Training in progress, step 2800, 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 201361312
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a26138f4d4689fa8b725947d495dbade33603970d0d8a9e87c0397e9d80c7171
|
| 3 |
size 201361312
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 102537812
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f14d6aacf8d6e623a81646d05b0bb1902fc343ef726e5e2b8552df92f76e54fc
|
| 3 |
size 102537812
|
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:cf247cd6fc12421a25e03b9f7afc154d921f893b113fe13081b8331bed547105
|
| 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 sha256:57890bbb36f903ddde92f175a8734aa474875f7006c17bc430e029ee05bcba34
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 1.602339506149292,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-2200",
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 200,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -18319,6 +18319,1414 @@
|
|
| 18319 |
"eval_samples_per_second": 8.737,
|
| 18320 |
"eval_steps_per_second": 4.374,
|
| 18321 |
"step": 2600
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18322 |
}
|
| 18323 |
],
|
| 18324 |
"logging_steps": 1,
|
|
@@ -18333,7 +19741,7 @@
|
|
| 18333 |
"early_stopping_threshold": 0.0
|
| 18334 |
},
|
| 18335 |
"attributes": {
|
| 18336 |
-
"early_stopping_patience_counter":
|
| 18337 |
}
|
| 18338 |
},
|
| 18339 |
"TrainerControl": {
|
|
@@ -18342,12 +19750,12 @@
|
|
| 18342 |
"should_evaluate": false,
|
| 18343 |
"should_log": false,
|
| 18344 |
"should_save": true,
|
| 18345 |
-
"should_training_stop":
|
| 18346 |
},
|
| 18347 |
"attributes": {}
|
| 18348 |
}
|
| 18349 |
},
|
| 18350 |
-
"total_flos": 2.
|
| 18351 |
"train_batch_size": 2,
|
| 18352 |
"trial_name": null,
|
| 18353 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 1.602339506149292,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-2200",
|
| 4 |
+
"epoch": 0.13363400985550822,
|
| 5 |
"eval_steps": 200,
|
| 6 |
+
"global_step": 2800,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 18319 |
"eval_samples_per_second": 8.737,
|
| 18320 |
"eval_steps_per_second": 4.374,
|
| 18321 |
"step": 2600
|
| 18322 |
+
},
|
| 18323 |
+
{
|
| 18324 |
+
"epoch": 0.1241364498693489,
|
| 18325 |
+
"grad_norm": 6.640268802642822,
|
| 18326 |
+
"learning_rate": 0.00019917472854104044,
|
| 18327 |
+
"loss": 6.6788,
|
| 18328 |
+
"step": 2601
|
| 18329 |
+
},
|
| 18330 |
+
{
|
| 18331 |
+
"epoch": 0.12418417630144014,
|
| 18332 |
+
"grad_norm": 6.098337650299072,
|
| 18333 |
+
"learning_rate": 0.0001991740873167308,
|
| 18334 |
+
"loss": 6.4633,
|
| 18335 |
+
"step": 2602
|
| 18336 |
+
},
|
| 18337 |
+
{
|
| 18338 |
+
"epoch": 0.1242319027335314,
|
| 18339 |
+
"grad_norm": 7.400532245635986,
|
| 18340 |
+
"learning_rate": 0.0001991734458444397,
|
| 18341 |
+
"loss": 6.3654,
|
| 18342 |
+
"step": 2603
|
| 18343 |
+
},
|
| 18344 |
+
{
|
| 18345 |
+
"epoch": 0.12427962916562266,
|
| 18346 |
+
"grad_norm": 6.543416500091553,
|
| 18347 |
+
"learning_rate": 0.00019917280412416882,
|
| 18348 |
+
"loss": 6.3276,
|
| 18349 |
+
"step": 2604
|
| 18350 |
+
},
|
| 18351 |
+
{
|
| 18352 |
+
"epoch": 0.1243273555977139,
|
| 18353 |
+
"grad_norm": 9.13476276397705,
|
| 18354 |
+
"learning_rate": 0.00019917216215591972,
|
| 18355 |
+
"loss": 7.5483,
|
| 18356 |
+
"step": 2605
|
| 18357 |
+
},
|
| 18358 |
+
{
|
| 18359 |
+
"epoch": 0.12437508202980516,
|
| 18360 |
+
"grad_norm": 6.472226142883301,
|
| 18361 |
+
"learning_rate": 0.000199171519939694,
|
| 18362 |
+
"loss": 7.5186,
|
| 18363 |
+
"step": 2606
|
| 18364 |
+
},
|
| 18365 |
+
{
|
| 18366 |
+
"epoch": 0.1244228084618964,
|
| 18367 |
+
"grad_norm": 6.057101249694824,
|
| 18368 |
+
"learning_rate": 0.0001991708774754933,
|
| 18369 |
+
"loss": 5.7719,
|
| 18370 |
+
"step": 2607
|
| 18371 |
+
},
|
| 18372 |
+
{
|
| 18373 |
+
"epoch": 0.12447053489398767,
|
| 18374 |
+
"grad_norm": 8.963512420654297,
|
| 18375 |
+
"learning_rate": 0.00019917023476331922,
|
| 18376 |
+
"loss": 6.3314,
|
| 18377 |
+
"step": 2608
|
| 18378 |
+
},
|
| 18379 |
+
{
|
| 18380 |
+
"epoch": 0.12451826132607892,
|
| 18381 |
+
"grad_norm": 6.93634557723999,
|
| 18382 |
+
"learning_rate": 0.00019916959180317335,
|
| 18383 |
+
"loss": 8.5299,
|
| 18384 |
+
"step": 2609
|
| 18385 |
+
},
|
| 18386 |
+
{
|
| 18387 |
+
"epoch": 0.12456598775817017,
|
| 18388 |
+
"grad_norm": 6.162479400634766,
|
| 18389 |
+
"learning_rate": 0.00019916894859505727,
|
| 18390 |
+
"loss": 5.6562,
|
| 18391 |
+
"step": 2610
|
| 18392 |
+
},
|
| 18393 |
+
{
|
| 18394 |
+
"epoch": 0.12461371419026142,
|
| 18395 |
+
"grad_norm": 5.48561429977417,
|
| 18396 |
+
"learning_rate": 0.00019916830513897266,
|
| 18397 |
+
"loss": 6.2665,
|
| 18398 |
+
"step": 2611
|
| 18399 |
+
},
|
| 18400 |
+
{
|
| 18401 |
+
"epoch": 0.12466144062235267,
|
| 18402 |
+
"grad_norm": 3.560473680496216,
|
| 18403 |
+
"learning_rate": 0.00019916766143492106,
|
| 18404 |
+
"loss": 3.051,
|
| 18405 |
+
"step": 2612
|
| 18406 |
+
},
|
| 18407 |
+
{
|
| 18408 |
+
"epoch": 0.12470916705444393,
|
| 18409 |
+
"grad_norm": 7.0898895263671875,
|
| 18410 |
+
"learning_rate": 0.00019916701748290416,
|
| 18411 |
+
"loss": 6.9835,
|
| 18412 |
+
"step": 2613
|
| 18413 |
+
},
|
| 18414 |
+
{
|
| 18415 |
+
"epoch": 0.12475689348653518,
|
| 18416 |
+
"grad_norm": 5.331742286682129,
|
| 18417 |
+
"learning_rate": 0.00019916637328292348,
|
| 18418 |
+
"loss": 6.0956,
|
| 18419 |
+
"step": 2614
|
| 18420 |
+
},
|
| 18421 |
+
{
|
| 18422 |
+
"epoch": 0.12480461991862643,
|
| 18423 |
+
"grad_norm": 8.475786209106445,
|
| 18424 |
+
"learning_rate": 0.00019916572883498068,
|
| 18425 |
+
"loss": 7.2774,
|
| 18426 |
+
"step": 2615
|
| 18427 |
+
},
|
| 18428 |
+
{
|
| 18429 |
+
"epoch": 0.12485234635071768,
|
| 18430 |
+
"grad_norm": 7.285261154174805,
|
| 18431 |
+
"learning_rate": 0.00019916508413907736,
|
| 18432 |
+
"loss": 9.4067,
|
| 18433 |
+
"step": 2616
|
| 18434 |
+
},
|
| 18435 |
+
{
|
| 18436 |
+
"epoch": 0.12490007278280894,
|
| 18437 |
+
"grad_norm": 6.492762565612793,
|
| 18438 |
+
"learning_rate": 0.00019916443919521513,
|
| 18439 |
+
"loss": 6.2198,
|
| 18440 |
+
"step": 2617
|
| 18441 |
+
},
|
| 18442 |
+
{
|
| 18443 |
+
"epoch": 0.1249477992149002,
|
| 18444 |
+
