999666420694,
|
| 42923 |
+
"grad_norm": 5.721414566040039,
|
| 42924 |
+
"learning_rate": 8.592226591978566e-05,
|
| 42925 |
+
"loss": 0.2359,
|
| 42926 |
+
"step": 6131
|
| 42927 |
+
},
|
| 42928 |
+
{
|
| 42929 |
+
"epoch": 0.24794040049733643,
|
| 42930 |
+
"grad_norm": 7.26845121383667,
|
| 42931 |
+
"learning_rate": 8.59178296732575e-05,
|
| 42932 |
+
"loss": 0.1464,
|
| 42933 |
+
"step": 6132
|
| 42934 |
+
},
|
| 42935 |
+
{
|
| 42936 |
+
"epoch": 0.24798083435260343,
|
| 42937 |
+
"grad_norm": 3.415755271911621,
|
| 42938 |
+
"learning_rate": 8.59133928424161e-05,
|
| 42939 |
+
"loss": 0.1107,
|
| 42940 |
+
"step": 6133
|
| 42941 |
+
},
|
| 42942 |
+
{
|
| 42943 |
+
"epoch": 0.24802126820787046,
|
| 42944 |
+
"grad_norm": 2.2114908695220947,
|
| 42945 |
+
"learning_rate": 8.59089554273336e-05,
|
| 42946 |
+
"loss": 0.0952,
|
| 42947 |
+
"step": 6134
|
| 42948 |
+
},
|
| 42949 |
+
{
|
| 42950 |
+
"epoch": 0.24806170206313746,
|
| 42951 |
+
"grad_norm": 4.224102973937988,
|
| 42952 |
+
"learning_rate": 8.590451742808222e-05,
|
| 42953 |
+
"loss": 0.1822,
|
| 42954 |
+
"step": 6135
|
| 42955 |
+
},
|
| 42956 |
+
{
|
| 42957 |
+
"epoch": 0.2481021359184045,
|
| 42958 |
+
"grad_norm": 5.626123428344727,
|
| 42959 |
+
"learning_rate": 8.590007884473413e-05,
|
| 42960 |
+
"loss": 0.1277,
|
| 42961 |
+
"step": 6136
|
| 42962 |
+
},
|
| 42963 |
+
{
|
| 42964 |
+
"epoch": 0.2481425697736715,
|
| 42965 |
+
"grad_norm": 2.7514543533325195,
|
| 42966 |
+
"learning_rate": 8.589563967736154e-05,
|
| 42967 |
+
"loss": 0.2924,
|
| 42968 |
+
"step": 6137
|
| 42969 |
+
},
|
| 42970 |
+
{
|
| 42971 |
+
"epoch": 0.2481830036289385,
|
| 42972 |
+
"grad_norm": 4.606257438659668,
|
| 42973 |
+
"learning_rate": 8.589119992603669e-05,
|
| 42974 |
+
"loss": 0.2058,
|
| 42975 |
+
"step": 6138
|
| 42976 |
+
},
|
| 42977 |
+
{
|
| 42978 |
+
"epoch": 0.24822343748420553,
|
| 42979 |
+
"grad_norm": 6.313202381134033,
|
| 42980 |
+
"learning_rate": 8.58867595908318e-05,
|
| 42981 |
+
"loss": 0.1745,
|
| 42982 |
+
"step": 6139
|
| 42983 |
+
},
|
| 42984 |
+
{
|
| 42985 |
+
"epoch": 0.24826387133947253,
|
| 42986 |
+
"grad_norm": 4.425397872924805,
|
| 42987 |
+
"learning_rate": 8.588231867181908e-05,
|
| 42988 |
+
"loss": 0.1273,
|
| 42989 |
+
"step": 6140
|
| 42990 |
+
},
|
| 42991 |
+
{
|
| 42992 |
+
"epoch": 0.24830430519473956,
|
| 42993 |
+
"grad_norm": 4.079669952392578,
|
| 42994 |
+
"learning_rate": 8.587787716907081e-05,
|
| 42995 |
+
"loss": 0.1124,
