Training in progress, step 800, 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 100966336
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ba6ec73b4c1f01c38c8a1fbfe9edfbc9b2f0b1b14df0cb2f371e7a5b5513d981
|
| 3 |
size 100966336
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 51613668
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4cecf779d92ea0121f02faecf8aab1b19dcc9365e8cf962e2baca405adbe756b
|
| 3 |
size 51613668
|
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 sha256:732a841e8b824ecb84ca891104a1d8c2b7dc7f6dee6f7dbbc378f580359dfe51
|
| 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 sha256:3e8775ee34d50e95cd30d80b301c9a1df173b48fcf7afed02023d13d99886046
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "miner_id_24/checkpoint-
|
| 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,
|
|
@@ -4239,6 +4239,1414 @@
|
|
| 4239 |
"eval_samples_per_second": 31.783,
|
| 4240 |
"eval_steps_per_second": 2.032,
|
| 4241 |
"step": 600
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4242 |
}
|
| 4243 |
],
|
| 4244 |
"logging_steps": 1,
|
|
@@ -4267,7 +5675,7 @@
|
|
| 4267 |
"attributes": {}
|
| 4268 |
}
|
| 4269 |
},
|
| 4270 |
-
"total_flos":
|
| 4271 |
"train_batch_size": 16,
|
| 4272 |
"trial_name": null,
|
| 4273 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.5467618703842163,
|
| 3 |
+
"best_model_checkpoint": "miner_id_24/checkpoint-800",
|
| 4 |
+
"epoch": 0.5893186003683242,
|
| 5 |
"eval_steps": 200,
|
| 6 |
+
"global_step": 800,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 4239 |
"eval_samples_per_second": 31.783,
|
| 4240 |
"eval_steps_per_second": 2.032,
|
| 4241 |
"step": 600
|
| 4242 |
+
},
|
| 4243 |
+
{
|
| 4244 |
+
"epoch": 0.4427255985267035,
|
| 4245 |
+
"grad_norm": 0.3408944308757782,
|
| 4246 |
+
"learning_rate": 0.0001903078361565149,
|
| 4247 |
+
"loss": 0.5812,
|
| 4248 |
+
"step": 601
|
| 4249 |
+
},
|
| 4250 |
+
{
|
| 4251 |
+
"epoch": 0.4434622467771639,
|
| 4252 |
+
"grad_norm": 0.3410055935382843,
|
| 4253 |
+
"learning_rate": 0.00019027442015197586,
|
| 4254 |
+
"loss": 0.5645,
|
| 4255 |
+
"step": 602
|
| 4256 |
+
},
|
| 4257 |
+
{
|
| 4258 |
+
"epoch": 0.44419889502762433,
|
| 4259 |
+
"grad_norm": 0.30768829584121704,
|
| 4260 |
+
"learning_rate": 0.00019024094958588296,
|
| 4261 |
+
"loss": 0.642,
|
| 4262 |
+
"step": 603
|
| 4263 |
+
},
|
| 4264 |
+
{
|
| 4265 |
+
"epoch": 0.4449355432780847,
|
| 4266 |
+
"grad_norm": 0.2911585867404938,
|
| 4267 |
+
"learning_rate": 0.00019020742447846566,
|
| 4268 |
+
"loss": 0.5639,
|
| 4269 |
+
"step": 604
|
| 4270 |
+
},
|
| 4271 |
+
{
|
| 4272 |
+
"epoch": 0.44567219152854515,
|
| 4273 |
+
"grad_norm": 0.3137848675251007,
|
| 4274 |
+
"learning_rate": 0.0001901738448499865,
|
| 4275 |
+
"loss": 0.466,
|
| 4276 |
+
"step": 605
|
| 4277 |
+
},
|
| 4278 |
+
{
|
| 4279 |
+
"epoch": 0.44640883977900553,
|
| 4280 |
+
"grad_norm": 0.2971612811088562,
|
| 4281 |
+
"learning_rate": 0.0001901402107207408,
|
| 4282 |
+
"loss": 0.5571,
|
| 4283 |
+
"step": 606
|
| 4284 |
+
},
|
| 4285 |
+
{
|
| 4286 |
+
"epoch": 0.4471454880294659,
|
| 4287 |
+
"grad_norm": 0.3212316632270813,
|
| 4288 |
+
"learning_rate": 0.00019010652211105694,
|
| 4289 |
+
"loss": 0.5885,
|
| 4290 |
+
"step": 607
|
| 4291 |
+
},
|
| 4292 |
+
{
|
| 4293 |
+
"epoch": 0.44788213627992635,
|
| 4294 |
+
"grad_norm": 0.35693487524986267,
|
| 4295 |
+
"learning_rate": 0.00019007277904129627,
|
| 4296 |
+
"loss": 0.6198,
|
| 4297 |
+
"step": 608
|
| 4298 |
+
},
|
| 4299 |
+
{
|
| 4300 |
+
"epoch": 0.4486187845303867,
|
| 4301 |
+
"grad_norm": 0.2876060903072357,
|
| 4302 |
+
"learning_rate": 0.0001900389815318529,
|
| 4303 |
+
"loss": 0.4793,
|
| 4304 |
+
"step": 609
|
| 4305 |
+
},
|
| 4306 |
+
{
|
| 4307 |
+
"epoch": 0.44935543278084716,
|
| 4308 |
+
"grad_norm": 0.3253692388534546,
|
| 4309 |
+
"learning_rate": 0.00019000512960315398,
|
| 4310 |
+
"loss": 0.5678,
|
| 4311 |
+
"step": 610
|
| 4312 |
+
},
|
| 4313 |
+
{
|
| 4314 |
+
"epoch": 0.45009208103130754,
|
| 4315 |
+
"grad_norm": 0.28405922651290894,
|
| 4316 |
+
"learning_rate": 0.0001899712232756595,
|
| 4317 |
+
"loss": 0.5775,
|
| 4318 |
+
"step": 611
|
| 4319 |
+
},
|
| 4320 |
+
{
|
| 4321 |
+
"epoch": 0.450828729281768,
|
| 4322 |
+
"grad_norm": 0.33833935856819153,
|
| 4323 |
+
"learning_rate": 0.00018993726256986228,
|
| 4324 |
+
"loss": 0.5041,
|
| 4325 |
+
"step": 612
|
| 4326 |
+
},
|
| 4327 |
+
{
|
| 4328 |
+
"epoch": 0.45156537753222836,
|
| 4329 |
+
"grad_norm": 0.2696160674095154,
|
| 4330 |
+
"learning_rate": 0.00018990324750628812,
|
| 4331 |
+
"loss": 0.5195,
|
| 4332 |
+
"step": 613
|
| 4333 |
+
},
|
| 4334 |
+
{
|
| 4335 |
+
"epoch": 0.45230202578268874,
|
| 4336 |
+
"grad_norm": 0.3042635917663574,
|
| 4337 |
+
"learning_rate": 0.00018986917810549558,
|
| 4338 |
+
"loss": 0.5394,
|
| 4339 |
+
"step": 614
|
| 4340 |
+
},
|
| 4341 |
+
{
|
| 4342 |
+
"epoch": 0.4530386740331492,
|
| 4343 |
+
"grad_norm": 0.40808773040771484,
|
| 4344 |
+
"learning_rate": 0.00018983505438807606,
|
| 4345 |
+
"loss": 0.5708,
|
| 4346 |
+
"step": 615
|
| 4347 |
+
},
|
| 4348 |
+
{
|
| 4349 |
+
"epoch": 0.45377532228360956,
|
| 4350 |
+
"grad_norm": 0.33351388573646545,
|
| 4351 |
+
"learning_rate": 0.0001898008763746539,
|
| 4352 |
+
"loss": 0.4879,
|
| 4353 |
+
"step": 616
|
| 4354 |
+
},
|
| 4355 |
+
{
|
| 4356 |
+
"epoch": 0.45451197053407,
|
| 4357 |
+
"grad_norm": 0.30167388916015625,
|
| 4358 |
+
"learning_rate": 0.00018976664408588606,
|
| 4359 |
+
"loss": 0.5058,
|
| 4360 |
+
"step": 617
|
| 4361 |
+
},
|
| 4362 |
+
{
|
| 4363 |
+
"epoch": 0.4552486187845304,
|
| 4364 |
+
"grad_norm": 0.32418936491012573,
|
| 4365 |
+
"learning_rate": 0.00018973235754246253,
|
| 4366 |
+
"loss": 0.5538,
|
| 4367 |
+
"step": 618
|
| 4368 |
+
},
|
| 4369 |
+
{
|
| 4370 |
+
"epoch": 0.4559852670349908,
|
| 4371 |
+
"grad_norm": 0.33115559816360474,
|
| 4372 |
+
"learning_rate": 0.00018969801676510593,
|
| 4373 |
+
"loss": 0.5728,