"grad_norm": 5.463344573974609,
|
| 18445 |
+
"learning_rate": 0.00019916379400339564,
|
| 18446 |
+
"loss": 6.9778,
|
| 18447 |
+
"step": 2618
|
| 18448 |
+
},
|
| 18449 |
+
{
|
| 18450 |
+
"epoch": 0.12499552564699144,
|
| 18451 |
+
"grad_norm": 5.793583393096924,
|
| 18452 |
+
"learning_rate": 0.00019916314856362045,
|
| 18453 |
+
"loss": 6.8516,
|
| 18454 |
+
"step": 2619
|
| 18455 |
+
},
|
| 18456 |
+
{
|
| 18457 |
+
"epoch": 0.1250432520790827,
|
| 18458 |
+
"grad_norm": 5.299160003662109,
|
| 18459 |
+
"learning_rate": 0.00019916250287589117,
|
| 18460 |
+
"loss": 5.9155,
|
| 18461 |
+
"step": 2620
|
| 18462 |
+
},
|
| 18463 |
+
{
|
| 18464 |
+
"epoch": 0.12509097851117396,
|
| 18465 |
+
"grad_norm": 5.868138790130615,
|
| 18466 |
+
"learning_rate": 0.00019916185694020947,
|
| 18467 |
+
"loss": 7.445,
|
| 18468 |
+
"step": 2621
|
| 18469 |
+
},
|
| 18470 |
+
{
|
| 18471 |
+
"epoch": 0.1251387049432652,
|
| 18472 |
+
"grad_norm": 6.175563335418701,
|
| 18473 |
+
"learning_rate": 0.00019916121075657693,
|
| 18474 |
+
"loss": 7.4463,
|
| 18475 |
+
"step": 2622
|
| 18476 |
+
},
|
| 18477 |
+
{
|
| 18478 |
+
"epoch": 0.12518643137535646,
|
| 18479 |
+
"grad_norm": 6.524340629577637,
|
| 18480 |
+
"learning_rate": 0.00019916056432499514,
|
| 18481 |
+
"loss": 5.5733,
|
| 18482 |
+
"step": 2623
|
| 18483 |
+
},
|
| 18484 |
+
{
|
| 18485 |
+
"epoch": 0.1252341578074477,
|
| 18486 |
+
"grad_norm": 5.758206367492676,
|
| 18487 |
+
"learning_rate": 0.0001991599176454658,
|
| 18488 |
+
"loss": 5.9721,
|
| 18489 |
+
"step": 2624
|
| 18490 |
+
},
|
| 18491 |
+
{
|
| 18492 |
+
"epoch": 0.12528188423953895,
|
| 18493 |
+
"grad_norm": 5.740457534790039,
|
| 18494 |
+
"learning_rate": 0.00019915927071799042,
|
| 18495 |
+
"loss": 6.1435,
|
| 18496 |
+
"step": 2625
|
| 18497 |
+
},
|
| 18498 |
+
{
|
| 18499 |
+
"epoch": 0.1253296106716302,
|
| 18500 |
+
"grad_norm": 5.045222759246826,
|
| 18501 |
+
"learning_rate": 0.00019915862354257072,
|
| 18502 |
+
"loss": 6.4285,
|
| 18503 |
+
"step": 2626
|
| 18504 |
+
},
|
| 18505 |
+
{
|
| 18506 |
+
"epoch": 0.12537733710372148,
|
| 18507 |
+
"grad_norm": 6.698198318481445,
|
| 18508 |
+
"learning_rate": 0.00019915797611920824,
|
| 18509 |
+
"loss": 7.0418,
|
| 18510 |
+
"step": 2627
|
| 18511 |
+
},
|
| 18512 |
+
{
|
| 18513 |
+
"epoch": 0.12542506353581273,
|
| 18514 |
+
"grad_norm": 5.3449602127075195,
|
| 18515 |
+
"learning_rate": 0.00019915732844790463,
|
| 18516 |
+
"loss": 5.8405,
|
| 18517 |
+
"step": 2628
|
| 18518 |
+
},
|
| 18519 |
+
{
|
| 18520 |
+
"epoch": 0.12547278996790398,
|
| 18521 |
+
"grad_norm": 8.230816841125488,
|
| 18522 |
+
"learning_rate": 0.0001991566805286615,
|
| 18523 |
+
"loss": 8.8511,
|
| 18524 |
+
"step": 2629
|
| 18525 |
+
},
|
| 18526 |
+
{
|
| 18527 |
+
"epoch": 0.12552051639999523,
|
| 18528 |
+
"grad_norm": 4.497639179229736,
|
| 18529 |
+
"learning_rate": 0.0001991560323614805,
|
| 18530 |
+
"loss": 5.1889,
|
| 18531 |
+
"step": 2630
|
| 18532 |
+
},
|
| 18533 |
+
{
|
| 18534 |
+
"epoch": 0.12556824283208648,
|
| 18535 |
+
"grad_norm": 7.639451503753662,
|
| 18536 |
+
"learning_rate": 0.00019915538394636321,
|
| 18537 |
+
"loss": 6.0825,
|
| 18538 |
+
"step": 2631
|
| 18539 |
+
},
|
| 18540 |
+
{
|
| 18541 |
+
"epoch": 0.12561596926417773,
|
| 18542 |
+
"grad_norm": 6.102507591247559,
|
| 18543 |
+
"learning_rate": 0.00019915473528331129,
|
| 18544 |
+
"loss": 5.6234,
|
| 18545 |
+
"step": 2632
|
| 18546 |
+
},
|
| 18547 |
+
{
|
| 18548 |
+
"epoch": 0.12566369569626898,
|
| 18549 |
+
"grad_norm": 5.559940338134766,
|
| 18550 |
+
"learning_rate": 0.00019915408637232634,
|
| 18551 |
+
"loss": 6.1196,
|
| 18552 |
+
"step": 2633
|
| 18553 |
+
},
|
| 18554 |
+
{
|
| 18555 |
+
"epoch": 0.12571142212836023,
|
| 18556 |
+
"grad_norm": 6.262628078460693,
|
| 18557 |
+
"learning_rate": 0.00019915343721341,
|
| 18558 |
+
"loss": 6.5523,
|
| 18559 |
+
"step": 2634
|
| 18560 |
+
},
|
| 18561 |
+
{
|
| 18562 |
+
"epoch": 0.1257591485604515,
|
| 18563 |
+
"grad_norm": 6.31156587600708,
|
| 18564 |
+
"learning_rate": 0.00019915278780656382,
|
| 18565 |
+
"loss": 6.9212,
|
| 18566 |
+
"step": 2635
|
| 18567 |
+
},
|
| 18568 |
+
{
|
| 18569 |
+
"epoch": 0.12580687499254276,
|
| 18570 |
+
"grad_norm": 5.397120475769043,
|
| 18571 |
+
"learning_rate": 0.0001991521381517895,
|
| 18572 |
+
"loss": 7.2011,
|
| 18573 |
+
"step": 2636
|
| 18574 |
+
},
|
| 18575 |
+
{
|
| 18576 |
+
"epoch": 0.125854601424634,
|
| 18577 |
+
"grad_norm": 5.896693229675293,
|
| 18578 |
+
"learning_rate": 0.00019915148824908868,
|
| 18579 |
+
"loss": 6.2354,
|
| 18580 |
+
"step": 2637
|
| 18581 |
+
},
|
| 18582 |
+
{
|
| 18583 |
+
"epoch": 0.12590232785672525,
|
| 18584 |
+
"grad_norm": 5.77620267868042,
|
| 18585 |
+
"learning_rate": 0.00019915083809846293,
|
| 18586 |
+
"loss": 5.7495,
|
| 18587 |
+
"step": 2638
|
| 18588 |
+
},
|
| 18589 |
+
{
|
| 18590 |
+
"epoch": 0.1259500542888165,
|
| 18591 |
+
"grad_norm": 4.042270660400391,
|
| 18592 |
+
"learning_rate": 0.00019915018769991387,
|
| 18593 |
+
"loss": 5.0322,
|
| 18594 |
+
"step": 2639
|
| 18595 |
+
},
|
| 18596 |
+
{
|
| 18597 |
+
"epoch": 0.12599778072090775,
|
| 18598 |
+
"grad_norm": 7.152609825134277,
|
| 18599 |
+
"learning_rate": 0.0001991495370534432,
|
| 18600 |
+
"loss": 7.4355,
|
| 18601 |
+
"step": 2640
|
| 18602 |
+
},
|
| 18603 |
+
{
|
| 18604 |
+
"epoch": 0.126045507152999,
|
| 18605 |
+
"grad_norm": 5.517868518829346,
|
| 18606 |
+
"learning_rate": 0.00019914888615905248,
|
| 18607 |
+
"loss": 5.7128,
|
| 18608 |
+
"step": 2641
|
| 18609 |
+
},
|
| 18610 |
+
{
|
| 18611 |
+
"epoch": 0.12609323358509025,
|
| 18612 |
+
"grad_norm": 6.580667495727539,
|
| 18613 |
+
"learning_rate": 0.00019914823501674333,
|
| 18614 |
+
"loss": 7.5049,
|
| 18615 |
+
"step": 2642
|
| 18616 |
+
},
|
| 18617 |
+
{
|
| 18618 |
+
"epoch": 0.1261409600171815,
|
| 18619 |
+
"grad_norm": 5.372723579406738,
|
| 18620 |
+
"learning_rate": 0.00019914758362651744,
|
| 18621 |
+
"loss": 6.6719,
|
| 18622 |
+
"step": 2643
|
| 18623 |
+
},
|
| 18624 |
+
{
|
| 18625 |
+
"epoch": 0.12618868644927278,
|
| 18626 |
+
"grad_norm": 6.25331974029541,
|
| 18627 |
+
"learning_rate": 0.0001991469319883764,
|
| 18628 |
+
"loss": 8.3484,
|
| 18629 |
+
"step": 2644
|
| 18630 |
+
},