|
| 42996 |
+
"step": 6141
|
| 42997 |
+
},
|
| 42998 |
+
{
|
| 42999 |
+
"epoch": 0.24834473905000656,
|
| 43000 |
+
"grad_norm": 2.386849880218506,
|
| 43001 |
+
"learning_rate": 8.587343508265923e-05,
|
| 43002 |
+
"loss": 0.1633,
|
| 43003 |
+
"step": 6142
|
| 43004 |
+
},
|
| 43005 |
+
{
|
| 43006 |
+
"epoch": 0.2483851729052736,
|
| 43007 |
+
"grad_norm": 7.280966758728027,
|
| 43008 |
+
"learning_rate": 8.58689924126566e-05,
|
| 43009 |
+
"loss": 0.2683,
|
| 43010 |
+
"step": 6143
|
| 43011 |
+
},
|
| 43012 |
+
{
|
| 43013 |
+
"epoch": 0.2484256067605406,
|
| 43014 |
+
"grad_norm": 2.142650604248047,
|
| 43015 |
+
"learning_rate": 8.586454915913517e-05,
|
| 43016 |
+
"loss": 0.1698,
|
| 43017 |
+
"step": 6144
|
| 43018 |
+
},
|
| 43019 |
+
{
|
| 43020 |
+
"epoch": 0.24846604061580763,
|
| 43021 |
+
"grad_norm": 7.144465923309326,
|
| 43022 |
+
"learning_rate": 8.586010532216728e-05,
|
| 43023 |
+
"loss": 0.1161,
|
| 43024 |
+
"step": 6145
|
| 43025 |
+
},
|
| 43026 |
+
{
|
| 43027 |
+
"epoch": 0.24850647447107463,
|
| 43028 |
+
"grad_norm": 4.681589126586914,
|
| 43029 |
+
"learning_rate": 8.585566090182518e-05,
|
| 43030 |
+
"loss": 0.1186,
|
| 43031 |
+
"step": 6146
|
| 43032 |
+
},
|
| 43033 |
+
{
|
| 43034 |
+
"epoch": 0.24854690832634166,
|
| 43035 |
+
"grad_norm": 3.8650546073913574,
|
| 43036 |
+
"learning_rate": 8.58512158981812e-05,
|
| 43037 |
+
"loss": 0.1997,
|
| 43038 |
+
"step": 6147
|
| 43039 |
+
},
|
| 43040 |
+
{
|
| 43041 |
+
"epoch": 0.24858734218160866,
|
| 43042 |
+
"grad_norm": 9.318419456481934,
|
| 43043 |
+
"learning_rate": 8.584677031130761e-05,
|
| 43044 |
+
"loss": 0.2923,
|
| 43045 |
+
"step": 6148
|
| 43046 |
+
},
|
| 43047 |
+
{
|
| 43048 |
+
"epoch": 0.24862777603687566,
|
| 43049 |
+
"grad_norm": 4.759102821350098,
|
| 43050 |
+
"learning_rate": 8.584232414127678e-05,
|
| 43051 |
+
"loss": 0.1804,
|
| 43052 |
+
"step": 6149
|
| 43053 |
+
},
|
| 43054 |
+
{
|
| 43055 |
+
"epoch": 0.2486682098921427,
|
| 43056 |
+
"grad_norm": 3.2844924926757812,
|
| 43057 |
+
"learning_rate": 8.5837877388161e-05,
|
| 43058 |
+
"loss": 0.1689,
|
| 43059 |
+
"step": 6150
|
| 43060 |
+
},
|
| 43061 |
+
{
|
| 43062 |
+
"epoch": 0.2487086437474097,
|
| 43063 |
+
"grad_norm": 9.399534225463867,
|
| 43064 |
+
"learning_rate": 8.583343005203263e-05,
|
| 43065 |
+
"loss": 0.3574,
|
| 43066 |
+
"step": 6151
|
| 43067 |
+
},
|
| 43068 |
+
{
|
| 43069 |
+
"epoch": 0.24874907760267673,
|
| 43070 |
+
"grad_norm": 4.286575794219971,