|
| 4374 |
+
"step": 619
|
| 4375 |
+
},
|
| 4376 |
+
{
|
| 4377 |
+
"epoch": 0.4567219152854512,
|
| 4378 |
+
"grad_norm": 0.31681492924690247,
|
| 4379 |
+
"learning_rate": 0.0001896636217745717,
|
| 4380 |
+
"loss": 0.4679,
|
| 4381 |
+
"step": 620
|
| 4382 |
+
},
|
| 4383 |
+
{
|
| 4384 |
+
"epoch": 0.4574585635359116,
|
| 4385 |
+
"grad_norm": 0.3244980573654175,
|
| 4386 |
+
"learning_rate": 0.0001896291725916481,
|
| 4387 |
+
"loss": 0.5237,
|
| 4388 |
+
"step": 621
|
| 4389 |
+
},
|
| 4390 |
+
{
|
| 4391 |
+
"epoch": 0.458195211786372,
|
| 4392 |
+
"grad_norm": 0.32656654715538025,
|
| 4393 |
+
"learning_rate": 0.00018959466923715605,
|
| 4394 |
+
"loss": 0.6019,
|
| 4395 |
+
"step": 622
|
| 4396 |
+
},
|
| 4397 |
+
{
|
| 4398 |
+
"epoch": 0.4589318600368324,
|
| 4399 |
+
"grad_norm": 0.36800074577331543,
|
| 4400 |
+
"learning_rate": 0.0001895601117319493,
|
| 4401 |
+
"loss": 0.5427,
|
| 4402 |
+
"step": 623
|
| 4403 |
+
},
|
| 4404 |
+
{
|
| 4405 |
+
"epoch": 0.45966850828729283,
|
| 4406 |
+
"grad_norm": 0.3162236213684082,
|
| 4407 |
+
"learning_rate": 0.00018952550009691422,
|
| 4408 |
+
"loss": 0.5323,
|
| 4409 |
+
"step": 624
|
| 4410 |
+
},
|
| 4411 |
+
{
|
| 4412 |
+
"epoch": 0.4604051565377532,
|
| 4413 |
+
"grad_norm": 0.3138112425804138,
|
| 4414 |
+
"learning_rate": 0.00018949083435297007,
|
| 4415 |
+
"loss": 0.6297,
|
| 4416 |
+
"step": 625
|
| 4417 |
+
},
|
| 4418 |
+
{
|
| 4419 |
+
"epoch": 0.46114180478821365,
|
| 4420 |
+
"grad_norm": 0.37662187218666077,
|
| 4421 |
+
"learning_rate": 0.0001894561145210686,
|
| 4422 |
+
"loss": 0.5936,
|
| 4423 |
+
"step": 626
|
| 4424 |
+
},
|
| 4425 |
+
{
|
| 4426 |
+
"epoch": 0.46187845303867403,
|
| 4427 |
+
"grad_norm": 0.3716781735420227,
|
| 4428 |
+
"learning_rate": 0.0001894213406221945,
|
| 4429 |
+
"loss": 0.5165,
|
| 4430 |
+
"step": 627
|
| 4431 |
+
},
|
| 4432 |
+
{
|
| 4433 |
+
"epoch": 0.4626151012891344,
|
| 4434 |
+
"grad_norm": 0.34003299474716187,
|
| 4435 |
+
"learning_rate": 0.00018938651267736485,
|
| 4436 |
+
"loss": 0.5975,
|
| 4437 |
+
"step": 628
|
| 4438 |
+
},
|
| 4439 |
+
{
|
| 4440 |
+
"epoch": 0.46335174953959485,
|
| 4441 |
+
"grad_norm": 0.3482089936733246,
|
| 4442 |
+
"learning_rate": 0.0001893516307076296,
|
| 4443 |
+
"loss": 0.5336,
|
| 4444 |
+
"step": 629
|
| 4445 |
+
},
|
| 4446 |
+
{
|
| 4447 |
+
"epoch": 0.46408839779005523,
|
| 4448 |
+
"grad_norm": 0.3335133194923401,
|
| 4449 |
+
"learning_rate": 0.0001893166947340713,
|
| 4450 |
+
"loss": 0.5662,
|
| 4451 |
+
"step": 630
|
| 4452 |
+
},
|
| 4453 |
+
{
|
| 4454 |
+
"epoch": 0.46482504604051567,
|
| 4455 |
+
"grad_norm": 0.3414890468120575,
|
| 4456 |
+
"learning_rate": 0.00018928170477780512,
|
| 4457 |
+
"loss": 0.6296,
|
| 4458 |
+
"step": 631
|
| 4459 |
+
},
|
| 4460 |
+
{
|
| 4461 |
+
"epoch": 0.46556169429097605,
|
| 4462 |
+
"grad_norm": 0.34633347392082214,
|
| 4463 |
+
"learning_rate": 0.0001892466608599789,
|
| 4464 |
+
"loss": 0.5826,
|
| 4465 |
+
"step": 632
|
| 4466 |
+
},
|
| 4467 |
+
{
|
| 4468 |
+
"epoch": 0.4662983425414365,
|
| 4469 |
+
"grad_norm": 0.32681214809417725,
|
| 4470 |
+
"learning_rate": 0.00018921156300177302,
|
| 4471 |
+
"loss": 0.4984,
|
| 4472 |
+
"step": 633
|
| 4473 |
+
},
|
| 4474 |
+
{
|
| 4475 |
+
"epoch": 0.46703499079189686,
|
| 4476 |
+
"grad_norm": 0.32555603981018066,
|
| 4477 |
+
"learning_rate": 0.00018917641122440055,
|
| 4478 |
+
"loss": 0.4791,
|
| 4479 |
+
"step": 634
|
| 4480 |
+
},
|
| 4481 |
+
{
|
| 4482 |
+
"epoch": 0.4677716390423573,
|
| 4483 |
+
"grad_norm": 0.2741229236125946,
|
| 4484 |
+
"learning_rate": 0.00018914120554910707,
|
| 4485 |
+
"loss": 0.4922,
|
| 4486 |
+
"step": 635
|
| 4487 |
+
},
|
| 4488 |
+
{
|
| 4489 |
+
"epoch": 0.4685082872928177,
|
| 4490 |
+
"grad_norm": 0.29967793822288513,
|
| 4491 |
+
"learning_rate": 0.00018910594599717083,
|
| 4492 |
+
"loss": 0.4675,
|
| 4493 |
+
"step": 636
|
| 4494 |
+
},
|
| 4495 |
+
{
|
| 4496 |
+
"epoch": 0.46924493554327806,
|
| 4497 |
+
"grad_norm": 0.3687807321548462,
|
| 4498 |
+
"learning_rate": 0.00018907063258990253,
|
| 4499 |
+
"loss": 0.6645,
|
| 4500 |
+
"step": 637
|
| 4501 |
+
},
|
| 4502 |
+
{
|
| 4503 |
+
"epoch": 0.4699815837937385,
|
| 4504 |
+
"grad_norm": 0.33013150095939636,
|
| 4505 |
+
"learning_rate": 0.00018903526534864548,
|
| 4506 |
+
"loss": 0.5858,
|
| 4507 |
+
"step": 638
|
| 4508 |
+
},
|
| 4509 |
+
{
|
| 4510 |
+
"epoch": 0.4707182320441989,
|
| 4511 |
+
"grad_norm": 0.3501259982585907,
|
| 4512 |
+
"learning_rate": 0.00018899984429477553,
|
| 4513 |
+
"loss": 0.5093,
|
| 4514 |
+
"step": 639
|
| 4515 |
+
},
|
| 4516 |
+
{
|
| 4517 |
+
"epoch": 0.4714548802946593,
|
| 4518 |
+
"grad_norm": 0.35059183835983276,
|
| 4519 |
+
"learning_rate": 0.00018896436944970102,
|
| 4520 |
+
"loss": 0.5427,
|
| 4521 |
+
"step": 640
|
| 4522 |
+
},
|
| 4523 |
+
{
|
| 4524 |
+
"epoch": 0.4721915285451197,
|
| 4525 |
+
"grad_norm": 0.3216591775417328,
|
| 4526 |
+
"learning_rate": 0.00018892884083486285,
|
| 4527 |
+
"loss": 0.5714,
|
| 4528 |
+
"step": 641
|
| 4529 |
+
},
|
| 4530 |
+
{
|
| 4531 |
+
"epoch": 0.47292817679558014,
|
| 4532 |
+
"grad_norm": 0.3616012632846832,
|
| 4533 |
+
"learning_rate": 0.00018889325847173442,
|
| 4534 |
+
"loss": 0.6256,
|
| 4535 |
+
"step": 642
|
| 4536 |
+
},
|
| 4537 |
+
{
|
| 4538 |
+
"epoch": 0.4736648250460405,
|
| 4539 |
+
"grad_norm": 0.34034106135368347,
|
| 4540 |
+
"learning_rate": 0.00018885762238182157,
|
| 4541 |
+
"loss": 0.5334,
|
| 4542 |
+
"step": 643
|
| 4543 |
+
},
|
| 4544 |
+
{
|
| 4545 |
+
"epoch": 0.4744014732965009,
|
| 4546 |
+
"grad_norm": 0.3253825008869171,
|
| 4547 |
+
"learning_rate": 0.00018882193258666258,
|
| 4548 |
+
"loss": 0.5722,
|
| 4549 |
+
"step": 644
|
| 4550 |
+
},
|
| 4551 |
+
{
|
| 4552 |
+
"epoch": 0.47513812154696133,
|
| 4553 |
+
"grad_norm": 0.31885895133018494,
|
| 4554 |
+
"learning_rate": 0.00018878618910782833,
|
| 4555 |
+
"loss": 0.4629,
|
| 4556 |
+
"step": 645
|