|
| 18631 |
+
{
|
| 18632 |
+
"epoch": 0.12623641288136403,
|
| 18633 |
+
"grad_norm": 7.487483024597168,
|
| 18634 |
+
"learning_rate": 0.00019914628010232182,
|
| 18635 |
+
"loss": 6.6997,
|
| 18636 |
+
"step": 2645
|
| 18637 |
+
},
|
| 18638 |
+
{
|
| 18639 |
+
"epoch": 0.12628413931345528,
|
| 18640 |
+
"grad_norm": 7.138690948486328,
|
| 18641 |
+
"learning_rate": 0.00019914562796835542,
|
| 18642 |
+
"loss": 6.7894,
|
| 18643 |
+
"step": 2646
|
| 18644 |
+
},
|
| 18645 |
+
{
|
| 18646 |
+
"epoch": 0.12633186574554653,
|
| 18647 |
+
"grad_norm": 4.610174179077148,
|
| 18648 |
+
"learning_rate": 0.0001991449755864787,
|
| 18649 |
+
"loss": 4.4906,
|
| 18650 |
+
"step": 2647
|
| 18651 |
+
},
|
| 18652 |
+
{
|
| 18653 |
+
"epoch": 0.12637959217763778,
|
| 18654 |
+
"grad_norm": 6.854246139526367,
|
| 18655 |
+
"learning_rate": 0.00019914432295669337,
|
| 18656 |
+
"loss": 7.2979,
|
| 18657 |
+
"step": 2648
|
| 18658 |
+
},
|
| 18659 |
+
{
|
| 18660 |
+
"epoch": 0.12642731860972903,
|
| 18661 |
+
"grad_norm": 8.805462837219238,
|
| 18662 |
+
"learning_rate": 0.00019914367007900106,
|
| 18663 |
+
"loss": 7.6631,
|
| 18664 |
+
"step": 2649
|
| 18665 |
+
},
|
| 18666 |
+
{
|
| 18667 |
+
"epoch": 0.12647504504182028,
|
| 18668 |
+
"grad_norm": 8.960370063781738,
|
| 18669 |
+
"learning_rate": 0.00019914301695340338,
|
| 18670 |
+
"loss": 6.965,
|
| 18671 |
+
"step": 2650
|
| 18672 |
+
},
|
| 18673 |
+
{
|
| 18674 |
+
"epoch": 0.12652277147391153,
|
| 18675 |
+
"grad_norm": 6.149599552154541,
|
| 18676 |
+
"learning_rate": 0.00019914236357990197,
|
| 18677 |
+
"loss": 5.7883,
|
| 18678 |
+
"step": 2651
|
| 18679 |
+
},
|
| 18680 |
+
{
|
| 18681 |
+
"epoch": 0.1265704979060028,
|
| 18682 |
+
"grad_norm": 5.623478889465332,
|
| 18683 |
+
"learning_rate": 0.00019914170995849847,
|
| 18684 |
+
"loss": 6.1539,
|
| 18685 |
+
"step": 2652
|
| 18686 |
+
},
|
| 18687 |
+
{
|
| 18688 |
+
"epoch": 0.12661822433809405,
|
| 18689 |
+
"grad_norm": 6.2518792152404785,
|
| 18690 |
+
"learning_rate": 0.00019914105608919452,
|
| 18691 |
+
"loss": 6.0241,
|
| 18692 |
+
"step": 2653
|
| 18693 |
+
},
|
| 18694 |
+
{
|
| 18695 |
+
"epoch": 0.1266659507701853,
|
| 18696 |
+
"grad_norm": 6.36302375793457,
|
| 18697 |
+
"learning_rate": 0.00019914040197199174,
|
| 18698 |
+
"loss": 6.4629,
|
| 18699 |
+
"step": 2654
|
| 18700 |
+
},
|
| 18701 |
+
{
|
| 18702 |
+
"epoch": 0.12671367720227655,
|
| 18703 |
+
"grad_norm": 6.179405212402344,
|
| 18704 |
+
"learning_rate": 0.0001991397476068918,
|
| 18705 |
+
"loss": 6.0301,
|
| 18706 |
+
"step": 2655
|
| 18707 |
+
},
|
| 18708 |
+
{
|
| 18709 |
+
"epoch": 0.1267614036343678,
|
| 18710 |
+
"grad_norm": 6.06155252456665,
|
| 18711 |
+
"learning_rate": 0.00019913909299389623,
|
| 18712 |
+
"loss": 6.2193,
|
| 18713 |
+
"step": 2656
|
| 18714 |
+
},
|
| 18715 |
+
{
|
| 18716 |
+
"epoch": 0.12680913006645905,
|
| 18717 |
+
"grad_norm": 6.631749629974365,
|
| 18718 |
+
"learning_rate": 0.0001991384381330068,
|
| 18719 |
+
"loss": 6.6485,
|
| 18720 |
+
"step": 2657
|
| 18721 |
+
},
|
| 18722 |
+
{
|
| 18723 |
+
"epoch": 0.1268568564985503,
|
| 18724 |
+
"grad_norm": 8.34609603881836,
|
| 18725 |
+
"learning_rate": 0.0001991377830242251,
|
| 18726 |
+
"loss": 6.7854,
|
| 18727 |
+
"step": 2658
|
| 18728 |
+
},
|
| 18729 |
+
{
|
| 18730 |
+
"epoch": 0.12690458293064155,
|
| 18731 |
+
"grad_norm": 6.579798221588135,
|
| 18732 |
+
"learning_rate": 0.00019913712766755275,
|
| 18733 |
+
"loss": 7.3782,
|
| 18734 |
+
"step": 2659
|
| 18735 |
+
},
|
| 18736 |
+
{
|
| 18737 |
+
"epoch": 0.12695230936273283,
|
| 18738 |
+
"grad_norm": 6.286202907562256,
|
| 18739 |
+
"learning_rate": 0.00019913647206299138,
|
| 18740 |
+
"loss": 6.2319,
|
| 18741 |
+
"step": 2660
|
| 18742 |
+
},
|
| 18743 |
+
{
|
| 18744 |
+
"epoch": 0.12700003579482408,
|
| 18745 |
+
"grad_norm": 6.803478240966797,
|
| 18746 |
+
"learning_rate": 0.00019913581621054267,
|
| 18747 |
+
"loss": 6.5881,
|
| 18748 |
+
"step": 2661
|
| 18749 |
+
},
|
| 18750 |
+
{
|
| 18751 |
+
"epoch": 0.12704776222691533,
|
| 18752 |
+
"grad_norm": 5.428170680999756,
|
| 18753 |
+
"learning_rate": 0.00019913516011020824,
|
| 18754 |
+
"loss": 6.6578,
|
| 18755 |
+
"step": 2662
|
| 18756 |
+
},
|
| 18757 |
+
{
|
| 18758 |
+
"epoch": 0.12709548865900658,
|
| 18759 |
+
"grad_norm": 5.8981032371521,
|
| 18760 |
+
"learning_rate": 0.0001991345037619897,
|
| 18761 |
+
"loss": 6.648,
|
| 18762 |
+
"step": 2663
|
| 18763 |
+
},
|
| 18764 |
+
{
|
| 18765 |
+
"epoch": 0.12714321509109783,
|
| 18766 |
+
"grad_norm": 6.136375427246094,
|
| 18767 |
+
"learning_rate": 0.00019913384716588876,
|
| 18768 |
+
"loss": 6.4674,
|
| 18769 |
+
"step": 2664
|
| 18770 |
+
},
|
| 18771 |
+
{
|
| 18772 |
+
"epoch": 0.12719094152318908,
|
| 18773 |
+
"grad_norm": 7.333561420440674,
|
| 18774 |
+
"learning_rate": 0.00019913319032190702,
|
| 18775 |
+
"loss": 7.6672,
|
| 18776 |
+
"step": 2665
|
| 18777 |
+
},
|
| 18778 |
+
{
|
| 18779 |
+
"epoch": 0.12723866795528033,
|
| 18780 |
+
"grad_norm": 6.941676616668701,
|
| 18781 |
+
"learning_rate": 0.00019913253323004609,
|
| 18782 |
+
"loss": 6.5824,
|
| 18783 |
+
"step": 2666
|
| 18784 |
+
},
|
| 18785 |
+
{
|
| 18786 |
+
"epoch": 0.12728639438737158,
|
| 18787 |
+
"grad_norm": 5.5567779541015625,
|
| 18788 |
+
"learning_rate": 0.00019913187589030767,
|
| 18789 |
+
"loss": 6.1721,
|
| 18790 |
+
"step": 2667
|
| 18791 |
+
},
|
| 18792 |
+
{
|
| 18793 |
+
"epoch": 0.12733412081946283,
|
| 18794 |
+
"grad_norm": 7.499778747558594,
|
| 18795 |
+
"learning_rate": 0.00019913121830269335,
|
| 18796 |
+
"loss": 7.4446,
|
| 18797 |
+
"step": 2668
|
| 18798 |
+
},
|
| 18799 |
+
{
|
| 18800 |
+
"epoch": 0.1273818472515541,
|
| 18801 |
+
"grad_norm": 8.435744285583496,
|
| 18802 |
+
"learning_rate": 0.00019913056046720485,
|
| 18803 |
+
"loss": 7.4201,
|
| 18804 |
+
"step": 2669
|
| 18805 |
+
},
|
| 18806 |
+
{
|
| 18807 |
+
"epoch": 0.12742957368364535,
|
| 18808 |
+
"grad_norm": 5.2992095947265625,
|
| 18809 |
+
"learning_rate": 0.0001991299023838437,
|
| 18810 |
+
"loss": 6.645,
|
| 18811 |
+
"step": 2670
|
| 18812 |
+
},
|
| 18813 |
+
{
|
| 18814 |
+
"epoch": 0.1274773001157366,
|
| 18815 |
+
"grad_norm": 5.7956223487854,
|
| 18816 |
+
"learning_rate": 0.0001991292440526117,