|
| 43071 |
+
"learning_rate": 8.582898213296402e-05,
|
| 43072 |
+
"loss": 0.2597,
|
| 43073 |
+
"step": 6152
|
| 43074 |
+
},
|
| 43075 |
+
{
|
| 43076 |
+
"epoch": 0.24878951145794373,
|
| 43077 |
+
"grad_norm": 5.7185139656066895,
|
| 43078 |
+
"learning_rate": 8.582453363102752e-05,
|
| 43079 |
+
"loss": 0.2317,
|
| 43080 |
+
"step": 6153
|
| 43081 |
+
},
|
| 43082 |
+
{
|
| 43083 |
+
"epoch": 0.24882994531321076,
|
| 43084 |
+
"grad_norm": 5.736219882965088,
|
| 43085 |
+
"learning_rate": 8.58200845462955e-05,
|
| 43086 |
+
"loss": 0.1812,
|
| 43087 |
+
"step": 6154
|
| 43088 |
+
},
|
| 43089 |
+
{
|
| 43090 |
+
"epoch": 0.24887037916847776,
|
| 43091 |
+
"grad_norm": 5.837985515594482,
|
| 43092 |
+
"learning_rate": 8.581563487884035e-05,
|
| 43093 |
+
"loss": 0.1813,
|
| 43094 |
+
"step": 6155
|
| 43095 |
+
},
|
| 43096 |
+
{
|
| 43097 |
+
"epoch": 0.2489108130237448,
|
| 43098 |
+
"grad_norm": 5.658602237701416,
|
| 43099 |
+
"learning_rate": 8.581118462873446e-05,
|
| 43100 |
+
"loss": 0.1545,
|
| 43101 |
+
"step": 6156
|
| 43102 |
+
},
|
| 43103 |
+
{
|
| 43104 |
+
"epoch": 0.2489512468790118,
|
| 43105 |
+
"grad_norm": 4.969576358795166,
|
| 43106 |
+
"learning_rate": 8.58067337960502e-05,
|
| 43107 |
+
"loss": 0.1486,
|
| 43108 |
+
"step": 6157
|
| 43109 |
+
},
|
| 43110 |
+
{
|
| 43111 |
+
"epoch": 0.24899168073427883,
|
| 43112 |
+
"grad_norm": 2.319491386413574,
|
| 43113 |
+
"learning_rate": 8.580228238085999e-05,
|
| 43114 |
+
"loss": 0.1649,
|
| 43115 |
+
"step": 6158
|
| 43116 |
+
},
|
| 43117 |
+
{
|
| 43118 |
+
"epoch": 0.24903211458954583,
|
| 43119 |
+
"grad_norm": 3.767211437225342,
|
| 43120 |
+
"learning_rate": 8.579783038323626e-05,
|
| 43121 |
+
"loss": 0.1828,
|
| 43122 |
+
"step": 6159
|
| 43123 |
+
},
|
| 43124 |
+
{
|
| 43125 |
+
"epoch": 0.24907254844481283,
|
| 43126 |
+
"grad_norm": 3.399191379547119,
|
| 43127 |
+
"learning_rate": 8.579337780325142e-05,
|
| 43128 |
+
"loss": 0.1291,
|
| 43129 |
+
"step": 6160
|
| 43130 |
+
},
|
| 43131 |
+
{
|
| 43132 |
+
"epoch": 0.24911298230007986,
|
| 43133 |
+
"grad_norm": 3.8261640071868896,
|
| 43134 |
+
"learning_rate": 8.57889246409779e-05,
|
| 43135 |
+
"loss": 0.1082,
|
| 43136 |
+
"step": 6161
|
| 43137 |
+
},
|
| 43138 |
+
{
|
| 43139 |
+
"epoch": 0.24915341615534686,
|
| 43140 |
+
"grad_norm": 2.4140303134918213,
|
| 43141 |
+
"learning_rate": 8.578447089648816e-05,
|
| 43142 |
+
"loss": 0.1318,
|
| 43143 |
+
"step": 6162
|
| 43144 |
+
},
|
| 43145 |