| 4557 |
+
},
|
| 4558 |
+
{
|
| 4559 |
+
"epoch": 0.4758747697974217,
|
| 4560 |
+
"grad_norm": 0.3061891198158264,
|
| 4561 |
+
"learning_rate": 0.00018875039196692198,
|
| 4562 |
+
"loss": 0.4436,
|
| 4563 |
+
"step": 646
|
| 4564 |
+
},
|
| 4565 |
+
{
|
| 4566 |
+
"epoch": 0.47661141804788215,
|
| 4567 |
+
"grad_norm": 0.32976722717285156,
|
| 4568 |
+
"learning_rate": 0.00018871454118557924,
|
| 4569 |
+
"loss": 0.5668,
|
| 4570 |
+
"step": 647
|
| 4571 |
+
},
|
| 4572 |
+
{
|
| 4573 |
+
"epoch": 0.47734806629834253,
|
| 4574 |
+
"grad_norm": 0.3309568464756012,
|
| 4575 |
+
"learning_rate": 0.00018867863678546823,
|
| 4576 |
+
"loss": 0.5349,
|
| 4577 |
+
"step": 648
|
| 4578 |
+
},
|
| 4579 |
+
{
|
| 4580 |
+
"epoch": 0.47808471454880297,
|
| 4581 |
+
"grad_norm": 0.3136848509311676,
|
| 4582 |
+
"learning_rate": 0.00018864267878828937,
|
| 4583 |
+
"loss": 0.5569,
|
| 4584 |
+
"step": 649
|
| 4585 |
+
},
|
| 4586 |
+
{
|
| 4587 |
+
"epoch": 0.47882136279926335,
|
| 4588 |
+
"grad_norm": 0.359239399433136,
|
| 4589 |
+
"learning_rate": 0.00018860666721577562,
|
| 4590 |
+
"loss": 0.544,
|
| 4591 |
+
"step": 650
|
| 4592 |
+
},
|
| 4593 |
+
{
|
| 4594 |
+
"epoch": 0.47955801104972373,
|
| 4595 |
+
"grad_norm": 0.30393654108047485,
|
| 4596 |
+
"learning_rate": 0.00018857060208969225,
|
| 4597 |
+
"loss": 0.4861,
|
| 4598 |
+
"step": 651
|
| 4599 |
+
},
|
| 4600 |
+
{
|
| 4601 |
+
"epoch": 0.48029465930018417,
|
| 4602 |
+
"grad_norm": 0.3133644461631775,
|
| 4603 |
+
"learning_rate": 0.00018853448343183682,
|
| 4604 |
+
"loss": 0.5472,
|
| 4605 |
+
"step": 652
|
| 4606 |
+
},
|
| 4607 |
+
{
|
| 4608 |
+
"epoch": 0.48103130755064455,
|
| 4609 |
+
"grad_norm": 0.2803438603878021,
|
| 4610 |
+
"learning_rate": 0.00018849831126403942,
|
| 4611 |
+
"loss": 0.5747,
|
| 4612 |
+
"step": 653
|
| 4613 |
+
},
|
| 4614 |
+
{
|
| 4615 |
+
"epoch": 0.481767955801105,
|
| 4616 |
+
"grad_norm": 0.34794047474861145,
|
| 4617 |
+
"learning_rate": 0.00018846208560816235,
|
| 4618 |
+
"loss": 0.5907,
|
| 4619 |
+
"step": 654
|
| 4620 |
+
},
|
| 4621 |
+
{
|
| 4622 |
+
"epoch": 0.48250460405156537,
|
| 4623 |
+
"grad_norm": 0.3284980058670044,
|
| 4624 |
+
"learning_rate": 0.00018842580648610025,
|
| 4625 |
+
"loss": 0.4845,
|
| 4626 |
+
"step": 655
|
| 4627 |
+
},
|
| 4628 |
+
{
|
| 4629 |
+
"epoch": 0.4832412523020258,
|
| 4630 |
+
"grad_norm": 0.3168398141860962,
|
| 4631 |
+
"learning_rate": 0.00018838947391978012,
|
| 4632 |
+
"loss": 0.5116,
|
| 4633 |
+
"step": 656
|
| 4634 |
+
},
|
| 4635 |
+
{
|
| 4636 |
+
"epoch": 0.4839779005524862,
|
| 4637 |
+
"grad_norm": 0.3016391694545746,
|
| 4638 |
+
"learning_rate": 0.00018835308793116123,
|
| 4639 |
+
"loss": 0.5041,
|
| 4640 |
+
"step": 657
|
| 4641 |
+
},
|
| 4642 |
+
{
|
| 4643 |
+
"epoch": 0.48471454880294657,
|
| 4644 |
+
"grad_norm": 0.37949711084365845,
|
| 4645 |
+
"learning_rate": 0.00018831664854223522,
|
| 4646 |
+
"loss": 0.5363,
|
| 4647 |
+
"step": 658
|
| 4648 |
+
},
|
| 4649 |
+
{
|
| 4650 |
+
"epoch": 0.485451197053407,
|
| 4651 |
+
"grad_norm": 0.3321974575519562,
|
| 4652 |
+
"learning_rate": 0.00018828015577502583,
|
| 4653 |
+
"loss": 0.4964,
|
| 4654 |
+
"step": 659
|
| 4655 |
+
},
|
| 4656 |
+
{
|
| 4657 |
+
"epoch": 0.4861878453038674,
|
| 4658 |
+
"grad_norm": 0.32142874598503113,
|
| 4659 |
+
"learning_rate": 0.00018824360965158921,
|
| 4660 |
+
"loss": 0.4954,
|
| 4661 |
+
"step": 660
|
| 4662 |
+
},
|
| 4663 |
+
{
|
| 4664 |
+
"epoch": 0.4869244935543278,
|
| 4665 |
+
"grad_norm": 0.31010639667510986,
|
| 4666 |
+
"learning_rate": 0.0001882070101940137,
|
| 4667 |
+
"loss": 0.5343,
|
| 4668 |
+
"step": 661
|
| 4669 |
+
},
|
| 4670 |
+
{
|
| 4671 |
+
"epoch": 0.4876611418047882,
|
| 4672 |
+
"grad_norm": 0.3347371220588684,
|
| 4673 |
+
"learning_rate": 0.00018817035742441988,
|
| 4674 |
+
"loss": 0.5106,
|
| 4675 |
+
"step": 662
|
| 4676 |
+
},
|
| 4677 |
+
{
|
| 4678 |
+
"epoch": 0.48839779005524864,
|
| 4679 |
+
"grad_norm": 0.3143859803676605,
|
| 4680 |
+
"learning_rate": 0.0001881336513649606,
|
| 4681 |
+
"loss": 0.5621,
|
| 4682 |
+
"step": 663
|
| 4683 |
+
},
|
| 4684 |
+
{
|
| 4685 |
+
"epoch": 0.489134438305709,
|
| 4686 |
+
"grad_norm": 0.3149314820766449,
|
| 4687 |
+
"learning_rate": 0.00018809689203782087,
|
| 4688 |
+
"loss": 0.5415,
|
| 4689 |
+
"step": 664
|
| 4690 |
+
},
|
| 4691 |
+
{
|
| 4692 |
+
"epoch": 0.48987108655616945,
|
| 4693 |
+
"grad_norm": 0.3495696783065796,
|
| 4694 |
+
"learning_rate": 0.00018806007946521792,
|
| 4695 |
+
"loss": 0.5333,
|
| 4696 |
+
"step": 665
|
| 4697 |
+
},
|
| 4698 |
+
{
|
| 4699 |
+
"epoch": 0.49060773480662984,
|
| 4700 |
+
"grad_norm": 0.33296969532966614,
|
| 4701 |
+
"learning_rate": 0.00018802321366940105,
|
| 4702 |
+
"loss": 0.5616,
|
| 4703 |
+
"step": 666
|
| 4704 |
+
},
|
| 4705 |
+
{
|
| 4706 |
+
"epoch": 0.4913443830570902,
|
| 4707 |
+
"grad_norm": 0.37067824602127075,
|
| 4708 |
+
"learning_rate": 0.0001879862946726519,
|
| 4709 |
+
"loss": 0.5168,
|
| 4710 |
+
"step": 667
|
| 4711 |
+
},
|
| 4712 |
+
{
|
| 4713 |
+
"epoch": 0.49208103130755065,
|
| 4714 |
+
"grad_norm": 0.34374094009399414,
|
| 4715 |
+
"learning_rate": 0.0001879493224972842,
|
| 4716 |
+
"loss": 0.575,
|
| 4717 |
+
"step": 668
|
| 4718 |
+
},
|
| 4719 |
+
{
|
| 4720 |
+
"epoch": 0.49281767955801103,
|
| 4721 |
+
"grad_norm": 0.30375105142593384,
|
| 4722 |
+
"learning_rate": 0.00018791229716564375,
|
| 4723 |
+
"loss": 0.547,
|
| 4724 |
+
"step": 669
|
| 4725 |
+
},
|
| 4726 |
+
{
|
| 4727 |
+
"epoch": 0.49355432780847147,
|
| 4728 |
+
"grad_norm": 0.3219543397426605,
|
| 4729 |
+
"learning_rate": 0.0001878752187001086,
|
| 4730 |
+
"loss": 0.5804,
|
| 4731 |
+
"step": 670
|
| 4732 |
+
},
|
| 4733 |
+
{
|
| 4734 |
+
"epoch": 0.49429097605893185,
|
| 4735 |
+
"grad_norm": 0.3195449113845825,
|
| 4736 |
+
"learning_rate": 0.00018783808712308876,
|
| 4737 |
+
"loss": 0.5321,
|
| 4738 |
+
"step": 671
|
| 4739 |
+
},
|
| 4740 |
+
{
|
| 4741 |