|
| 18817 |
+
"loss": 5.3825,
|
| 18818 |
+
"step": 2671
|
| 18819 |
+
},
|
| 18820 |
+
{
|
| 18821 |
+
"epoch": 0.12752502654782785,
|
| 18822 |
+
"grad_norm": 5.5730509757995605,
|
| 18823 |
+
"learning_rate": 0.00019912858547351038,
|
| 18824 |
+
"loss": 5.0827,
|
| 18825 |
+
"step": 2672
|
| 18826 |
+
},
|
| 18827 |
+
{
|
| 18828 |
+
"epoch": 0.1275727529799191,
|
| 18829 |
+
"grad_norm": 6.531050205230713,
|
| 18830 |
+
"learning_rate": 0.00019912792664654143,
|
| 18831 |
+
"loss": 6.4528,
|
| 18832 |
+
"step": 2673
|
| 18833 |
+
},
|
| 18834 |
+
{
|
| 18835 |
+
"epoch": 0.12762047941201035,
|
| 18836 |
+
"grad_norm": 5.897129535675049,
|
| 18837 |
+
"learning_rate": 0.00019912726757170644,
|
| 18838 |
+
"loss": 6.4735,
|
| 18839 |
+
"step": 2674
|
| 18840 |
+
},
|
| 18841 |
+
{
|
| 18842 |
+
"epoch": 0.1276682058441016,
|
| 18843 |
+
"grad_norm": 5.176238536834717,
|
| 18844 |
+
"learning_rate": 0.00019912660824900717,
|
| 18845 |
+
"loss": 5.9754,
|
| 18846 |
+
"step": 2675
|
| 18847 |
+
},
|
| 18848 |
+
{
|
| 18849 |
+
"epoch": 0.12771593227619285,
|
| 18850 |
+
"grad_norm": 5.605959415435791,
|
| 18851 |
+
"learning_rate": 0.00019912594867844516,
|
| 18852 |
+
"loss": 5.5582,
|
| 18853 |
+
"step": 2676
|
| 18854 |
+
},
|
| 18855 |
+
{
|
| 18856 |
+
"epoch": 0.12776365870828413,
|
| 18857 |
+
"grad_norm": 6.690275192260742,
|
| 18858 |
+
"learning_rate": 0.0001991252888600221,
|
| 18859 |
+
"loss": 6.5006,
|
| 18860 |
+
"step": 2677
|
| 18861 |
+
},
|
| 18862 |
+
{
|
| 18863 |
+
"epoch": 0.12781138514037538,
|
| 18864 |
+
"grad_norm": 8.319185256958008,
|
| 18865 |
+
"learning_rate": 0.00019912462879373966,
|
| 18866 |
+
"loss": 6.3769,
|
| 18867 |
+
"step": 2678
|
| 18868 |
+
},
|
| 18869 |
+
{
|
| 18870 |
+
"epoch": 0.12785911157246663,
|
| 18871 |
+
"grad_norm": 5.938503265380859,
|
| 18872 |
+
"learning_rate": 0.00019912396847959947,
|
| 18873 |
+
"loss": 5.9018,
|
| 18874 |
+
"step": 2679
|
| 18875 |
+
},
|
| 18876 |
+
{
|
| 18877 |
+
"epoch": 0.12790683800455788,
|
| 18878 |
+
"grad_norm": 6.656704902648926,
|
| 18879 |
+
"learning_rate": 0.0001991233079176032,
|
| 18880 |
+
"loss": 4.3682,
|
| 18881 |
+
"step": 2680
|
| 18882 |
+
},
|
| 18883 |
+
{
|
| 18884 |
+
"epoch": 0.12795456443664913,
|
| 18885 |
+
"grad_norm": 6.898440361022949,
|
| 18886 |
+
"learning_rate": 0.00019912264710775243,
|
| 18887 |
+
"loss": 7.5726,
|
| 18888 |
+
"step": 2681
|
| 18889 |
+
},
|
| 18890 |
+
{
|
| 18891 |
+
"epoch": 0.12800229086874038,
|
| 18892 |
+
"grad_norm": 5.528750419616699,
|
| 18893 |
+
"learning_rate": 0.0001991219860500489,
|
| 18894 |
+
"loss": 4.7983,
|
| 18895 |
+
"step": 2682
|
| 18896 |
+
},
|
| 18897 |
+
{
|
| 18898 |
+
"epoch": 0.12805001730083163,
|
| 18899 |
+
"grad_norm": 6.4117817878723145,
|
| 18900 |
+
"learning_rate": 0.00019912132474449423,
|
| 18901 |
+
"loss": 6.4503,
|
| 18902 |
+
"step": 2683
|
| 18903 |
+
},
|
| 18904 |
+
{
|
| 18905 |
+
"epoch": 0.12809774373292288,
|
| 18906 |
+
"grad_norm": 6.662499904632568,
|
| 18907 |
+
"learning_rate": 0.00019912066319109008,
|
| 18908 |
+
"loss": 7.6737,
|
| 18909 |
+
"step": 2684
|
| 18910 |
+
},
|
| 18911 |
+
{
|
| 18912 |
+
"epoch": 0.12814547016501415,
|
| 18913 |
+
"grad_norm": 6.276695728302002,
|
| 18914 |
+
"learning_rate": 0.0001991200013898381,
|
| 18915 |
+
"loss": 6.7972,
|
| 18916 |
+
"step": 2685
|
| 18917 |
+
},
|
| 18918 |
+
{
|
| 18919 |
+
"epoch": 0.1281931965971054,
|
| 18920 |
+
"grad_norm": 6.167166709899902,
|
| 18921 |
+
"learning_rate": 0.00019911933934073993,
|
| 18922 |
+
"loss": 6.0611,
|
| 18923 |
+
"step": 2686
|
| 18924 |
+
},
|
| 18925 |
+
{
|
| 18926 |
+
"epoch": 0.12824092302919665,
|
| 18927 |
+
"grad_norm": 6.943721294403076,
|
| 18928 |
+
"learning_rate": 0.00019911867704379722,
|
| 18929 |
+
"loss": 6.0425,
|
| 18930 |
+
"step": 2687
|
| 18931 |
+
},
|
| 18932 |
+
{
|
| 18933 |
+
"epoch": 0.1282886494612879,
|
| 18934 |
+
"grad_norm": 6.147932052612305,
|
| 18935 |
+
"learning_rate": 0.0001991180144990117,
|
| 18936 |
+
"loss": 5.8839,
|
| 18937 |
+
"step": 2688
|
| 18938 |
+
},
|
| 18939 |
+
{
|
| 18940 |
+
"epoch": 0.12833637589337915,
|
| 18941 |
+
"grad_norm": 6.889720439910889,
|
| 18942 |
+
"learning_rate": 0.0001991173517063849,
|
| 18943 |
+
"loss": 5.5848,
|
| 18944 |
+
"step": 2689
|
| 18945 |
+
},
|
| 18946 |
+
{
|
| 18947 |
+
"epoch": 0.1283841023254704,
|
| 18948 |
+
"grad_norm": 5.700843811035156,
|
| 18949 |
+
"learning_rate": 0.00019911668866591858,
|
| 18950 |
+
"loss": 5.2723,
|
| 18951 |
+
"step": 2690
|
| 18952 |
+
},
|
| 18953 |
+
{
|
| 18954 |
+
"epoch": 0.12843182875756165,
|
| 18955 |
+
"grad_norm": 6.69716215133667,
|
| 18956 |
+
"learning_rate": 0.0001991160253776144,
|
| 18957 |
+
"loss": 6.8941,
|
| 18958 |
+
"step": 2691
|
| 18959 |
+
},
|
| 18960 |
+
{
|
| 18961 |
+
"epoch": 0.1284795551896529,
|
| 18962 |
+
"grad_norm": 9.107110977172852,
|
| 18963 |
+
"learning_rate": 0.00019911536184147397,
|
| 18964 |
+
"loss": 8.4972,
|
| 18965 |
+
"step": 2692
|
| 18966 |
+
},
|
| 18967 |
+
{
|
| 18968 |
+
"epoch": 0.12852728162174415,
|
| 18969 |
+
"grad_norm": 6.4476776123046875,
|
| 18970 |
+
"learning_rate": 0.00019911469805749895,
|
| 18971 |
+
"loss": 7.1879,
|
| 18972 |
+
"step": 2693
|
| 18973 |
+
},
|
| 18974 |
+
{
|
| 18975 |
+
"epoch": 0.12857500805383543,
|
| 18976 |
+
"grad_norm": 6.365779399871826,
|
| 18977 |
+
"learning_rate": 0.000199114034025691,
|
| 18978 |
+
"loss": 5.6422,
|
| 18979 |
+
"step": 2694
|
| 18980 |
+
},
|
| 18981 |
+
{
|
| 18982 |
+
"epoch": 0.12862273448592668,
|
| 18983 |
+
"grad_norm": 5.086087703704834,
|
| 18984 |
+
"learning_rate": 0.00019911336974605179,
|
| 18985 |
+
"loss": 4.4789,
|
| 18986 |
+
"step": 2695
|
| 18987 |
+
},
|
| 18988 |
+
{
|
| 18989 |
+
"epoch": 0.12867046091801793,
|
| 18990 |
+
"grad_norm": 4.119086265563965,
|
| 18991 |
+
"learning_rate": 0.000199112705218583,
|
| 18992 |
+
"loss": 4.5278,
|
| 18993 |
+
"step": 2696
|
| 18994 |
+
},
|
| 18995 |
+
{
|
| 18996 |
+
"epoch": 0.12871818735010918,
|
| 18997 |
+
"grad_norm": 5.206000804901123,
|
| 18998 |
+
"learning_rate": 0.0001991120404432863,
|
| 18999 |
+
"loss": 5.2369,
|
| 19000 |
+
"step": 2697
|
| 19001 |
+
},
|
| 19002 |
+
{
|
| 19003 |
+
"epoch": 0.12876591378220043,
|
| 19004 |