+
{
|
| 43146 |
+
"epoch": 0.2491938500106139,
|
| 43147 |
+
"grad_norm": 7.104363918304443,
|
| 43148 |
+
"learning_rate": 8.578001656985466e-05,
|
| 43149 |
+
"loss": 0.1731,
|
| 43150 |
+
"step": 6163
|
| 43151 |
+
},
|
| 43152 |
+
{
|
| 43153 |
+
"epoch": 0.2492342838658809,
|
| 43154 |
+
"grad_norm": 7.832826137542725,
|
| 43155 |
+
"learning_rate": 8.577556166114984e-05,
|
| 43156 |
+
"loss": 0.299,
|
| 43157 |
+
"step": 6164
|
| 43158 |
+
},
|
| 43159 |
+
{
|
| 43160 |
+
"epoch": 0.24927471772114793,
|
| 43161 |
+
"grad_norm": 8.464363098144531,
|
| 43162 |
+
"learning_rate": 8.577110617044618e-05,
|
| 43163 |
+
"loss": 0.3275,
|
| 43164 |
+
"step": 6165
|
| 43165 |
+
},
|
| 43166 |
+
{
|
| 43167 |
+
"epoch": 0.24931515157641493,
|
| 43168 |
+
"grad_norm": 4.684128761291504,
|
| 43169 |
+
"learning_rate": 8.576665009781616e-05,
|
| 43170 |
+
"loss": 0.2326,
|
| 43171 |
+
"step": 6166
|
| 43172 |
+
},
|
| 43173 |
+
{
|
| 43174 |
+
"epoch": 0.24935558543168196,
|
| 43175 |
+
"grad_norm": 7.360625743865967,
|
| 43176 |
+
"learning_rate": 8.576219344333227e-05,
|
| 43177 |
+
"loss": 0.2637,
|
| 43178 |
+
"step": 6167
|
| 43179 |
+
},
|
| 43180 |
+
{
|
| 43181 |
+
"epoch": 0.24939601928694896,
|
| 43182 |
+
"grad_norm": 5.580816268920898,
|
| 43183 |
+
"learning_rate": 8.575773620706704e-05,
|
| 43184 |
+
"loss": 0.2078,
|
| 43185 |
+
"step": 6168
|
| 43186 |
+
},
|
| 43187 |
+
{
|
| 43188 |
+
"epoch": 0.24943645314221596,
|
| 43189 |
+
"grad_norm": 4.870668411254883,
|
| 43190 |
+
"learning_rate": 8.575327838909295e-05,
|
| 43191 |
+
"loss": 0.1566,
|
| 43192 |
+
"step": 6169
|
| 43193 |
+
},
|
| 43194 |
+
{
|
| 43195 |
+
"epoch": 0.249476886997483,
|
| 43196 |
+
"grad_norm": 3.1624975204467773,
|
| 43197 |
+
"learning_rate": 8.574881998948252e-05,
|
| 43198 |
+
"loss": 0.2021,
|
| 43199 |
+
"step": 6170
|
| 43200 |
+
},
|
| 43201 |
+
{
|
| 43202 |
+
"epoch": 0.24951732085275,
|
| 43203 |
+
"grad_norm": 2.021010398864746,
|
| 43204 |
+
"learning_rate": 8.574436100830829e-05,
|
| 43205 |
+
"loss": 0.0836,
|
| 43206 |
+
"step": 6171
|
| 43207 |
+
},
|
| 43208 |
+
{
|
| 43209 |
+
"epoch": 0.24955775470801703,
|
| 43210 |
+
"grad_norm": 2.359396457672119,
|
| 43211 |
+
"learning_rate": 8.573990144564281e-05,
|
| 43212 |
+
"loss": 0.1261,
|
| 43213 |
+
"step": 6172
|
| 43214 |
+
},
|
| 43215 |
+
{
|
| 43216 |
+
"epoch": 0.24959818856328403,
|
| 43217 |
+
"grad_norm": 14.166296005249023,
|
| 43218 |
+
"learning_rate": 8.57354413015586e-05,