+
"epoch": 0.4950276243093923,
|
| 4742 |
+
"grad_norm": 0.3505839407444,
|
| 4743 |
+
"learning_rate": 0.00018780090245702652,
|
| 4744 |
+
"loss": 0.5573,
|
| 4745 |
+
"step": 672
|
| 4746 |
+
},
|
| 4747 |
+
{
|
| 4748 |
+
"epoch": 0.49576427255985267,
|
| 4749 |
+
"grad_norm": 0.3471439778804779,
|
| 4750 |
+
"learning_rate": 0.0001877636647243961,
|
| 4751 |
+
"loss": 0.5552,
|
| 4752 |
+
"step": 673
|
| 4753 |
+
},
|
| 4754 |
+
{
|
| 4755 |
+
"epoch": 0.49650092081031305,
|
| 4756 |
+
"grad_norm": 0.31486374139785767,
|
| 4757 |
+
"learning_rate": 0.00018772637394770392,
|
| 4758 |
+
"loss": 0.4646,
|
| 4759 |
+
"step": 674
|
| 4760 |
+
},
|
| 4761 |
+
{
|
| 4762 |
+
"epoch": 0.4972375690607735,
|
| 4763 |
+
"grad_norm": 0.3076501786708832,
|
| 4764 |
+
"learning_rate": 0.00018768903014948837,
|
| 4765 |
+
"loss": 0.5305,
|
| 4766 |
+
"step": 675
|
| 4767 |
+
},
|
| 4768 |
+
{
|
| 4769 |
+
"epoch": 0.49797421731123387,
|
| 4770 |
+
"grad_norm": 0.32506081461906433,
|
| 4771 |
+
"learning_rate": 0.0001876516333523199,
|
| 4772 |
+
"loss": 0.5336,
|
| 4773 |
+
"step": 676
|
| 4774 |
+
},
|
| 4775 |
+
{
|
| 4776 |
+
"epoch": 0.4987108655616943,
|
| 4777 |
+
"grad_norm": 0.3549332320690155,
|
| 4778 |
+
"learning_rate": 0.00018761418357880102,
|
| 4779 |
+
"loss": 0.5418,
|
| 4780 |
+
"step": 677
|
| 4781 |
+
},
|
| 4782 |
+
{
|
| 4783 |
+
"epoch": 0.4994475138121547,
|
| 4784 |
+
"grad_norm": 0.3804044723510742,
|
| 4785 |
+
"learning_rate": 0.00018757668085156627,
|
| 4786 |
+
"loss": 0.6278,
|
| 4787 |
+
"step": 678
|
| 4788 |
+
},
|
| 4789 |
+
{
|
| 4790 |
+
"epoch": 0.5001841620626151,
|
| 4791 |
+
"grad_norm": 0.3792779743671417,
|
| 4792 |
+
"learning_rate": 0.0001875391251932821,
|
| 4793 |
+
"loss": 0.5879,
|
| 4794 |
+
"step": 679
|
| 4795 |
+
},
|
| 4796 |
+
{
|
| 4797 |
+
"epoch": 0.5009208103130756,
|
| 4798 |
+
"grad_norm": 0.3581220805644989,
|
| 4799 |
+
"learning_rate": 0.00018750151662664708,
|
| 4800 |
+
"loss": 0.5712,
|
| 4801 |
+
"step": 680
|
| 4802 |
+
},
|
| 4803 |
+
{
|
| 4804 |
+
"epoch": 0.5016574585635359,
|
| 4805 |
+
"grad_norm": 0.30549612641334534,
|
| 4806 |
+
"learning_rate": 0.00018746385517439174,
|
| 4807 |
+
"loss": 0.4795,
|
| 4808 |
+
"step": 681
|
| 4809 |
+
},
|
| 4810 |
+
{
|
| 4811 |
+
"epoch": 0.5023941068139963,
|
| 4812 |
+
"grad_norm": 0.3250758945941925,
|
| 4813 |
+
"learning_rate": 0.00018742614085927843,
|
| 4814 |
+
"loss": 0.5336,
|
| 4815 |
+
"step": 682
|
| 4816 |
+
},
|
| 4817 |
+
{
|
| 4818 |
+
"epoch": 0.5031307550644567,
|
| 4819 |
+
"grad_norm": 0.3441884219646454,
|
| 4820 |
+
"learning_rate": 0.00018738837370410166,
|
| 4821 |
+
"loss": 0.6303,
|
| 4822 |
+
"step": 683
|
| 4823 |
+
},
|
| 4824 |
+
{
|
| 4825 |
+
"epoch": 0.5038674033149171,
|
| 4826 |
+
"grad_norm": 0.33731353282928467,
|
| 4827 |
+
"learning_rate": 0.0001873505537316877,
|
| 4828 |
+
"loss": 0.4025,
|
| 4829 |
+
"step": 684
|
| 4830 |
+
},
|
| 4831 |
+
{
|
| 4832 |
+
"epoch": 0.5046040515653776,
|
| 4833 |
+
"grad_norm": 0.3341776430606842,
|
| 4834 |
+
"learning_rate": 0.0001873126809648948,
|
| 4835 |
+
"loss": 0.5485,
|
| 4836 |
+
"step": 685
|
| 4837 |
+
},
|
| 4838 |
+
{
|
| 4839 |
+
"epoch": 0.505340699815838,
|
| 4840 |
+
"grad_norm": 0.3634783625602722,
|
| 4841 |
+
"learning_rate": 0.00018727475542661321,
|
| 4842 |
+
"loss": 0.517,
|
| 4843 |
+
"step": 686
|
| 4844 |
+
},
|
| 4845 |
+
{
|
| 4846 |
+
"epoch": 0.5060773480662983,
|
| 4847 |
+
"grad_norm": 0.3433530032634735,
|
| 4848 |
+
"learning_rate": 0.00018723677713976494,
|
| 4849 |
+
"loss": 0.6425,
|
| 4850 |
+
"step": 687
|
| 4851 |
+
},
|
| 4852 |
+
{
|
| 4853 |
+
"epoch": 0.5068139963167587,
|
| 4854 |
+
"grad_norm": 0.28929224610328674,
|
| 4855 |
+
"learning_rate": 0.00018719874612730398,
|
| 4856 |
+
"loss": 0.5271,
|
| 4857 |
+
"step": 688
|
| 4858 |
+
},
|
| 4859 |
+
{
|
| 4860 |
+
"epoch": 0.5075506445672191,
|
| 4861 |
+
"grad_norm": 0.31161174178123474,
|
| 4862 |
+
"learning_rate": 0.00018716066241221605,
|
| 4863 |
+
"loss": 0.4219,
|
| 4864 |
+
"step": 689
|
| 4865 |
+
},
|
| 4866 |
+
{
|
| 4867 |
+
"epoch": 0.5082872928176796,
|
| 4868 |
+
"grad_norm": 0.35673725605010986,
|
| 4869 |
+
"learning_rate": 0.00018712252601751895,
|
| 4870 |
+
"loss": 0.4235,
|
| 4871 |
+
"step": 690
|
| 4872 |
+
},
|
| 4873 |
+
{
|
| 4874 |
+
"epoch": 0.50902394106814,
|
| 4875 |
+
"grad_norm": 0.3995974063873291,
|
| 4876 |
+
"learning_rate": 0.0001870843369662621,
|
| 4877 |
+
"loss": 0.5381,
|
| 4878 |
+
"step": 691
|
| 4879 |
+
},
|
| 4880 |
+
{
|
| 4881 |
+
"epoch": 0.5097605893186004,
|
| 4882 |
+
"grad_norm": 0.4052712023258209,
|
| 4883 |
+
"learning_rate": 0.00018704609528152685,
|
| 4884 |
+
"loss": 0.5407,
|
| 4885 |
+
"step": 692
|
| 4886 |
+
},
|
| 4887 |
+
{
|
| 4888 |
+
"epoch": 0.5104972375690607,
|
| 4889 |
+
"grad_norm": 0.3908758759498596,
|
| 4890 |
+
"learning_rate": 0.00018700780098642635,
|
| 4891 |
+
"loss": 0.6067,
|
| 4892 |
+
"step": 693
|
| 4893 |
+
},
|
| 4894 |
+
{
|
| 4895 |
+
"epoch": 0.5112338858195212,
|
| 4896 |
+
"grad_norm": 0.3414273262023926,
|
| 4897 |
+
"learning_rate": 0.00018696945410410555,
|
| 4898 |
+
"loss": 0.5591,
|
| 4899 |
+
"step": 694
|
| 4900 |
+
},
|
| 4901 |
+
{
|
| 4902 |
+
"epoch": 0.5119705340699816,
|
| 4903 |
+
"grad_norm": 0.3640691339969635,
|
| 4904 |
+
"learning_rate": 0.00018693105465774113,
|
| 4905 |
+
"loss": 0.5719,
|
| 4906 |
+
"step": 695
|
| 4907 |
+
},
|
| 4908 |
+
{
|
| 4909 |
+
"epoch": 0.512707182320442,
|
| 4910 |
+
"grad_norm": 0.3373907804489136,
|
| 4911 |
+
"learning_rate": 0.00018689260267054167,
|
| 4912 |
+
"loss": 0.5307,
|
| 4913 |
+
"step": 696
|
| 4914 |
+
},
|
| 4915 |
+
{
|
| 4916 |
+
"epoch": 0.5134438305709024,
|
| 4917 |
+
"grad_norm": 0.34063127636909485,
|
| 4918 |
+
"learning_rate": 0.00018685409816574734,
|
| 4919 |
+
"loss": 0.5852,
|
| 4920 |
+
"step": 697
|
| 4921 |
+
},
|
| 4922 |
+
{
|
| 4923 |
+
"epoch": 0.5141804788213628,
|
| 4924 |
+