+
"grad_norm": 4.440513610839844,
|
| 19005 |
+
"learning_rate": 0.00019911137542016333,
|
| 19006 |
+
"loss": 5.9666,
|
| 19007 |
+
"step": 2698
|
| 19008 |
+
},
|
| 19009 |
+
{
|
| 19010 |
+
"epoch": 0.12881364021429167,
|
| 19011 |
+
"grad_norm": 8.883622169494629,
|
| 19012 |
+
"learning_rate": 0.0001991107101492157,
|
| 19013 |
+
"loss": 7.3967,
|
| 19014 |
+
"step": 2699
|
| 19015 |
+
},
|
| 19016 |
+
{
|
| 19017 |
+
"epoch": 0.12886136664638292,
|
| 19018 |
+
"grad_norm": 6.536487579345703,
|
| 19019 |
+
"learning_rate": 0.00019911004463044516,
|
| 19020 |
+
"loss": 6.1909,
|
| 19021 |
+
"step": 2700
|
| 19022 |
+
},
|
| 19023 |
+
{
|
| 19024 |
+
"epoch": 0.12890909307847417,
|
| 19025 |
+
"grad_norm": 4.277031898498535,
|
| 19026 |
+
"learning_rate": 0.00019910937886385334,
|
| 19027 |
+
"loss": 4.1263,
|
| 19028 |
+
"step": 2701
|
| 19029 |
+
},
|
| 19030 |
+
{
|
| 19031 |
+
"epoch": 0.12895681951056545,
|
| 19032 |
+
"grad_norm": 5.570382118225098,
|
| 19033 |
+
"learning_rate": 0.00019910871284944192,
|
| 19034 |
+
"loss": 6.8265,
|
| 19035 |
+
"step": 2702
|
| 19036 |
+
},
|
| 19037 |
+
{
|
| 19038 |
+
"epoch": 0.1290045459426567,
|
| 19039 |
+
"grad_norm": 4.961894989013672,
|
| 19040 |
+
"learning_rate": 0.0001991080465872125,
|
| 19041 |
+
"loss": 4.9882,
|
| 19042 |
+
"step": 2703
|
| 19043 |
+
},
|
| 19044 |
+
{
|
| 19045 |
+
"epoch": 0.12905227237474795,
|
| 19046 |
+
"grad_norm": 5.48018741607666,
|
| 19047 |
+
"learning_rate": 0.00019910738007716683,
|
| 19048 |
+
"loss": 6.3094,
|
| 19049 |
+
"step": 2704
|
| 19050 |
+
},
|
| 19051 |
+
{
|
| 19052 |
+
"epoch": 0.1290999988068392,
|
| 19053 |
+
"grad_norm": 5.8967671394348145,
|
| 19054 |
+
"learning_rate": 0.00019910671331930654,
|
| 19055 |
+
"loss": 7.0458,
|
| 19056 |
+
"step": 2705
|
| 19057 |
+
},
|
| 19058 |
+
{
|
| 19059 |
+
"epoch": 0.12914772523893045,
|
| 19060 |
+
"grad_norm": 7.660339832305908,
|
| 19061 |
+
"learning_rate": 0.00019910604631363327,
|
| 19062 |
+
"loss": 7.6331,
|
| 19063 |
+
"step": 2706
|
| 19064 |
+
},
|
| 19065 |
+
{
|
| 19066 |
+
"epoch": 0.1291954516710217,
|
| 19067 |
+
"grad_norm": 5.007476329803467,
|
| 19068 |
+
"learning_rate": 0.00019910537906014873,
|
| 19069 |
+
"loss": 6.0415,
|
| 19070 |
+
"step": 2707
|
| 19071 |
+
},
|
| 19072 |
+
{
|
| 19073 |
+
"epoch": 0.12924317810311295,
|
| 19074 |
+
"grad_norm": 7.231809616088867,
|
| 19075 |
+
"learning_rate": 0.00019910471155885459,
|
| 19076 |
+
"loss": 6.9033,
|
| 19077 |
+
"step": 2708
|
| 19078 |
+
},
|
| 19079 |
+
{
|
| 19080 |
+
"epoch": 0.1292909045352042,
|
| 19081 |
+
"grad_norm": 7.8866753578186035,
|
| 19082 |
+
"learning_rate": 0.0001991040438097525,
|
| 19083 |
+
"loss": 8.1581,
|
| 19084 |
+
"step": 2709
|
| 19085 |
+
},
|
| 19086 |
+
{
|
| 19087 |
+
"epoch": 0.12933863096729548,
|
| 19088 |
+
"grad_norm": 4.540760517120361,
|
| 19089 |
+
"learning_rate": 0.00019910337581284415,
|
| 19090 |
+
"loss": 3.9487,
|
| 19091 |
+
"step": 2710
|
| 19092 |
+
},
|
| 19093 |
+
{
|
| 19094 |
+
"epoch": 0.12938635739938673,
|
| 19095 |
+
"grad_norm": 6.535691261291504,
|
| 19096 |
+
"learning_rate": 0.0001991027075681312,
|
| 19097 |
+
"loss": 6.8769,
|
| 19098 |
+
"step": 2711
|
| 19099 |
+
},
|
| 19100 |
+
{
|
| 19101 |
+
"epoch": 0.12943408383147798,
|
| 19102 |
+
"grad_norm": 7.321287631988525,
|
| 19103 |
+
"learning_rate": 0.00019910203907561527,
|
| 19104 |
+
"loss": 7.6307,
|
| 19105 |
+
"step": 2712
|
| 19106 |
+
},
|
| 19107 |
+
{
|
| 19108 |
+
"epoch": 0.12948181026356922,
|
| 19109 |
+
"grad_norm": 8.327848434448242,
|
| 19110 |
+
"learning_rate": 0.0001991013703352981,
|
| 19111 |
+
"loss": 6.3343,
|
| 19112 |
+
"step": 2713
|
| 19113 |
+
},
|
| 19114 |
+
{
|
| 19115 |
+
"epoch": 0.12952953669566047,
|
| 19116 |
+
"grad_norm": 4.456166744232178,
|
| 19117 |
+
"learning_rate": 0.00019910070134718133,
|
| 19118 |
+
"loss": 4.3715,
|
| 19119 |
+
"step": 2714
|
| 19120 |
+
},
|
| 19121 |
+
{
|
| 19122 |
+
"epoch": 0.12957726312775172,
|
| 19123 |
+
"grad_norm": 7.4280781745910645,
|
| 19124 |
+
"learning_rate": 0.00019910003211126668,
|
| 19125 |
+
"loss": 7.9487,
|
| 19126 |
+
"step": 2715
|
| 19127 |
+
},
|
| 19128 |
+
{
|
| 19129 |
+
"epoch": 0.12962498955984297,
|
| 19130 |
+
"grad_norm": 6.105531215667725,
|
| 19131 |
+
"learning_rate": 0.00019909936262755572,
|
| 19132 |
+
"loss": 5.3206,
|
| 19133 |
+
"step": 2716
|
| 19134 |
+
},
|
| 19135 |
+
{
|
| 19136 |
+
"epoch": 0.12967271599193422,
|
| 19137 |
+
"grad_norm": 6.178251266479492,
|
| 19138 |
+
"learning_rate": 0.00019909869289605023,
|
| 19139 |
+
"loss": 7.429,
|
| 19140 |
+
"step": 2717
|
| 19141 |
+
},
|
| 19142 |
+
{
|
| 19143 |
+
"epoch": 0.12972044242402547,
|
| 19144 |
+
"grad_norm": 7.50347375869751,
|
| 19145 |
+
"learning_rate": 0.00019909802291675187,
|
| 19146 |
+
"loss": 6.884,
|
| 19147 |
+
"step": 2718
|
| 19148 |
+
},
|
| 19149 |
+
{
|
| 19150 |
+
"epoch": 0.12976816885611675,
|
| 19151 |
+
"grad_norm": 8.008402824401855,
|
| 19152 |
+
"learning_rate": 0.00019909735268966222,
|
| 19153 |
+
"loss": 6.4672,
|
| 19154 |
+
"step": 2719
|
| 19155 |
+
},
|
| 19156 |
+
{
|
| 19157 |
+
"epoch": 0.129815895288208,
|
| 19158 |
+
"grad_norm": 6.246631622314453,
|
| 19159 |
+
"learning_rate": 0.0001990966822147831,
|
| 19160 |
+
"loss": 5.7752,
|
| 19161 |
+
"step": 2720
|
| 19162 |
+
},
|
| 19163 |
+
{
|
| 19164 |
+
"epoch": 0.12986362172029925,
|
| 19165 |
+
"grad_norm": 5.764650821685791,
|
| 19166 |
+
"learning_rate": 0.00019909601149211605,
|
| 19167 |
+
"loss": 6.088,
|
| 19168 |
+
"step": 2721
|
| 19169 |
+
},
|
| 19170 |
+
{
|
| 19171 |
+
"epoch": 0.1299113481523905,
|
| 19172 |
+
"grad_norm": 6.757874488830566,
|
| 19173 |
+
"learning_rate": 0.00019909534052166286,
|
| 19174 |
+
"loss": 6.4215,
|
| 19175 |
+
"step": 2722
|
| 19176 |
+
},
|
| 19177 |
+
{
|
| 19178 |
+
"epoch": 0.12995907458448175,
|
| 19179 |
+
"grad_norm": 7.595560073852539,
|
| 19180 |
+
"learning_rate": 0.0001990946693034251,
|
| 19181 |
+
"loss": 7.4822,
|
| 19182 |
+
"step": 2723
|
| 19183 |
+
},
|
| 19184 |
+
{
|
| 19185 |
+
"epoch": 0.130006801016573,
|
| 19186 |
+
"grad_norm": 6.582478046417236,
|
| 19187 |
+
"learning_rate": 0.00019909399783740454,
|
| 19188 |
+
"loss": 6.6361,
|
| 19189 |
+
"step": 2724
|