|
| 43219 |
+
"loss": 0.3957,
|
| 43220 |
+
"step": 6173
|
| 43221 |
+
},
|
| 43222 |
+
{
|
| 43223 |
+
"epoch": 0.24963862241855106,
|
| 43224 |
+
"grad_norm": 4.486057281494141,
|
| 43225 |
+
"learning_rate": 8.573098057612822e-05,
|
| 43226 |
+
"loss": 0.1729,
|
| 43227 |
+
"step": 6174
|
| 43228 |
+
},
|
| 43229 |
+
{
|
| 43230 |
+
"epoch": 0.24967905627381806,
|
| 43231 |
+
"grad_norm": 4.341570854187012,
|
| 43232 |
+
"learning_rate": 8.572651926942428e-05,
|
| 43233 |
+
"loss": 0.0956,
|
| 43234 |
+
"step": 6175
|
| 43235 |
+
},
|
| 43236 |
+
{
|
| 43237 |
+
"epoch": 0.2497194901290851,
|
| 43238 |
+
"grad_norm": 6.579067707061768,
|
| 43239 |
+
"learning_rate": 8.572205738151931e-05,
|
| 43240 |
+
"loss": 0.1692,
|
| 43241 |
+
"step": 6176
|
| 43242 |
+
},
|
| 43243 |
+
{
|
| 43244 |
+
"epoch": 0.2497599239843521,
|
| 43245 |
+
"grad_norm": 4.9769206047058105,
|
| 43246 |
+
"learning_rate": 8.571759491248593e-05,
|
| 43247 |
+
"loss": 0.1608,
|
| 43248 |
+
"step": 6177
|
| 43249 |
+
},
|
| 43250 |
+
{
|
| 43251 |
+
"epoch": 0.24980035783961912,
|
| 43252 |
+
"grad_norm": 11.039288520812988,
|
| 43253 |
+
"learning_rate": 8.57131318623967e-05,
|
| 43254 |
+
"loss": 0.3244,
|
| 43255 |
+
"step": 6178
|
| 43256 |
+
},
|
| 43257 |
+
{
|
| 43258 |
+
"epoch": 0.24984079169488613,
|
| 43259 |
+
"grad_norm": 2.834900140762329,
|
| 43260 |
+
"learning_rate": 8.570866823132425e-05,
|
| 43261 |
+
"loss": 0.1042,
|
| 43262 |
+
"step": 6179
|
| 43263 |
+
},
|
| 43264 |
+
{
|
| 43265 |
+
"epoch": 0.24988122555015313,
|
| 43266 |
+
"grad_norm": 5.615602016448975,
|
| 43267 |
+
"learning_rate": 8.57042040193412e-05,
|
| 43268 |
+
"loss": 0.1991,
|
| 43269 |
+
"step": 6180
|
| 43270 |
+
},
|
| 43271 |
+
{
|
| 43272 |
+
"epoch": 0.24992165940542016,
|
| 43273 |
+
"grad_norm": 5.0720086097717285,
|
| 43274 |
+
"learning_rate": 8.569973922652015e-05,
|
| 43275 |
+
"loss": 0.1036,
|
| 43276 |
+
"step": 6181
|
| 43277 |
+
},
|
| 43278 |
+
{
|
| 43279 |
+
"epoch": 0.24996209326068716,
|
| 43280 |
+
"grad_norm": 4.303607940673828,
|
| 43281 |
+
"learning_rate": 8.569527385293375e-05,
|
| 43282 |
+
"loss": 0.2971,
|
| 43283 |
+
"step": 6182
|
| 43284 |
+
},
|
| 43285 |
+
{
|
| 43286 |
+
"epoch": 0.25000252711595417,
|
| 43287 |
+
"grad_norm": 2.68996262550354,
|
| 43288 |
+
"learning_rate": 8.569080789865463e-05,
|
| 43289 |
+
"loss": 0.0876,
|
| 43290 |
+
"step": 6183
|
| 43291 |
+
},
|
| 43292 |
+
{
|
| 43293 |
+
"epoch": 0.2500429609712212,
