Training in progress, step 1044, 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 83945296
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d691bdf43b3bd1e09cb5c98703d7b30cb1bd022703d7d074533325df99594ba1
|
3 |
size 83945296
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 43123028
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f87ea960c8416520eeddfa88984d166ed0ec6a6f9e50ff1e2249678eb646b81f
|
3 |
size 43123028
|
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 sha256:362bf5ea5581bfd88a319d55517e06e0c66f6a7250dd88328201d694417113e1
|
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:ac730905d01cb821bce3ec3eb28c0bf93a86a906be4c6ab47513122b4f73e726
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
5 |
"eval_steps": 348,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -4903,6 +4903,2450 @@
|
|
4903 |
"eval_samples_per_second": 27.246,
|
4904 |
"eval_steps_per_second": 13.623,
|
4905 |
"step": 696
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4906 |
}
|
4907 |
],
|
4908 |
"logging_steps": 1,
|
@@ -4922,7 +7366,7 @@
|
|
4922 |
"attributes": {}
|
4923 |
}
|
4924 |
},
|
4925 |
-
"total_flos":
|
4926 |
"train_batch_size": 2,
|
4927 |
"trial_name": null,
|
4928 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.20472595352485537,
|
5 |
"eval_steps": 348,
|
6 |
+
"global_step": 1044,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
4903 |
"eval_samples_per_second": 27.246,
|
4904 |
"eval_steps_per_second": 13.623,
|
4905 |
"step": 696
|
4906 |
+
},
|
4907 |
+
{
|
4908 |
+
"epoch": 0.13668006667320326,
|
4909 |
+
"grad_norm": 3.3614909648895264,
|
4910 |
+
"learning_rate": 0.00010056953873626384,
|
4911 |
+
"loss": 0.7098,
|
4912 |
+
"step": 697
|
4913 |
+
},
|
4914 |
+
{
|
4915 |
+
"epoch": 0.13687616432983626,
|
4916 |
+
"grad_norm": 6.330589294433594,
|
4917 |
+
"learning_rate": 0.00010034172442413501,
|
4918 |
+
"loss": 1.97,
|
4919 |
+
"step": 698
|
4920 |
+
},
|
4921 |
+
{
|
4922 |
+
"epoch": 0.13707226198646927,
|
4923 |
+
"grad_norm": 6.881412506103516,
|
4924 |
+
"learning_rate": 0.00010011390833844143,
|
4925 |
+
"loss": 2.2338,
|
4926 |
+
"step": 699
|
4927 |
+
},
|
4928 |
+
{
|
4929 |
+
"epoch": 0.13726835964310227,
|
4930 |
+
"grad_norm": 6.3669114112854,
|
4931 |
+
"learning_rate": 9.988609166155859e-05,
|
4932 |
+
"loss": 1.2176,
|
4933 |
+
"step": 700
|
4934 |
+
},
|
4935 |
+
{
|
4936 |
+
"epoch": 0.13746445729973528,
|
4937 |
+
"grad_norm": 4.456204414367676,
|
4938 |
+
"learning_rate": 9.965827557586498e-05,
|
4939 |
+
"loss": 2.5429,
|
4940 |
+
"step": 701
|
4941 |
+
},
|
4942 |
+
{
|
4943 |
+
"epoch": 0.13766055495636828,
|
4944 |
+
"grad_norm": 11.206175804138184,
|
4945 |
+
"learning_rate": 9.943046126373618e-05,
|
4946 |
+
"loss": 0.9134,
|
4947 |
+
"step": 702
|
4948 |
+
},
|
4949 |
+
{
|
4950 |
+
"epoch": 0.13785665261300128,
|
4951 |
+
"grad_norm": 3.9151828289031982,
|
4952 |
+
"learning_rate": 9.920264990753837e-05,
|
4953 |
+
"loss": 0.8662,
|
4954 |
+
"step": 703
|
4955 |
+
},
|
4956 |
+
{
|
4957 |
+
"epoch": 0.1380527502696343,
|
4958 |
+
"grad_norm": 9.173954963684082,
|
4959 |
+
"learning_rate": 9.897484268962243e-05,
|
4960 |
+