| 19190 |
+
},
|
| 19191 |
+
{
|
| 19192 |
+
"epoch": 0.13005452744866425,
|
| 19193 |
+
"grad_norm": 5.431997299194336,
|
| 19194 |
+
"learning_rate": 0.00019909332612360285,
|
| 19195 |
+
"loss": 6.3359,
|
| 19196 |
+
"step": 2725
|
| 19197 |
+
},
|
| 19198 |
+
{
|
| 19199 |
+
"epoch": 0.1301022538807555,
|
| 19200 |
+
"grad_norm": 5.795412540435791,
|
| 19201 |
+
"learning_rate": 0.00019909265416202164,
|
| 19202 |
+
"loss": 7.2463,
|
| 19203 |
+
"step": 2726
|
| 19204 |
+
},
|
| 19205 |
+
{
|
| 19206 |
+
"epoch": 0.13014998031284677,
|
| 19207 |
+
"grad_norm": 6.594300270080566,
|
| 19208 |
+
"learning_rate": 0.00019909198195266266,
|
| 19209 |
+
"loss": 5.7899,
|
| 19210 |
+
"step": 2727
|
| 19211 |
+
},
|
| 19212 |
+
{
|
| 19213 |
+
"epoch": 0.13019770674493802,
|
| 19214 |
+
"grad_norm": 6.408435344696045,
|
| 19215 |
+
"learning_rate": 0.00019909130949552755,
|
| 19216 |
+
"loss": 7.3634,
|
| 19217 |
+
"step": 2728
|
| 19218 |
+
},
|
| 19219 |
+
{
|
| 19220 |
+
"epoch": 0.13024543317702927,
|
| 19221 |
+
"grad_norm": 6.5589919090271,
|
| 19222 |
+
"learning_rate": 0.000199090636790618,
|
| 19223 |
+
"loss": 6.6704,
|
| 19224 |
+
"step": 2729
|
| 19225 |
+
},
|
| 19226 |
+
{
|
| 19227 |
+
"epoch": 0.13029315960912052,
|
| 19228 |
+
"grad_norm": 6.496981620788574,
|
| 19229 |
+
"learning_rate": 0.0001990899638379357,
|
| 19230 |
+
"loss": 6.1406,
|
| 19231 |
+
"step": 2730
|
| 19232 |
+
},
|
| 19233 |
+
{
|
| 19234 |
+
"epoch": 0.13034088604121177,
|
| 19235 |
+
"grad_norm": 4.503755569458008,
|
| 19236 |
+
"learning_rate": 0.00019908929063748235,
|
| 19237 |
+
"loss": 5.0579,
|
| 19238 |
+
"step": 2731
|
| 19239 |
+
},
|
| 19240 |
+
{
|
| 19241 |
+
"epoch": 0.13038861247330302,
|
| 19242 |
+
"grad_norm": 7.117122173309326,
|
| 19243 |
+
"learning_rate": 0.0001990886171892596,
|
| 19244 |
+
"loss": 7.4509,
|
| 19245 |
+
"step": 2732
|
| 19246 |
+
},
|
| 19247 |
+
{
|
| 19248 |
+
"epoch": 0.13043633890539427,
|
| 19249 |
+
"grad_norm": 6.266332149505615,
|
| 19250 |
+
"learning_rate": 0.00019908794349326913,
|
| 19251 |
+
"loss": 6.4409,
|
| 19252 |
+
"step": 2733
|
| 19253 |
+
},
|
| 19254 |
+
{
|
| 19255 |
+
"epoch": 0.13048406533748552,
|
| 19256 |
+
"grad_norm": 6.61918306350708,
|
| 19257 |
+
"learning_rate": 0.00019908726954951267,
|
| 19258 |
+
"loss": 7.2602,
|
| 19259 |
+
"step": 2734
|
| 19260 |
+
},
|
| 19261 |
+
{
|
| 19262 |
+
"epoch": 0.1305317917695768,
|
| 19263 |
+
"grad_norm": 6.399169445037842,
|
| 19264 |
+
"learning_rate": 0.00019908659535799187,
|
| 19265 |
+
"loss": 7.1302,
|
| 19266 |
+
"step": 2735
|
| 19267 |
+
},
|
| 19268 |
+
{
|
| 19269 |
+
"epoch": 0.13057951820166805,
|
| 19270 |
+
"grad_norm": 7.0500898361206055,
|
| 19271 |
+
"learning_rate": 0.00019908592091870842,
|
| 19272 |
+
"loss": 7.6507,
|
| 19273 |
+
"step": 2736
|
| 19274 |
+
},
|
| 19275 |
+
{
|
| 19276 |
+
"epoch": 0.1306272446337593,
|
| 19277 |
+
"grad_norm": 5.893949508666992,
|
| 19278 |
+
"learning_rate": 0.00019908524623166401,
|
| 19279 |
+
"loss": 6.7635,
|
| 19280 |
+
"step": 2737
|
| 19281 |
+
},
|
| 19282 |
+
{
|
| 19283 |
+
"epoch": 0.13067497106585055,
|
| 19284 |
+
"grad_norm": 6.369676113128662,
|
| 19285 |
+
"learning_rate": 0.0001990845712968603,
|
| 19286 |
+
"loss": 7.2699,
|
| 19287 |
+
"step": 2738
|
| 19288 |
+
},
|
| 19289 |
+
{
|
| 19290 |
+
"epoch": 0.1307226974979418,
|
| 19291 |
+
"grad_norm": 6.691352844238281,
|
| 19292 |
+
"learning_rate": 0.00019908389611429905,
|
| 19293 |
+
"loss": 7.8422,
|
| 19294 |
+
"step": 2739
|
| 19295 |
+
},
|
| 19296 |
+
{
|
| 19297 |
+
"epoch": 0.13077042393003305,
|
| 19298 |
+
"grad_norm": 6.010880470275879,
|
| 19299 |
+
"learning_rate": 0.00019908322068398188,
|
| 19300 |
+
"loss": 7.8595,
|
| 19301 |
+
"step": 2740
|
| 19302 |
+
},
|
| 19303 |
+
{
|
| 19304 |
+
"epoch": 0.1308181503621243,
|
| 19305 |
+
"grad_norm": 6.41002893447876,
|
| 19306 |
+
"learning_rate": 0.0001990825450059105,
|
| 19307 |
+
"loss": 6.4199,
|
| 19308 |
+
"step": 2741
|
| 19309 |
+
},
|
| 19310 |
+
{
|
| 19311 |
+
"epoch": 0.13086587679421555,
|
| 19312 |
+
"grad_norm": 6.322947025299072,
|
| 19313 |
+
"learning_rate": 0.0001990818690800866,
|
| 19314 |
+
"loss": 5.793,
|
| 19315 |
+
"step": 2742
|
| 19316 |
+
},
|
| 19317 |
+
{
|
| 19318 |
+
"epoch": 0.1309136032263068,
|
| 19319 |
+
"grad_norm": 6.068185329437256,
|
| 19320 |
+
"learning_rate": 0.0001990811929065119,
|
| 19321 |
+
"loss": 5.8207,
|
| 19322 |
+
"step": 2743
|
| 19323 |
+
},
|
| 19324 |
+
{
|
| 19325 |
+
"epoch": 0.13096132965839807,
|
| 19326 |
+
"grad_norm": 4.704338073730469,
|
| 19327 |
+
"learning_rate": 0.00019908051648518803,
|
| 19328 |
+
"loss": 5.3277,
|
| 19329 |
+
"step": 2744
|
| 19330 |
+
},
|
| 19331 |
+
{
|
| 19332 |
+
"epoch": 0.13100905609048932,
|
| 19333 |
+
"grad_norm": 5.012578964233398,
|
| 19334 |
+
"learning_rate": 0.00019907983981611672,
|
| 19335 |
+
"loss": 5.0888,
|
| 19336 |
+
"step": 2745
|
| 19337 |
+
},
|
| 19338 |
+
{
|
| 19339 |
+
"epoch": 0.13105678252258057,
|
| 19340 |
+
"grad_norm": 6.814549922943115,
|
| 19341 |
+
"learning_rate": 0.00019907916289929966,
|
| 19342 |
+
"loss": 7.5414,
|
| 19343 |
+
"step": 2746
|
| 19344 |
+
},
|
| 19345 |
+
{
|
| 19346 |
+
"epoch": 0.13110450895467182,
|
| 19347 |
+
"grad_norm": 5.401822090148926,
|
| 19348 |
+
"learning_rate": 0.00019907848573473853,
|
| 19349 |
+
"loss": 5.8487,
|
| 19350 |
+
"step": 2747
|
| 19351 |
+
},
|
| 19352 |
+
{
|
| 19353 |
+
"epoch": 0.13115223538676307,
|
| 19354 |
+
"grad_norm": 6.063152313232422,
|
| 19355 |
+
"learning_rate": 0.00019907780832243506,
|
| 19356 |
+
"loss": 7.1534,
|
| 19357 |
+
"step": 2748
|
| 19358 |
+
},
|
| 19359 |
+
{
|
| 19360 |
+
"epoch": 0.13119996181885432,
|
| 19361 |
+
"grad_norm": 4.4824395179748535,
|
| 19362 |
+
"learning_rate": 0.00019907713066239088,
|
| 19363 |
+
"loss": 4.1908,
|
| 19364 |
+
"step": 2749
|
| 19365 |
+
},
|
| 19366 |
+
{
|
| 19367 |
+
"epoch": 0.13124768825094557,
|
| 19368 |
+
"grad_norm": 7.166471004486084,
|
| 19369 |
+
"learning_rate": 0.00019907645275460775,
|
| 19370 |
+
"loss": 7.4206,
|
| 19371 |
+
"step": 2750
|
| 19372 |
+
},
|
| 19373 |
+
{
|
| 19374 |
+
"epoch": 0.13129541468303682,
|
| 19375 |
+
"grad_norm": 8.25527286529541,
|
| 19376 |
+