|
| 43294 |
+
"grad_norm": 3.8559789657592773,
|
| 43295 |
+
"learning_rate": 8.568634136375547e-05,
|
| 43296 |
+
"loss": 0.1207,
|
| 43297 |
+
"step": 6184
|
| 43298 |
+
},
|
| 43299 |
+
{
|
| 43300 |
+
"epoch": 0.2500833948264882,
|
| 43301 |
+
"grad_norm": 6.243417739868164,
|
| 43302 |
+
"learning_rate": 8.568187424830889e-05,
|
| 43303 |
+
"loss": 0.2087,
|
| 43304 |
+
"step": 6185
|
| 43305 |
+
},
|
| 43306 |
+
{
|
| 43307 |
+
"epoch": 0.25012382868175526,
|
| 43308 |
+
"grad_norm": 5.405614376068115,
|
| 43309 |
+
"learning_rate": 8.567740655238761e-05,
|
| 43310 |
+
"loss": 0.1594,
|
| 43311 |
+
"step": 6186
|
| 43312 |
+
},
|
| 43313 |
+
{
|
| 43314 |
+
"epoch": 0.25016426253702223,
|
| 43315 |
+
"grad_norm": 3.042715311050415,
|
| 43316 |
+
"learning_rate": 8.567293827606427e-05,
|
| 43317 |
+
"loss": 0.0963,
|
| 43318 |
+
"step": 6187
|
| 43319 |
+
},
|
| 43320 |
+
{
|
| 43321 |
+
"epoch": 0.25020469639228926,
|
| 43322 |
+
"grad_norm": 8.87225341796875,
|
| 43323 |
+
"learning_rate": 8.566846941941158e-05,
|
| 43324 |
+
"loss": 0.243,
|
| 43325 |
+
"step": 6188
|
| 43326 |
+
},
|
| 43327 |
+
{
|
| 43328 |
+
"epoch": 0.2502451302475563,
|
| 43329 |
+
"grad_norm": 3.334763288497925,
|
| 43330 |
+
"learning_rate": 8.566399998250224e-05,
|
| 43331 |
+
"loss": 0.0835,
|
| 43332 |
+
"step": 6189
|
| 43333 |
+
},
|
| 43334 |
+
{
|
| 43335 |
+
"epoch": 0.2502855641028233,
|
| 43336 |
+
"grad_norm": 5.4543538093566895,
|
| 43337 |
+
"learning_rate": 8.565952996540893e-05,
|
| 43338 |
+
"loss": 0.1584,
|
| 43339 |
+
"step": 6190
|
| 43340 |
+
},
|
| 43341 |
+
{
|
| 43342 |
+
"epoch": 0.2503259979580903,
|
| 43343 |
+
"grad_norm": 6.160861968994141,
|
| 43344 |
+
"learning_rate": 8.565505936820441e-05,
|
| 43345 |
+
"loss": 0.1298,
|
| 43346 |
+
"step": 6191
|
| 43347 |
+
},
|
| 43348 |
+
{
|
| 43349 |
+
"epoch": 0.2503664318133573,
|
| 43350 |
+
"grad_norm": 5.069806098937988,
|
| 43351 |
+
"learning_rate": 8.565058819096139e-05,
|
| 43352 |
+
"loss": 0.2268,
|
| 43353 |
+
"step": 6192
|
| 43354 |
+
},
|
| 43355 |
+
{
|
| 43356 |
+
"epoch": 0.25040686566862436,
|
| 43357 |
+
"grad_norm": 5.266380786895752,
|
| 43358 |
+
"learning_rate": 8.564611643375263e-05,
|
| 43359 |
+
"loss": 0.1274,
|
| 43360 |
+
"step": 6193
|
| 43361 |
+
},
|
| 43362 |
+
{
|
| 43363 |
+
"epoch": 0.25044729952389133,
|
| 43364 |
+
"grad_norm": 5.647245407104492,
|
| 43365 |
+
"learning_rate": 8.564164409665083e-05,
|
| 43366 |
+
"loss": 0.1708,
|
| 43367 |
+