"grad_norm": 0.30992797017097473,
|
| 4925 |
+
"learning_rate": 0.0001868155411666302,
|
| 4926 |
+
"loss": 0.4501,
|
| 4927 |
+
"step": 698
|
| 4928 |
+
},
|
| 4929 |
+
{
|
| 4930 |
+
"epoch": 0.5149171270718232,
|
| 4931 |
+
"grad_norm": 0.38110870122909546,
|
| 4932 |
+
"learning_rate": 0.00018677693169649392,
|
| 4933 |
+
"loss": 0.539,
|
| 4934 |
+
"step": 699
|
| 4935 |
+
},
|
| 4936 |
+
{
|
| 4937 |
+
"epoch": 0.5156537753222836,
|
| 4938 |
+
"grad_norm": 0.3289574086666107,
|
| 4939 |
+
"learning_rate": 0.00018673826977867397,
|
| 4940 |
+
"loss": 0.6027,
|
| 4941 |
+
"step": 700
|
| 4942 |
+
},
|
| 4943 |
+
{
|
| 4944 |
+
"epoch": 0.516390423572744,
|
| 4945 |
+
"grad_norm": 0.34149253368377686,
|
| 4946 |
+
"learning_rate": 0.00018669955543653744,
|
| 4947 |
+
"loss": 0.5224,
|
| 4948 |
+
"step": 701
|
| 4949 |
+
},
|
| 4950 |
+
{
|
| 4951 |
+
"epoch": 0.5171270718232044,
|
| 4952 |
+
"grad_norm": 0.35705652832984924,
|
| 4953 |
+
"learning_rate": 0.00018666078869348317,
|
| 4954 |
+
"loss": 0.6056,
|
| 4955 |
+
"step": 702
|
| 4956 |
+
},
|
| 4957 |
+
{
|
| 4958 |
+
"epoch": 0.5178637200736649,
|
| 4959 |
+
"grad_norm": 0.3409496247768402,
|
| 4960 |
+
"learning_rate": 0.00018662196957294167,
|
| 4961 |
+
"loss": 0.5263,
|
| 4962 |
+
"step": 703
|
| 4963 |
+
},
|
| 4964 |
+
{
|
| 4965 |
+
"epoch": 0.5186003683241253,
|
| 4966 |
+
"grad_norm": 0.3538173735141754,
|
| 4967 |
+
"learning_rate": 0.00018658309809837503,
|
| 4968 |
+
"loss": 0.5534,
|
| 4969 |
+
"step": 704
|
| 4970 |
+
},
|
| 4971 |
+
{
|
| 4972 |
+
"epoch": 0.5193370165745856,
|
| 4973 |
+
"grad_norm": 0.3721802830696106,
|
| 4974 |
+
"learning_rate": 0.0001865441742932771,
|
| 4975 |
+
"loss": 0.5129,
|
| 4976 |
+
"step": 705
|
| 4977 |
+
},
|
| 4978 |
+
{
|
| 4979 |
+
"epoch": 0.520073664825046,
|
| 4980 |
+
"grad_norm": 0.32195401191711426,
|
| 4981 |
+
"learning_rate": 0.00018650519818117327,
|
| 4982 |
+
"loss": 0.4938,
|
| 4983 |
+
"step": 706
|
| 4984 |
+
},
|
| 4985 |
+
{
|
| 4986 |
+
"epoch": 0.5208103130755064,
|
| 4987 |
+
"grad_norm": 0.3473074734210968,
|
| 4988 |
+
"learning_rate": 0.00018646616978562053,
|
| 4989 |
+
"loss": 0.5828,
|
| 4990 |
+
"step": 707
|
| 4991 |
+
},
|
| 4992 |
+
{
|
| 4993 |
+
"epoch": 0.5215469613259669,
|
| 4994 |
+
"grad_norm": 0.3335668742656708,
|
| 4995 |
+
"learning_rate": 0.00018642708913020758,
|
| 4996 |
+
"loss": 0.4803,
|
| 4997 |
+
"step": 708
|
| 4998 |
+
},
|
| 4999 |
+
{
|
| 5000 |
+
"epoch": 0.5222836095764273,
|
| 5001 |
+
"grad_norm": 0.3301938772201538,
|
| 5002 |
+
"learning_rate": 0.00018638795623855458,
|
| 5003 |
+
"loss": 0.4469,
|
| 5004 |
+
"step": 709
|
| 5005 |
+
},
|
| 5006 |
+
{
|
| 5007 |
+
"epoch": 0.5230202578268877,
|
| 5008 |
+
"grad_norm": 0.32511696219444275,
|
| 5009 |
+
"learning_rate": 0.00018634877113431336,
|
| 5010 |
+
"loss": 0.5289,
|
| 5011 |
+
"step": 710
|
| 5012 |
+
},
|
| 5013 |
+
{
|
| 5014 |
+
"epoch": 0.523756906077348,
|
| 5015 |
+
"grad_norm": 0.33410945534706116,
|
| 5016 |
+
"learning_rate": 0.00018630953384116722,
|
| 5017 |
+
"loss": 0.5081,
|
| 5018 |
+
"step": 711
|
| 5019 |
+
},
|
| 5020 |
+
{
|
| 5021 |
+
"epoch": 0.5244935543278084,
|
| 5022 |
+
"grad_norm": 0.38132521510124207,
|
| 5023 |
+
"learning_rate": 0.00018627024438283107,
|
| 5024 |
+
"loss": 0.5885,
|
| 5025 |
+
"step": 712
|
| 5026 |
+
},
|
| 5027 |
+
{
|
| 5028 |
+
"epoch": 0.5252302025782689,
|
| 5029 |
+
"grad_norm": 0.3487853407859802,
|
| 5030 |
+
"learning_rate": 0.00018623090278305134,
|
| 5031 |
+
"loss": 0.5313,
|
| 5032 |
+
"step": 713
|
| 5033 |
+
},
|
| 5034 |
+
{
|
| 5035 |
+
"epoch": 0.5259668508287293,
|
| 5036 |
+
"grad_norm": 0.34762606024742126,
|
| 5037 |
+
"learning_rate": 0.000186191509065606,
|
| 5038 |
+
"loss": 0.5679,
|
| 5039 |
+
"step": 714
|
| 5040 |
+
},
|
| 5041 |
+
{
|
| 5042 |
+
"epoch": 0.5267034990791897,
|
| 5043 |
+
"grad_norm": 0.38201674818992615,
|
| 5044 |
+
"learning_rate": 0.00018615206325430433,
|
| 5045 |
+
"loss": 0.5614,
|
| 5046 |
+
"step": 715
|
| 5047 |
+
},
|
| 5048 |
+
{
|
| 5049 |
+
"epoch": 0.52744014732965,
|
| 5050 |
+
"grad_norm": 0.3226095139980316,
|
| 5051 |
+
"learning_rate": 0.00018611256537298738,
|
| 5052 |
+
"loss": 0.5076,
|
| 5053 |
+
"step": 716
|
| 5054 |
+
},
|
| 5055 |
+
{
|
| 5056 |
+
"epoch": 0.5281767955801105,
|
| 5057 |
+
"grad_norm": 0.34292536973953247,
|
| 5058 |
+
"learning_rate": 0.0001860730154455275,
|
| 5059 |
+
"loss": 0.5718,
|
| 5060 |
+
"step": 717
|
| 5061 |
+
},
|
| 5062 |
+
{
|
| 5063 |
+
"epoch": 0.5289134438305709,
|
| 5064 |
+
"grad_norm": 0.3531748652458191,
|
| 5065 |
+
"learning_rate": 0.00018603341349582853,
|
| 5066 |
+
"loss": 0.5445,
|
| 5067 |
+
"step": 718
|
| 5068 |
+
},
|
| 5069 |
+
{
|
| 5070 |
+
"epoch": 0.5296500920810313,
|
| 5071 |
+
"grad_norm": 0.37051910161972046,
|
| 5072 |
+
"learning_rate": 0.00018599375954782575,
|
| 5073 |
+
"loss": 0.5493,
|
| 5074 |
+
"step": 719
|
| 5075 |
+
},
|
| 5076 |
+
{
|
| 5077 |
+
"epoch": 0.5303867403314917,
|
| 5078 |
+
"grad_norm": 0.3461134433746338,
|
| 5079 |
+
"learning_rate": 0.0001859540536254859,
|
| 5080 |
+
"loss": 0.4829,
|
| 5081 |
+
"step": 720
|
| 5082 |
+
},
|
| 5083 |
+
{
|
| 5084 |
+
"epoch": 0.5311233885819521,
|
| 5085 |
+
"grad_norm": 0.3038894534111023,
|
| 5086 |
+
"learning_rate": 0.00018591429575280703,
|
| 5087 |
+
"loss": 0.5114,
|
| 5088 |
+
"step": 721
|
| 5089 |
+
},
|
| 5090 |
+
{
|
| 5091 |
+
"epoch": 0.5318600368324126,
|
| 5092 |
+
"grad_norm": 0.32695645093917847,
|
| 5093 |
+
"learning_rate": 0.00018587448595381876,
|
| 5094 |
+
"loss": 0.4403,
|
| 5095 |
+
"step": 722
|
| 5096 |
+
},
|
| 5097 |
+
{
|
| 5098 |
+
"epoch": 0.532596685082873,
|
| 5099 |
+
"grad_norm": 0.32673871517181396,
|
| 5100 |
+
"learning_rate": 0.0001858346242525819,
|
| 5101 |
+
"loss": 0.4795,
|
| 5102 |
+
"step": 723
|
| 5103 |
+
},
|
| 5104 |
+
{
|
| 5105 |
+
"epoch": 0.5333333333333333,
|
| 5106 |
+
"grad_norm": 0.3549819588661194,
|
| 5107 |
+