"learning_rate": 0.0001990757745990873,
|
| 19377 |
+
"loss": 8.5766,
|
| 19378 |
+
"step": 2751
|
| 19379 |
+
},
|
| 19380 |
+
{
|
| 19381 |
+
"epoch": 0.1313431411151281,
|
| 19382 |
+
"grad_norm": 5.488870143890381,
|
| 19383 |
+
"learning_rate": 0.0001990750961958313,
|
| 19384 |
+
"loss": 5.4217,
|
| 19385 |
+
"step": 2752
|
| 19386 |
+
},
|
| 19387 |
+
{
|
| 19388 |
+
"epoch": 0.13139086754721935,
|
| 19389 |
+
"grad_norm": 7.472160339355469,
|
| 19390 |
+
"learning_rate": 0.00019907441754484138,
|
| 19391 |
+
"loss": 7.9112,
|
| 19392 |
+
"step": 2753
|
| 19393 |
+
},
|
| 19394 |
+
{
|
| 19395 |
+
"epoch": 0.1314385939793106,
|
| 19396 |
+
"grad_norm": 6.098803997039795,
|
| 19397 |
+
"learning_rate": 0.00019907373864611925,
|
| 19398 |
+
"loss": 5.8821,
|
| 19399 |
+
"step": 2754
|
| 19400 |
+
},
|
| 19401 |
+
{
|
| 19402 |
+
"epoch": 0.13148632041140185,
|
| 19403 |
+
"grad_norm": 5.743438243865967,
|
| 19404 |
+
"learning_rate": 0.00019907305949966664,
|
| 19405 |
+
"loss": 6.1745,
|
| 19406 |
+
"step": 2755
|
| 19407 |
+
},
|
| 19408 |
+
{
|
| 19409 |
+
"epoch": 0.1315340468434931,
|
| 19410 |
+
"grad_norm": 5.455843448638916,
|
| 19411 |
+
"learning_rate": 0.00019907238010548523,
|
| 19412 |
+
"loss": 5.7785,
|
| 19413 |
+
"step": 2756
|
| 19414 |
+
},
|
| 19415 |
+
{
|
| 19416 |
+
"epoch": 0.13158177327558435,
|
| 19417 |
+
"grad_norm": 5.50465726852417,
|
| 19418 |
+
"learning_rate": 0.00019907170046357673,
|
| 19419 |
+
"loss": 5.6001,
|
| 19420 |
+
"step": 2757
|
| 19421 |
+
},
|
| 19422 |
+
{
|
| 19423 |
+
"epoch": 0.1316294997076756,
|
| 19424 |
+
"grad_norm": 6.299639701843262,
|
| 19425 |
+
"learning_rate": 0.0001990710205739428,
|
| 19426 |
+
"loss": 8.1939,
|
| 19427 |
+
"step": 2758
|
| 19428 |
+
},
|
| 19429 |
+
{
|
| 19430 |
+
"epoch": 0.13167722613976685,
|
| 19431 |
+
"grad_norm": 6.706847667694092,
|
| 19432 |
+
"learning_rate": 0.0001990703404365852,
|
| 19433 |
+
"loss": 6.7525,
|
| 19434 |
+
"step": 2759
|
| 19435 |
+
},
|
| 19436 |
+
{
|
| 19437 |
+
"epoch": 0.13172495257185812,
|
| 19438 |
+
"grad_norm": 8.025195121765137,
|
| 19439 |
+
"learning_rate": 0.00019906966005150557,
|
| 19440 |
+
"loss": 7.904,
|
| 19441 |
+
"step": 2760
|
| 19442 |
+
},
|
| 19443 |
+
{
|
| 19444 |
+
"epoch": 0.13177267900394937,
|
| 19445 |
+
"grad_norm": 9.603436470031738,
|
| 19446 |
+
"learning_rate": 0.00019906897941870564,
|
| 19447 |
+
"loss": 6.8963,
|
| 19448 |
+
"step": 2761
|
| 19449 |
+
},
|
| 19450 |
+
{
|
| 19451 |
+
"epoch": 0.13182040543604062,
|
| 19452 |
+
"grad_norm": 6.5206217765808105,
|
| 19453 |
+
"learning_rate": 0.00019906829853818714,
|
| 19454 |
+
"loss": 6.6906,
|
| 19455 |
+
"step": 2762
|
| 19456 |
+
},
|
| 19457 |
+
{
|
| 19458 |
+
"epoch": 0.13186813186813187,
|
| 19459 |
+
"grad_norm": 5.41235876083374,
|
| 19460 |
+
"learning_rate": 0.00019906761740995173,
|
| 19461 |
+
"loss": 6.1441,
|
| 19462 |
+
"step": 2763
|
| 19463 |
+
},
|
| 19464 |
+
{
|
| 19465 |
+
"epoch": 0.13191585830022312,
|
| 19466 |
+
"grad_norm": 5.746955871582031,
|
| 19467 |
+
"learning_rate": 0.00019906693603400112,
|
| 19468 |
+
"loss": 5.5281,
|
| 19469 |
+
"step": 2764
|
| 19470 |
+
},
|
| 19471 |
+
{
|
| 19472 |
+
"epoch": 0.13196358473231437,
|
| 19473 |
+
"grad_norm": 6.709540843963623,
|
| 19474 |
+
"learning_rate": 0.000199066254410337,
|
| 19475 |
+
"loss": 6.3966,
|
| 19476 |
+
"step": 2765
|
| 19477 |
+
},
|
| 19478 |
+
{
|
| 19479 |
+
"epoch": 0.13201131116440562,
|
| 19480 |
+
"grad_norm": 5.8970417976379395,
|
| 19481 |
+
"learning_rate": 0.00019906557253896115,
|
| 19482 |
+
"loss": 6.9662,
|
| 19483 |
+
"step": 2766
|
| 19484 |
+
},
|
| 19485 |
+
{
|
| 19486 |
+
"epoch": 0.13205903759649687,
|
| 19487 |
+
"grad_norm": 5.076374053955078,
|
| 19488 |
+
"learning_rate": 0.00019906489041987517,
|
| 19489 |
+
"loss": 5.1472,
|
| 19490 |
+
"step": 2767
|
| 19491 |
+
},
|
| 19492 |
+
{
|
| 19493 |
+
"epoch": 0.13210676402858812,
|
| 19494 |
+
"grad_norm": 6.5956902503967285,
|
| 19495 |
+
"learning_rate": 0.00019906420805308083,
|
| 19496 |
+
"loss": 7.3836,
|
| 19497 |
+
"step": 2768
|
| 19498 |
+
},
|
| 19499 |
+
{
|
| 19500 |
+
"epoch": 0.1321544904606794,
|
| 19501 |
+
"grad_norm": 5.3043928146362305,
|
| 19502 |
+
"learning_rate": 0.00019906352543857983,
|
| 19503 |
+
"loss": 6.8485,
|
| 19504 |
+
"step": 2769
|
| 19505 |
+
},
|
| 19506 |
+
{
|
| 19507 |
+
"epoch": 0.13220221689277065,
|
| 19508 |
+
"grad_norm": 7.224258899688721,
|
| 19509 |
+
"learning_rate": 0.00019906284257637386,
|
| 19510 |
+
"loss": 6.2999,
|
| 19511 |
+
"step": 2770
|
| 19512 |
+
},
|
| 19513 |
+
{
|
| 19514 |
+
"epoch": 0.1322499433248619,
|
| 19515 |
+
"grad_norm": 4.787478923797607,
|
| 19516 |
+
"learning_rate": 0.00019906215946646465,
|
| 19517 |
+
"loss": 4.6901,
|
| 19518 |
+
"step": 2771
|
| 19519 |
+
},
|
| 19520 |
+
{
|
| 19521 |
+
"epoch": 0.13229766975695315,
|
| 19522 |
+
"grad_norm": 5.938333511352539,
|
| 19523 |
+
"learning_rate": 0.0001990614761088539,
|
| 19524 |
+
"loss": 7.4666,
|
| 19525 |
+
"step": 2772
|
| 19526 |
+
},
|
| 19527 |
+
{
|
| 19528 |
+
"epoch": 0.1323453961890444,
|
| 19529 |
+
"grad_norm": 6.651096820831299,
|
| 19530 |
+
"learning_rate": 0.0001990607925035433,
|
| 19531 |
+
"loss": 7.231,
|
| 19532 |
+
"step": 2773
|
| 19533 |
+
},
|
| 19534 |
+
{
|
| 19535 |
+
"epoch": 0.13239312262113564,
|
| 19536 |
+
"grad_norm": 4.987978935241699,
|
| 19537 |
+
"learning_rate": 0.00019906010865053454,
|
| 19538 |
+
"loss": 5.4362,
|
| 19539 |
+
"step": 2774
|
| 19540 |
+
},
|
| 19541 |
+
{
|
| 19542 |
+
"epoch": 0.1324408490532269,
|
| 19543 |
+
"grad_norm": 6.788287162780762,
|
| 19544 |
+
"learning_rate": 0.0001990594245498294,
|
| 19545 |
+
"loss": 7.0538,
|
| 19546 |
+
"step": 2775
|
| 19547 |
+
},
|
| 19548 |
+
{
|
| 19549 |
+
"epoch": 0.13248857548531814,
|
| 19550 |
+
"grad_norm": 5.706051826477051,
|
| 19551 |
+
"learning_rate": 0.00019905874020142953,
|
| 19552 |
+
"loss": 6.9948,
|
| 19553 |
+
"step": 2776
|
| 19554 |
+
},
|
| 19555 |
+
{
|
| 19556 |
+
"epoch": 0.13253630191740942,
|
| 19557 |
+
"grad_norm": 7.513382434844971,
|
| 19558 |
+
"learning_rate": 0.0001990580556053367,
|
| 19559 |
+
"loss": 6.8821,
|
| 19560 |
+
"step": 2777
|
| 19561 |
+
},
|
| 19562 |
+
{
|
| 19563 |
+
"epoch": 0.13258402834950067,