"step": 6194
|
| 43368 |
+
},
|
| 43369 |
+
{
|
| 43370 |
+
"epoch": 0.25048773337915836,
|
| 43371 |
+
"grad_norm": 8.582027435302734,
|
| 43372 |
+
"learning_rate": 8.563717117972876e-05,
|
| 43373 |
+
"loss": 0.193,
|
| 43374 |
+
"step": 6195
|
| 43375 |
+
},
|
| 43376 |
+
{
|
| 43377 |
+
"epoch": 0.2505281672344254,
|
| 43378 |
+
"grad_norm": 4.458442687988281,
|
| 43379 |
+
"learning_rate": 8.563269768305922e-05,
|
| 43380 |
+
"loss": 0.3056,
|
| 43381 |
+
"step": 6196
|
| 43382 |
+
},
|
| 43383 |
+
{
|
| 43384 |
+
"epoch": 0.2505686010896924,
|
| 43385 |
+
"grad_norm": 3.2652697563171387,
|
| 43386 |
+
"learning_rate": 8.562822360671495e-05,
|
| 43387 |
+
"loss": 0.1163,
|
| 43388 |
+
"step": 6197
|
| 43389 |
+
},
|
| 43390 |
+
{
|
| 43391 |
+
"epoch": 0.2506090349449594,
|
| 43392 |
+
"grad_norm": 3.184579849243164,
|
| 43393 |
+
"learning_rate": 8.562374895076874e-05,
|
| 43394 |
+
"loss": 0.1153,
|
| 43395 |
+
"step": 6198
|
| 43396 |
+
},
|
| 43397 |
+
{
|
| 43398 |
+
"epoch": 0.2506494688002264,
|
| 43399 |
+
"grad_norm": 4.188009262084961,
|
| 43400 |
+
"learning_rate": 8.561927371529341e-05,
|
| 43401 |
+
"loss": 0.1579,
|
| 43402 |
+
"step": 6199
|
| 43403 |
+
},
|
| 43404 |
+
{
|
| 43405 |
+
"epoch": 0.25068990265549346,
|
| 43406 |
+
"grad_norm": 1.9505306482315063,
|
| 43407 |
+
"learning_rate": 8.561479790036175e-05,
|
| 43408 |
+
"loss": 0.0645,
|
| 43409 |
+
"step": 6200
|
| 43410 |
+
},
|
| 43411 |
+
{
|
| 43412 |
+
"epoch": 0.25073033651076043,
|
| 43413 |
+
"grad_norm": 4.53508996963501,
|
| 43414 |
+
"learning_rate": 8.561032150604655e-05,
|
| 43415 |
+
"loss": 0.1473,
|
| 43416 |
+
"step": 6201
|
| 43417 |
+
},
|
| 43418 |
+
{
|
| 43419 |
+
"epoch": 0.25077077036602746,
|
| 43420 |
+
"grad_norm": 3.168933391571045,
|
| 43421 |
+
"learning_rate": 8.560584453242064e-05,
|
| 43422 |
+
"loss": 0.1491,
|
| 43423 |
+
"step": 6202
|
| 43424 |
+
},
|
| 43425 |
+
{
|
| 43426 |
+
"epoch": 0.2508112042212945,
|
| 43427 |
+
"grad_norm": 3.8186163902282715,
|
| 43428 |
+
"learning_rate": 8.560136697955687e-05,
|
| 43429 |
+
"loss": 0.1645,
|
| 43430 |
+
"step": 6203
|
| 43431 |
+
},
|
| 43432 |
+
{
|
| 43433 |
+
"epoch": 0.2508516380765615,
|
| 43434 |
+
"grad_norm": 4.747062683105469,
|
| 43435 |
+
"learning_rate": 8.559688884752808e-05,
|
| 43436 |
+
"loss": 0.1574,
|
| 43437 |
+
"step": 6204
|
| 43438 |
+
},
|
| 43439 |
+
{
|
| 43440 |
+
"epoch": 0.2508920719318285,
|
| 43441 |
+
"grad_norm": 4.399488925933838,
|
| 43442 |