"learning_rate": 0.0001857947106731888,
|
| 5108 |
+
"loss": 0.5153,
|
| 5109 |
+
"step": 724
|
| 5110 |
+
},
|
| 5111 |
+
{
|
| 5112 |
+
"epoch": 0.5340699815837937,
|
| 5113 |
+
"grad_norm": 0.30546438694000244,
|
| 5114 |
+
"learning_rate": 0.0001857547452397631,
|
| 5115 |
+
"loss": 0.5103,
|
| 5116 |
+
"step": 725
|
| 5117 |
+
},
|
| 5118 |
+
{
|
| 5119 |
+
"epoch": 0.5348066298342542,
|
| 5120 |
+
"grad_norm": 0.3295595943927765,
|
| 5121 |
+
"learning_rate": 0.00018571472797645973,
|
| 5122 |
+
"loss": 0.5673,
|
| 5123 |
+
"step": 726
|
| 5124 |
+
},
|
| 5125 |
+
{
|
| 5126 |
+
"epoch": 0.5355432780847146,
|
| 5127 |
+
"grad_norm": 0.3666938841342926,
|
| 5128 |
+
"learning_rate": 0.00018567465890746498,
|
| 5129 |
+
"loss": 0.5581,
|
| 5130 |
+
"step": 727
|
| 5131 |
+
},
|
| 5132 |
+
{
|
| 5133 |
+
"epoch": 0.536279926335175,
|
| 5134 |
+
"grad_norm": 0.31538668274879456,
|
| 5135 |
+
"learning_rate": 0.0001856345380569965,
|
| 5136 |
+
"loss": 0.4924,
|
| 5137 |
+
"step": 728
|
| 5138 |
+
},
|
| 5139 |
+
{
|
| 5140 |
+
"epoch": 0.5370165745856353,
|
| 5141 |
+
"grad_norm": 0.362741619348526,
|
| 5142 |
+
"learning_rate": 0.00018559436544930317,
|
| 5143 |
+
"loss": 0.5471,
|
| 5144 |
+
"step": 729
|
| 5145 |
+
},
|
| 5146 |
+
{
|
| 5147 |
+
"epoch": 0.5377532228360957,
|
| 5148 |
+
"grad_norm": 0.3251877725124359,
|
| 5149 |
+
"learning_rate": 0.0001855541411086652,
|
| 5150 |
+
"loss": 0.5362,
|
| 5151 |
+
"step": 730
|
| 5152 |
+
},
|
| 5153 |
+
{
|
| 5154 |
+
"epoch": 0.5384898710865562,
|
| 5155 |
+
"grad_norm": 0.34566330909729004,
|
| 5156 |
+
"learning_rate": 0.00018551386505939401,
|
| 5157 |
+
"loss": 0.4682,
|
| 5158 |
+
"step": 731
|
| 5159 |
+
},
|
| 5160 |
+
{
|
| 5161 |
+
"epoch": 0.5392265193370166,
|
| 5162 |
+
"grad_norm": 0.333143025636673,
|
| 5163 |
+
"learning_rate": 0.00018547353732583234,
|
| 5164 |
+
"loss": 0.4079,
|
| 5165 |
+
"step": 732
|
| 5166 |
+
},
|
| 5167 |
+
{
|
| 5168 |
+
"epoch": 0.539963167587477,
|
| 5169 |
+
"grad_norm": 0.29367437958717346,
|
| 5170 |
+
"learning_rate": 0.0001854331579323541,
|
| 5171 |
+
"loss": 0.4531,
|
| 5172 |
+
"step": 733
|
| 5173 |
+
},
|
| 5174 |
+
{
|
| 5175 |
+
"epoch": 0.5406998158379374,
|
| 5176 |
+
"grad_norm": 0.37323516607284546,
|
| 5177 |
+
"learning_rate": 0.00018539272690336453,
|
| 5178 |
+
"loss": 0.5318,
|
| 5179 |
+
"step": 734
|
| 5180 |
+
},
|
| 5181 |
+
{
|
| 5182 |
+
"epoch": 0.5414364640883977,
|
| 5183 |
+
"grad_norm": 0.32561689615249634,
|
| 5184 |
+
"learning_rate": 0.00018535224426329992,
|
| 5185 |
+
"loss": 0.5257,
|
| 5186 |
+
"step": 735
|
| 5187 |
+
},
|
| 5188 |
+
{
|
| 5189 |
+
"epoch": 0.5421731123388582,
|
| 5190 |
+
"grad_norm": 0.3341492712497711,
|
| 5191 |
+
"learning_rate": 0.0001853117100366279,
|
| 5192 |
+
"loss": 0.5465,
|
| 5193 |
+
"step": 736
|
| 5194 |
+
},
|
| 5195 |
+
{
|
| 5196 |
+
"epoch": 0.5429097605893186,
|
| 5197 |
+
"grad_norm": 0.30662253499031067,
|
| 5198 |
+
"learning_rate": 0.0001852711242478472,
|
| 5199 |
+
"loss": 0.5225,
|
| 5200 |
+
"step": 737
|
| 5201 |
+
},
|
| 5202 |
+
{
|
| 5203 |
+
"epoch": 0.543646408839779,
|
| 5204 |
+
"grad_norm": 0.34401625394821167,
|
| 5205 |
+
"learning_rate": 0.00018523048692148774,
|
| 5206 |
+
"loss": 0.5564,
|
| 5207 |
+
"step": 738
|
| 5208 |
+
},
|
| 5209 |
+
{
|
| 5210 |
+
"epoch": 0.5443830570902394,
|
| 5211 |
+
"grad_norm": 0.3269616365432739,
|
| 5212 |
+
"learning_rate": 0.00018518979808211054,
|
| 5213 |
+
"loss": 0.5546,
|
| 5214 |
+
"step": 739
|
| 5215 |
+
},
|
| 5216 |
+
{
|
| 5217 |
+
"epoch": 0.5451197053406999,
|
| 5218 |
+
"grad_norm": 0.3114912807941437,
|
| 5219 |
+
"learning_rate": 0.00018514905775430786,
|
| 5220 |
+
"loss": 0.5057,
|
| 5221 |
+
"step": 740
|
| 5222 |
+
},
|
| 5223 |
+
{
|
| 5224 |
+
"epoch": 0.5458563535911602,
|
| 5225 |
+
"grad_norm": 0.3612748384475708,
|
| 5226 |
+
"learning_rate": 0.000185108265962703,
|
| 5227 |
+
"loss": 0.5308,
|
| 5228 |
+
"step": 741
|
| 5229 |
+
},
|
| 5230 |
+
{
|
| 5231 |
+
"epoch": 0.5465930018416206,
|
| 5232 |
+
"grad_norm": 0.3298930525779724,
|
| 5233 |
+
"learning_rate": 0.00018506742273195035,
|
| 5234 |
+
"loss": 0.5606,
|
| 5235 |
+
"step": 742
|
| 5236 |
+
},
|
| 5237 |
+
{
|
| 5238 |
+
"epoch": 0.547329650092081,
|
| 5239 |
+
"grad_norm": 0.29278379678726196,
|
| 5240 |
+
"learning_rate": 0.00018502652808673545,
|
| 5241 |
+
"loss": 0.4259,
|
| 5242 |
+
"step": 743
|
| 5243 |
+
},
|
| 5244 |
+
{
|
| 5245 |
+
"epoch": 0.5480662983425414,
|
| 5246 |
+
"grad_norm": 0.2903897762298584,
|
| 5247 |
+
"learning_rate": 0.00018498558205177487,
|
| 5248 |
+
"loss": 0.4847,
|
| 5249 |
+
"step": 744
|
| 5250 |
+
},
|
| 5251 |
+
{
|
| 5252 |
+
"epoch": 0.5488029465930019,
|
| 5253 |
+
"grad_norm": 0.3876440227031708,
|
| 5254 |
+
"learning_rate": 0.00018494458465181623,
|
| 5255 |
+
"loss": 0.578,
|
| 5256 |
+
"step": 745
|
| 5257 |
+
},
|
| 5258 |
+
{
|
| 5259 |
+
"epoch": 0.5495395948434623,
|
| 5260 |
+
"grad_norm": 0.30334675312042236,
|
| 5261 |
+
"learning_rate": 0.0001849035359116383,
|
| 5262 |
+
"loss": 0.4999,
|
| 5263 |
+
"step": 746
|
| 5264 |
+
},
|
| 5265 |
+
{
|
| 5266 |
+
"epoch": 0.5502762430939226,
|
| 5267 |
+
"grad_norm": 0.3237926959991455,
|
| 5268 |
+
"learning_rate": 0.0001848624358560507,
|
| 5269 |
+
"loss": 0.6113,
|
| 5270 |
+
"step": 747
|
| 5271 |
+
},
|
| 5272 |
+
{
|
| 5273 |
+
"epoch": 0.551012891344383,
|
| 5274 |
+
"grad_norm": 0.3559538424015045,
|
| 5275 |
+
"learning_rate": 0.00018482128450989427,
|
| 5276 |
+
"loss": 0.5203,
|
| 5277 |
+
"step": 748
|
| 5278 |
+
},
|
| 5279 |
+
{
|
| 5280 |
+
"epoch": 0.5517495395948434,
|
| 5281 |
+
"grad_norm": 0.29950273036956787,
|
| 5282 |
+
"learning_rate": 0.00018478008189804063,
|
| 5283 |
+
"loss": 0.4669,
|
| 5284 |
+
"step": 749
|
| 5285 |
+
},
|
| 5286 |
+
{
|
| 5287 |
+
"epoch": 0.5524861878453039,
|
| 5288 |
+
"grad_norm": 0.3407313823699951,
|
| 5289 |
+
"learning_rate": 0.0001847388280453926,
|
| 5290 |
+
"loss": 0.5565,