|
| 19564 |
+
"grad_norm": 7.37601375579834,
|
| 19565 |
+
"learning_rate": 0.00019905737076155258,
|
| 19566 |
+
"loss": 8.9933,
|
| 19567 |
+
"step": 2778
|
| 19568 |
+
},
|
| 19569 |
+
{
|
| 19570 |
+
"epoch": 0.13263175478159192,
|
| 19571 |
+
"grad_norm": 4.514930248260498,
|
| 19572 |
+
"learning_rate": 0.00019905668567007887,
|
| 19573 |
+
"loss": 4.4424,
|
| 19574 |
+
"step": 2779
|
| 19575 |
+
},
|
| 19576 |
+
{
|
| 19577 |
+
"epoch": 0.13267948121368317,
|
| 19578 |
+
"grad_norm": 6.982491970062256,
|
| 19579 |
+
"learning_rate": 0.0001990560003309173,
|
| 19580 |
+
"loss": 7.462,
|
| 19581 |
+
"step": 2780
|
| 19582 |
+
},
|
| 19583 |
+
{
|
| 19584 |
+
"epoch": 0.13272720764577442,
|
| 19585 |
+
"grad_norm": 7.289805889129639,
|
| 19586 |
+
"learning_rate": 0.0001990553147440696,
|
| 19587 |
+
"loss": 6.7003,
|
| 19588 |
+
"step": 2781
|
| 19589 |
+
},
|
| 19590 |
+
{
|
| 19591 |
+
"epoch": 0.13277493407786567,
|
| 19592 |
+
"grad_norm": 5.799755573272705,
|
| 19593 |
+
"learning_rate": 0.00019905462890953745,
|
| 19594 |
+
"loss": 6.4667,
|
| 19595 |
+
"step": 2782
|
| 19596 |
+
},
|
| 19597 |
+
{
|
| 19598 |
+
"epoch": 0.13282266050995692,
|
| 19599 |
+
"grad_norm": 5.055026531219482,
|
| 19600 |
+
"learning_rate": 0.0001990539428273226,
|
| 19601 |
+
"loss": 5.4229,
|
| 19602 |
+
"step": 2783
|
| 19603 |
+
},
|
| 19604 |
+
{
|
| 19605 |
+
"epoch": 0.13287038694204817,
|
| 19606 |
+
"grad_norm": 4.6647844314575195,
|
| 19607 |
+
"learning_rate": 0.00019905325649742674,
|
| 19608 |
+
"loss": 4.7695,
|
| 19609 |
+
"step": 2784
|
| 19610 |
+
},
|
| 19611 |
+
{
|
| 19612 |
+
"epoch": 0.13291811337413945,
|
| 19613 |
+
"grad_norm": 7.971372604370117,
|
| 19614 |
+
"learning_rate": 0.00019905256991985162,
|
| 19615 |
+
"loss": 6.1439,
|
| 19616 |
+
"step": 2785
|
| 19617 |
+
},
|
| 19618 |
+
{
|
| 19619 |
+
"epoch": 0.1329658398062307,
|
| 19620 |
+
"grad_norm": 6.43125581741333,
|
| 19621 |
+
"learning_rate": 0.00019905188309459891,
|
| 19622 |
+
"loss": 6.2026,
|
| 19623 |
+
"step": 2786
|
| 19624 |
+
},
|
| 19625 |
+
{
|
| 19626 |
+
"epoch": 0.13301356623832195,
|
| 19627 |
+
"grad_norm": 4.872565269470215,
|
| 19628 |
+
"learning_rate": 0.00019905119602167036,
|
| 19629 |
+
"loss": 4.4752,
|
| 19630 |
+
"step": 2787
|
| 19631 |
+
},
|
| 19632 |
+
{
|
| 19633 |
+
"epoch": 0.1330612926704132,
|
| 19634 |
+
"grad_norm": 5.977827548980713,
|
| 19635 |
+
"learning_rate": 0.00019905050870106767,
|
| 19636 |
+
"loss": 5.6266,
|
| 19637 |
+
"step": 2788
|
| 19638 |
+
},
|
| 19639 |
+
{
|
| 19640 |
+
"epoch": 0.13310901910250444,
|
| 19641 |
+
"grad_norm": 7.002156734466553,
|
| 19642 |
+
"learning_rate": 0.0001990498211327926,
|
| 19643 |
+
"loss": 7.0112,
|
| 19644 |
+
"step": 2789
|
| 19645 |
+
},
|
| 19646 |
+
{
|
| 19647 |
+
"epoch": 0.1331567455345957,
|
| 19648 |
+
"grad_norm": 6.144350528717041,
|
| 19649 |
+
"learning_rate": 0.0001990491333168468,
|
| 19650 |
+
"loss": 6.1978,
|
| 19651 |
+
"step": 2790
|
| 19652 |
+
},
|
| 19653 |
+
{
|
| 19654 |
+
"epoch": 0.13320447196668694,
|
| 19655 |
+
"grad_norm": 7.993871212005615,
|
| 19656 |
+
"learning_rate": 0.000199048445253232,
|
| 19657 |
+
"loss": 6.4143,
|
| 19658 |
+
"step": 2791
|
| 19659 |
+
},
|
| 19660 |
+
{
|
| 19661 |
+
"epoch": 0.1332521983987782,
|
| 19662 |
+
"grad_norm": 6.021445274353027,
|
| 19663 |
+
"learning_rate": 0.00019904775694195,
|
| 19664 |
+
"loss": 5.4286,
|
| 19665 |
+
"step": 2792
|
| 19666 |
+
},
|
| 19667 |
+
{
|
| 19668 |
+
"epoch": 0.13329992483086944,
|
| 19669 |
+
"grad_norm": 5.774000644683838,
|
| 19670 |
+
"learning_rate": 0.00019904706838300243,
|
| 19671 |
+
"loss": 5.5089,
|
| 19672 |
+
"step": 2793
|
| 19673 |
+
},
|
| 19674 |
+
{
|
| 19675 |
+
"epoch": 0.13334765126296072,
|
| 19676 |
+
"grad_norm": 6.7085185050964355,
|
| 19677 |
+
"learning_rate": 0.00019904637957639108,
|
| 19678 |
+
"loss": 7.9083,
|
| 19679 |
+
"step": 2794
|
| 19680 |
+
},
|
| 19681 |
+
{
|
| 19682 |
+
"epoch": 0.13339537769505197,
|
| 19683 |
+
"grad_norm": 7.979789733886719,
|
| 19684 |
+
"learning_rate": 0.00019904569052211764,
|
| 19685 |
+
"loss": 6.8796,
|
| 19686 |
+
"step": 2795
|
| 19687 |
+
},
|
| 19688 |
+
{
|
| 19689 |
+
"epoch": 0.13344310412714322,
|
| 19690 |
+
"grad_norm": 9.78246784210205,
|
| 19691 |
+
"learning_rate": 0.0001990450012201838,
|
| 19692 |
+
"loss": 9.3304,
|
| 19693 |
+
"step": 2796
|
| 19694 |
+
},
|
| 19695 |
+
{
|
| 19696 |
+
"epoch": 0.13349083055923447,
|
| 19697 |
+
"grad_norm": 5.552200794219971,
|
| 19698 |
+
"learning_rate": 0.00019904431167059135,
|
| 19699 |
+
"loss": 5.1558,
|
| 19700 |
+
"step": 2797
|
| 19701 |
+
},
|
| 19702 |
+
{
|
| 19703 |
+
"epoch": 0.13353855699132572,
|
| 19704 |
+
"grad_norm": 6.368596076965332,
|
| 19705 |
+
"learning_rate": 0.00019904362187334198,
|
| 19706 |
+
"loss": 6.2315,
|
| 19707 |
+
"step": 2798
|
| 19708 |
+
},
|
| 19709 |
+
{
|
| 19710 |
+
"epoch": 0.13358628342341697,
|
| 19711 |
+
"grad_norm": 4.584041118621826,
|
| 19712 |
+
"learning_rate": 0.00019904293182843738,
|
| 19713 |
+
"loss": 5.0416,
|
| 19714 |
+
"step": 2799
|
| 19715 |
+
},
|
| 19716 |
+
{
|
| 19717 |
+
"epoch": 0.13363400985550822,
|
| 19718 |
+
"grad_norm": 6.690604209899902,
|
| 19719 |
+
"learning_rate": 0.00019904224153587935,
|
| 19720 |
+
"loss": 6.5086,
|
| 19721 |
+
"step": 2800
|
| 19722 |
+
},
|
| 19723 |
+
{
|
| 19724 |
+
"epoch": 0.13363400985550822,
|
| 19725 |
+
"eval_loss": 1.6191966533660889,
|
| 19726 |
+
"eval_runtime": 96.5269,
|
| 19727 |
+
"eval_samples_per_second": 8.733,
|
| 19728 |
+
"eval_steps_per_second": 4.372,
|
| 19729 |
+
"step": 2800
|
| 19730 |
}
|
| 19731 |
],
|
| 19732 |
"logging_steps": 1,
|
|
|
|
| 19741 |
"early_stopping_threshold": 0.0
|
| 19742 |
},
|
| 19743 |
"attributes": {
|
| 19744 |
+
"early_stopping_patience_counter": 3
|
| 19745 |
}
|
| 19746 |
},
|
| 19747 |
"TrainerControl": {
|
|
|
|
| 19750 |
"should_evaluate": false,
|
| 19751 |
"should_log": false,
|
| 19752 |
"should_save": true,
|
| 19753 |
+
"should_training_stop": true
|
| 19754 |
},
|
| 19755 |
"attributes": {}
|
| 19756 |
}
|
| 19757 |
},
|
| 19758 |
+
"total_flos": 2.596245419851776e+17,
|
| 19759 |
"train_batch_size": 2,
|
| 19760 |
"trial_name": null,
|
| 19761 |
"trial_params": null
|