+
"learning_rate": 8.55924101364071e-05,
|
| 43443 |
+
"loss": 0.2242,
|
| 43444 |
+
"step": 6205
|
| 43445 |
+
},
|
| 43446 |
+
{
|
| 43447 |
+
"epoch": 0.25093250578709553,
|
| 43448 |
+
"grad_norm": 6.263681411743164,
|
| 43449 |
+
"learning_rate": 8.55879308462668e-05,
|
| 43450 |
+
"loss": 0.2975,
|
| 43451 |
+
"step": 6206
|
| 43452 |
+
},
|
| 43453 |
+
{
|
| 43454 |
+
"epoch": 0.25097293964236256,
|
| 43455 |
+
"grad_norm": 5.284090518951416,
|
| 43456 |
+
"learning_rate": 8.558345097718006e-05,
|
| 43457 |
+
"loss": 0.1785,
|
| 43458 |
+
"step": 6207
|
| 43459 |
+
},
|
| 43460 |
+
{
|
| 43461 |
+
"epoch": 0.2510133734976296,
|
| 43462 |
+
"grad_norm": 5.420502185821533,
|
| 43463 |
+
"learning_rate": 8.557897052921975e-05,
|
| 43464 |
+
"loss": 0.1035,
|
| 43465 |
+
"step": 6208
|
| 43466 |
+
},
|
| 43467 |
+
{
|
| 43468 |
+
"epoch": 0.25105380735289656,
|
| 43469 |
+
"grad_norm": 4.130248069763184,
|
| 43470 |
+
"learning_rate": 8.557448950245874e-05,
|
| 43471 |
+
"loss": 0.2202,
|
| 43472 |
+
"step": 6209
|
| 43473 |
+
},
|
| 43474 |
+
{
|
| 43475 |
+
"epoch": 0.2510942412081636,
|
| 43476 |
+
"grad_norm": 3.97829008102417,
|
| 43477 |
+
"learning_rate": 8.557000789696997e-05,
|
| 43478 |
+
"loss": 0.0936,
|
| 43479 |
+
"step": 6210
|
| 43480 |
+
},
|
| 43481 |
+
{
|
| 43482 |
+
"epoch": 0.2511346750634306,
|
| 43483 |
+
"grad_norm": 5.584280967712402,
|
| 43484 |
+
"learning_rate": 8.556552571282631e-05,
|
| 43485 |
+
"loss": 0.2014,
|
| 43486 |
+
"step": 6211
|
| 43487 |
+
},
|
| 43488 |
+
{
|
| 43489 |
+
"epoch": 0.2511751089186976,
|
| 43490 |
+
"grad_norm": 4.516365051269531,
|
| 43491 |
+
"learning_rate": 8.55610429501007e-05,
|
| 43492 |
+
"loss": 0.2308,
|
| 43493 |
+
"step": 6212
|
| 43494 |
+
},
|
| 43495 |
+
{
|
| 43496 |
+
"epoch": 0.25121554277396463,
|
| 43497 |
+
"grad_norm": 4.201866626739502,
|
| 43498 |
+
"learning_rate": 8.555655960886603e-05,
|
| 43499 |
+
"loss": 0.1656,
|
| 43500 |
+
"step": 6213
|
| 43501 |
+
},
|
| 43502 |
+
{
|
| 43503 |
+
"epoch": 0.25125597662923166,
|
| 43504 |
+
"grad_norm": 5.899411201477051,
|
| 43505 |
+
"learning_rate": 8.555207568919529e-05,
|
| 43506 |
+
"loss": 0.2022,
|
| 43507 |
+
"step": 6214
|
| 43508 |
}
|
| 43509 |
],
|
| 43510 |
"logging_steps": 1,
|
|
|
|
| 43524 |
"attributes": {}
|
| 43525 |
}
|
| 43526 |
},
|
| 43527 |
+
"total_flos": 3.8434118781055795e+17,
|
| 43528 |
"train_batch_size": 4,
|
| 43529 |
"trial_name": null,
|
| 43530 |
"trial_params": null
|