|
| 5291 |
+
"step": 750
|
| 5292 |
+
},
|
| 5293 |
+
{
|
| 5294 |
+
"epoch": 0.5532228360957643,
|
| 5295 |
+
"grad_norm": 0.3306919038295746,
|
| 5296 |
+
"learning_rate": 0.0001846975229768838,
|
| 5297 |
+
"loss": 0.5453,
|
| 5298 |
+
"step": 751
|
| 5299 |
+
},
|
| 5300 |
+
{
|
| 5301 |
+
"epoch": 0.5539594843462247,
|
| 5302 |
+
"grad_norm": 0.34809431433677673,
|
| 5303 |
+
"learning_rate": 0.0001846561667174789,
|
| 5304 |
+
"loss": 0.4539,
|
| 5305 |
+
"step": 752
|
| 5306 |
+
},
|
| 5307 |
+
{
|
| 5308 |
+
"epoch": 0.554696132596685,
|
| 5309 |
+
"grad_norm": 0.32188156247138977,
|
| 5310 |
+
"learning_rate": 0.0001846147592921735,
|
| 5311 |
+
"loss": 0.5743,
|
| 5312 |
+
"step": 753
|
| 5313 |
+
},
|
| 5314 |
+
{
|
| 5315 |
+
"epoch": 0.5554327808471455,
|
| 5316 |
+
"grad_norm": 0.37432029843330383,
|
| 5317 |
+
"learning_rate": 0.00018457330072599407,
|
| 5318 |
+
"loss": 0.5594,
|
| 5319 |
+
"step": 754
|
| 5320 |
+
},
|
| 5321 |
+
{
|
| 5322 |
+
"epoch": 0.5561694290976059,
|
| 5323 |
+
"grad_norm": 0.2952982187271118,
|
| 5324 |
+
"learning_rate": 0.00018453179104399806,
|
| 5325 |
+
"loss": 0.4873,
|
| 5326 |
+
"step": 755
|
| 5327 |
+
},
|
| 5328 |
+
{
|
| 5329 |
+
"epoch": 0.5569060773480663,
|
| 5330 |
+
"grad_norm": 0.3097144365310669,
|
| 5331 |
+
"learning_rate": 0.0001844902302712738,
|
| 5332 |
+
"loss": 0.4535,
|
| 5333 |
+
"step": 756
|
| 5334 |
+
},
|
| 5335 |
+
{
|
| 5336 |
+
"epoch": 0.5576427255985267,
|
| 5337 |
+
"grad_norm": 0.36936330795288086,
|
| 5338 |
+
"learning_rate": 0.0001844486184329404,
|
| 5339 |
+
"loss": 0.5327,
|
| 5340 |
+
"step": 757
|
| 5341 |
+
},
|
| 5342 |
+
{
|
| 5343 |
+
"epoch": 0.5583793738489871,
|
| 5344 |
+
"grad_norm": 0.34994593262672424,
|
| 5345 |
+
"learning_rate": 0.00018440695555414798,
|
| 5346 |
+
"loss": 0.5562,
|
| 5347 |
+
"step": 758
|
| 5348 |
+
},
|
| 5349 |
+
{
|
| 5350 |
+
"epoch": 0.5591160220994476,
|
| 5351 |
+
"grad_norm": 0.33190274238586426,
|
| 5352 |
+
"learning_rate": 0.0001843652416600774,
|
| 5353 |
+
"loss": 0.5494,
|
| 5354 |
+
"step": 759
|
| 5355 |
+
},
|
| 5356 |
+
{
|
| 5357 |
+
"epoch": 0.5598526703499079,
|
| 5358 |
+
"grad_norm": 0.31601038575172424,
|
| 5359 |
+
"learning_rate": 0.00018432347677594044,
|
| 5360 |
+
"loss": 0.4593,
|
| 5361 |
+
"step": 760
|
| 5362 |
+
},
|
| 5363 |
+
{
|
| 5364 |
+
"epoch": 0.5605893186003683,
|
| 5365 |
+
"grad_norm": 0.34895774722099304,
|
| 5366 |
+
"learning_rate": 0.00018428166092697966,
|
| 5367 |
+
"loss": 0.488,
|
| 5368 |
+
"step": 761
|
| 5369 |
+
},
|
| 5370 |
+
{
|
| 5371 |
+
"epoch": 0.5613259668508287,
|
| 5372 |
+
"grad_norm": 0.31067827343940735,
|
| 5373 |
+
"learning_rate": 0.00018423979413846836,
|
| 5374 |
+
"loss": 0.484,
|
| 5375 |
+
"step": 762
|
| 5376 |
+
},
|
| 5377 |
+
{
|
| 5378 |
+
"epoch": 0.5620626151012892,
|
| 5379 |
+
"grad_norm": 0.3028845191001892,
|
| 5380 |
+
"learning_rate": 0.00018419787643571075,
|
| 5381 |
+
"loss": 0.443,
|
| 5382 |
+
"step": 763
|
| 5383 |
+
},
|
| 5384 |
+
{
|
| 5385 |
+
"epoch": 0.5627992633517496,
|
| 5386 |
+
"grad_norm": 0.33658021688461304,
|
| 5387 |
+
"learning_rate": 0.0001841559078440417,
|
| 5388 |
+
"loss": 0.575,
|
| 5389 |
+
"step": 764
|
| 5390 |
+
},
|
| 5391 |
+
{
|
| 5392 |
+
"epoch": 0.56353591160221,
|
| 5393 |
+
"grad_norm": 0.3199828863143921,
|
| 5394 |
+
"learning_rate": 0.00018411388838882688,
|
| 5395 |
+
"loss": 0.4989,
|
| 5396 |
+
"step": 765
|
| 5397 |
+
},
|
| 5398 |
+
{
|
| 5399 |
+
"epoch": 0.5642725598526703,
|
| 5400 |
+
"grad_norm": 0.35100269317626953,
|
| 5401 |
+
"learning_rate": 0.00018407181809546277,
|
| 5402 |
+
"loss": 0.5856,
|
| 5403 |
+
"step": 766
|
| 5404 |
+
},
|
| 5405 |
+
{
|
| 5406 |
+
"epoch": 0.5650092081031307,
|
| 5407 |
+
"grad_norm": 0.3549753725528717,
|
| 5408 |
+
"learning_rate": 0.00018402969698937645,
|
| 5409 |
+
"loss": 0.5744,
|
| 5410 |
+
"step": 767
|
| 5411 |
+
},
|
| 5412 |
+
{
|
| 5413 |
+
"epoch": 0.5657458563535912,
|
| 5414 |
+
"grad_norm": 0.3407423794269562,
|
| 5415 |
+
"learning_rate": 0.00018398752509602578,
|
| 5416 |
+
"loss": 0.4986,
|
| 5417 |
+
"step": 768
|
| 5418 |
+
},
|
| 5419 |
+
{
|
| 5420 |
+
"epoch": 0.5664825046040516,
|
| 5421 |
+
"grad_norm": 0.35427817702293396,
|
| 5422 |
+
"learning_rate": 0.00018394530244089937,
|
| 5423 |
+
"loss": 0.4373,
|
| 5424 |
+
"step": 769
|
| 5425 |
+
},
|
| 5426 |
+
{
|
| 5427 |
+
"epoch": 0.567219152854512,
|
| 5428 |
+
"grad_norm": 0.3336740732192993,
|
| 5429 |
+
"learning_rate": 0.00018390302904951636,
|
| 5430 |
+
"loss": 0.5063,
|
| 5431 |
+
"step": 770
|
| 5432 |
+
},
|
| 5433 |
+
{
|
| 5434 |
+
"epoch": 0.5679558011049723,
|
| 5435 |
+
"grad_norm": 0.34777185320854187,
|
| 5436 |
+
"learning_rate": 0.0001838607049474267,
|
| 5437 |
+
"loss": 0.4951,
|
| 5438 |
+
"step": 771
|
| 5439 |
+
},
|
| 5440 |
+
{
|
| 5441 |
+
"epoch": 0.5686924493554327,
|
| 5442 |
+
"grad_norm": 0.3432445228099823,
|
| 5443 |
+
"learning_rate": 0.00018381833016021092,
|
| 5444 |
+
"loss": 0.5407,
|
| 5445 |
+
"step": 772
|
| 5446 |
+
},
|
| 5447 |
+
{
|
| 5448 |
+
"epoch": 0.5694290976058932,
|
| 5449 |
+
"grad_norm": 0.3722617030143738,
|
| 5450 |
+
"learning_rate": 0.00018377590471348022,
|
| 5451 |
+
"loss": 0.5542,
|
| 5452 |
+
"step": 773
|
| 5453 |
+
},
|
| 5454 |
+
{
|
| 5455 |
+
"epoch": 0.5701657458563536,
|
| 5456 |
+
"grad_norm": 0.3160288631916046,
|
| 5457 |
+
"learning_rate": 0.00018373342863287637,
|
| 5458 |
+
"loss": 0.5267,
|
| 5459 |
+
"step": 774
|
| 5460 |
+
},
|
| 5461 |
+
{
|
| 5462 |
+
"epoch": 0.570902394106814,
|
| 5463 |
+
"grad_norm": 0.336774617433548,
|
| 5464 |
+
"learning_rate": 0.00018369090194407175,
|
| 5465 |
+
"loss": 0.5109,
|
| 5466 |
+
"step": 775
|
| 5467 |
+
},
|
| 5468 |
+
{
|
| 5469 |
+
"epoch": 0.5716390423572744,
|
| 5470 |
+
"grad_norm": 0.3267696499824524,
|
| 5471 |
+
"learning_rate": 0.0001836483246727694,
|
| 5472 |
+
"loss": 0.5216,
|
| 5473 |
+
"step": 776
|
| 5474 |
+
},
|
| 5475 |
+
{
|
| 5476 |
+
"epoch": 0.5723756906077349,
|
| 5477 |
+
"grad_norm": 0.3363187313079834,
|
| 5478 |
+
"learning_rate": 0.00018360569684470286,
|
| 5479 |
+
"loss": 0.4185,
|
| 5480 |
+
"step": 777
|
| 5481 |
+
},
|
| 5482 |
+
{
|
| 5483 |
+
"epoch": 0.5731123388581952,
|
| 5484 |
+
"grad_norm": 0.31178250908851624,
|
| 5485 |
+
"learning_rate": 0.0001835630184856362,
|
| 5486 |
+
"loss": 0.5133,
|
| 5487 |
+
"step": 778
|
| 5488 |
+
},
|
| 5489 |
+
{
|
| 5490 |
+
"epoch": 0.5738489871086556,
|
| 5491 |
+
"grad_norm": 0.3427146375179291,
|
| 5492 |
+
"learning_rate": 0.00018352028962136414,
|
| 5493 |
+
"loss": 0.5549,
|
| 5494 |
+
"step": 779
|
| 5495 |
+
},
|
| 5496 |
+
{
|
| 5497 |
+
"epoch": 0.574585635359116,
|
| 5498 |
+
"grad_norm": 0.3670169413089752,
|
| 5499 |
+
"learning_rate": 0.00018347751027771183,
|
| 5500 |
+
"loss": 0.5594,
|
| 5501 |
+
"step": 780
|
| 5502 |
+
},
|
| 5503 |
+
{
|
| 5504 |
+
"epoch": 0.5753222836095764,
|
| 5505 |
+
"grad_norm": 0.31756749749183655,
|
| 5506 |
+
"learning_rate": 0.00018343468048053496,
|
| 5507 |
+
"loss": 0.4551,
|
| 5508 |
+
"step": 781
|
| 5509 |
+
},
|
| 5510 |
+
{
|
| 5511 |
+
"epoch": 0.5760589318600369,
|
| 5512 |
+
"grad_norm": 0.3160708546638489,
|
| 5513 |
+
"learning_rate": 0.00018339180025571972,
|
| 5514 |
+
"loss": 0.509,
|
| 5515 |
+
"step": 782
|
| 5516 |
+
},
|
| 5517 |
+
{
|
| 5518 |
+
"epoch": 0.5767955801104973,
|
| 5519 |
+
"grad_norm": 0.31845369935035706,
|
| 5520 |
+
"learning_rate": 0.00018334886962918275,
|
| 5521 |
+
"loss": 0.5067,
|
| 5522 |
+
"step": 783
|
| 5523 |
+
},
|
| 5524 |
+
{
|
| 5525 |
+
"epoch": 0.5775322283609576,
|
| 5526 |
+
"grad_norm": 0.333650141954422,
|
| 5527 |
+
"learning_rate": 0.0001833058886268712,
|
| 5528 |
+
"loss": 0.4835,
|
| 5529 |
+
"step": 784
|
| 5530 |
+
},
|
| 5531 |
+
{
|
| 5532 |
+
"epoch": 0.578268876611418,
|
| 5533 |
+
"grad_norm": 0.32727667689323425,
|
| 5534 |
+
"learning_rate": 0.0001832628572747627,
|
| 5535 |
+
"loss": 0.4618,
|
| 5536 |
+
"step": 785
|
| 5537 |
+
},
|
| 5538 |
+
{
|
| 5539 |
+
"epoch": 0.5790055248618785,
|
| 5540 |
+
"grad_norm": 0.3380400538444519,
|
| 5541 |
+
"learning_rate": 0.00018321977559886512,
|
| 5542 |
+
"loss": 0.4818,
|
| 5543 |
+
"step": 786
|
| 5544 |
+
},
|
| 5545 |
+
{
|
| 5546 |
+
"epoch": 0.5797421731123389,
|
| 5547 |
+
"grad_norm": 0.35377421975135803,
|
| 5548 |
+
"learning_rate": 0.00018317664362521696,
|
| 5549 |
+
"loss": 0.4374,
|
| 5550 |
+
"step": 787
|
| 5551 |
+
},
|
| 5552 |
+
{
|
| 5553 |
+
"epoch": 0.5804788213627993,
|
| 5554 |
+
"grad_norm": 0.31174394488334656,
|
| 5555 |
+
"learning_rate": 0.00018313346137988705,
|
| 5556 |
+
"loss": 0.5116,
|
| 5557 |
+
"step": 788
|
| 5558 |
+
},
|
| 5559 |
+
{
|
| 5560 |
+
"epoch": 0.5812154696132596,
|
| 5561 |
+
"grad_norm": 0.33696267008781433,
|
| 5562 |
+
"learning_rate": 0.00018309022888897457,
|
| 5563 |
+
"loss": 0.5162,
|
| 5564 |
+
"step": 789
|
| 5565 |
+
},
|
| 5566 |
+
{
|
| 5567 |
+
"epoch": 0.58195211786372,
|
| 5568 |
+
"grad_norm": 0.42592304944992065,
|
| 5569 |
+
"learning_rate": 0.0001830469461786091,
|
| 5570 |
+
"loss": 0.5078,
|
| 5571 |
+
"step": 790
|
| 5572 |
+
},
|
| 5573 |
+
{
|
| 5574 |
+
"epoch": 0.5826887661141805,
|
| 5575 |
+
"grad_norm": 0.339857816696167,
|
| 5576 |
+
"learning_rate": 0.00018300361327495054,
|
| 5577 |
+
"loss": 0.5453,
|
| 5578 |
+
"step": 791
|
| 5579 |
+
},
|
| 5580 |
+
{
|
| 5581 |
+
"epoch": 0.5834254143646409,
|
| 5582 |
+
"grad_norm": 0.32044872641563416,
|
| 5583 |
+
"learning_rate": 0.00018296023020418918,
|
| 5584 |
+
"loss": 0.4915,
|
| 5585 |
+
"step": 792
|
| 5586 |
+
},
|
| 5587 |
+
{
|
| 5588 |
+
"epoch": 0.5841620626151013,
|
| 5589 |
+
"grad_norm": 0.3149367570877075,
|
| 5590 |
+
"learning_rate": 0.0001829167969925456,
|
| 5591 |
+
"loss": 0.4404,
|
| 5592 |
+
"step": 793
|
| 5593 |
+
},
|
| 5594 |
+
{
|
| 5595 |
+
"epoch": 0.5848987108655617,
|
| 5596 |
+
"grad_norm": 0.3723057210445404,
|
| 5597 |
+
"learning_rate": 0.00018287331366627062,
|
| 5598 |
+
"loss": 0.5716,
|
| 5599 |
+
"step": 794
|
| 5600 |
+
},
|
| 5601 |
+
{
|
| 5602 |
+
"epoch": 0.585635359116022,
|
| 5603 |
+
"grad_norm": 0.40884530544281006,
|
| 5604 |
+
"learning_rate": 0.00018282978025164551,
|
| 5605 |
+
"loss": 0.4867,
|
| 5606 |
+
"step": 795
|
| 5607 |
+
},
|
| 5608 |
+
{
|
| 5609 |
+
"epoch": 0.5863720073664825,
|
| 5610 |
+
"grad_norm": 0.3199060559272766,
|
| 5611 |
+
"learning_rate": 0.00018278619677498167,
|
| 5612 |
+
"loss": 0.486,
|
| 5613 |
+
"step": 796
|
| 5614 |
+
},
|
| 5615 |
+
{
|
| 5616 |
+
"epoch": 0.5871086556169429,
|
| 5617 |
+
"grad_norm": 0.33268362283706665,
|
| 5618 |
+
"learning_rate": 0.00018274256326262082,
|
| 5619 |
+
"loss": 0.5297,
|
| 5620 |
+
"step": 797
|
| 5621 |
+
},
|
| 5622 |
+
{
|
| 5623 |
+
"epoch": 0.5878453038674033,
|
| 5624 |
+
"grad_norm": 0.3343597948551178,
|
| 5625 |
+
"learning_rate": 0.0001826988797409349,
|
| 5626 |
+
"loss": 0.563,
|
| 5627 |
+
"step": 798
|
| 5628 |
+
},
|
| 5629 |
+
{
|
| 5630 |
+
"epoch": 0.5885819521178637,
|
| 5631 |
+
"grad_norm": 0.3317117691040039,
|
| 5632 |
+
"learning_rate": 0.0001826551462363261,
|
| 5633 |
+
"loss": 0.5536,
|
| 5634 |
+
"step": 799
|
| 5635 |
+
},
|
| 5636 |
+
{
|
| 5637 |
+
"epoch": 0.5893186003683242,
|
| 5638 |
+
"grad_norm": 0.3641378879547119,
|
| 5639 |
+
"learning_rate": 0.00018261136277522677,
|
| 5640 |
+
"loss": 0.4998,
|
| 5641 |
+
"step": 800
|
| 5642 |
+
},
|
| 5643 |
+
{
|
| 5644 |
+
"epoch": 0.5893186003683242,
|
| 5645 |
+
"eval_loss": 0.5467618703842163,
|
| 5646 |
+
"eval_runtime": 6.8864,
|
| 5647 |
+
"eval_samples_per_second": 31.802,
|
| 5648 |
+
"eval_steps_per_second": 2.033,
|
| 5649 |
+
"step": 800
|
| 5650 |
}
|
| 5651 |
],
|
| 5652 |
"logging_steps": 1,
|
|
|
|
| 5675 |
"attributes": {}
|
| 5676 |
}
|
| 5677 |
},
|
| 5678 |
+
"total_flos": 8.065908019612877e+16,
|
| 5679 |
"train_batch_size": 16,
|
| 5680 |
"trial_name": null,
|
| 5681 |
"trial_params": null
|