"loss": 1.7328,
|
4961 |
+
"step": 704
|
4962 |
+
},
|
4963 |
+
{
|
4964 |
+
"epoch": 0.1382488479262673,
|
4965 |
+
"grad_norm": 9.788105010986328,
|
4966 |
+
"learning_rate": 9.874704079231791e-05,
|
4967 |
+
"loss": 1.5085,
|
4968 |
+
"step": 705
|
4969 |
+
},
|
4970 |
+
{
|
4971 |
+
"epoch": 0.1384449455829003,
|
4972 |
+
"grad_norm": 4.705010414123535,
|
4973 |
+
"learning_rate": 9.851924539792656e-05,
|
4974 |
+
"loss": 1.7789,
|
4975 |
+
"step": 706
|
4976 |
+
},
|
4977 |
+
{
|
4978 |
+
"epoch": 0.1386410432395333,
|
4979 |
+
"grad_norm": 9.228858947753906,
|
4980 |
+
"learning_rate": 9.82914576887165e-05,
|
4981 |
+
"loss": 1.3564,
|
4982 |
+
"step": 707
|
4983 |
+
},
|
4984 |
+
{
|
4985 |
+
"epoch": 0.13883714089616628,
|
4986 |
+
"grad_norm": 13.22546100616455,
|
4987 |
+
"learning_rate": 9.806367884691594e-05,
|
4988 |
+
"loss": 1.752,
|
4989 |
+
"step": 708
|
4990 |
+
},
|
4991 |
+
{
|
4992 |
+
"epoch": 0.13903323855279928,
|
4993 |
+
"grad_norm": 8.163661003112793,
|
4994 |
+
"learning_rate": 9.783591005470698e-05,
|
4995 |
+
"loss": 1.3811,
|
4996 |
+
"step": 709
|
4997 |
+
},
|
4998 |
+
{
|
4999 |
+
"epoch": 0.1392293362094323,
|
5000 |
+
"grad_norm": 8.644879341125488,
|
5001 |
+
"learning_rate": 9.760815249421973e-05,
|
5002 |
+
"loss": 1.8551,
|
5003 |
+
"step": 710
|
5004 |
+
},
|
5005 |
+
{
|
5006 |
+
"epoch": 0.1394254338660653,
|
5007 |
+
"grad_norm": 4.906625270843506,
|
5008 |
+
"learning_rate": 9.738040734752582e-05,
|
5009 |
+
"loss": 1.3423,
|
5010 |
+
"step": 711
|
5011 |
+
},
|
5012 |
+
{
|
5013 |
+
"epoch": 0.1396215315226983,
|
5014 |
+
"grad_norm": 7.070640563964844,
|
5015 |
+
"learning_rate": 9.715267579663262e-05,
|
5016 |
+
"loss": 1.1465,
|
5017 |
+
"step": 712
|
5018 |
+
},
|
5019 |
+
{
|
5020 |
+
"epoch": 0.1398176291793313,
|
5021 |
+
"grad_norm": 4.564503192901611,
|
5022 |
+
"learning_rate": 9.692495902347678e-05,
|
5023 |
+
"loss": 1.167,
|
5024 |
+
"step": 713
|
5025 |
+
},
|
5026 |
+
{
|
5027 |
+
"epoch": 0.1400137268359643,
|
5028 |
+
"grad_norm": 6.235091686248779,
|
5029 |
+
"learning_rate": 9.669725820991841e-05,
|
5030 |
+
"loss": 1.0448,
|
5031 |
+
"step": 714
|
5032 |
+
},
|
5033 |
+
{
|
5034 |
+
"epoch": 0.1402098244925973,
|
5035 |
+
"grad_norm": 6.636250019073486,
|
5036 |
+
"learning_rate": 9.646957453773469e-05,
|
5037 |
+
"loss": 1.1572,
|
5038 |
+
"step": 715
|
5039 |
+
},
|
5040 |
+
{
|
5041 |
+
"epoch": 0.1404059221492303,
|
5042 |
+
"grad_norm": 4.873953342437744,
|
5043 |
+
"learning_rate": 9.62419091886138e-05,
|
5044 |
+
"loss": 1.6995,
|
5045 |
+
"step": 716
|
5046 |
+
},
|
5047 |
+
{
|
5048 |
+
"epoch": 0.14060201980586332,
|
5049 |
+
"grad_norm": 7.3861589431762695,
|
5050 |
+
"learning_rate": 9.601426334414898e-05,
|
5051 |
+
"loss": 1.6366,
|
5052 |
+
"step": 717
|
5053 |
+
},
|
5054 |
+
{
|
5055 |
+
"epoch": 0.14079811746249632,
|
5056 |
+
"grad_norm": 4.689702033996582,
|
5057 |
+
"learning_rate": 9.578663818583203e-05,
|
5058 |
+
"loss": 0.6136,
|
5059 |
+
"step": 718
|
5060 |
+
},
|
5061 |
+
{
|
5062 |
+
"epoch": 0.14099421511912932,
|
5063 |
+
"grad_norm": 6.917242527008057,
|
5064 |
+
"learning_rate": 9.555903489504761e-05,
|
5065 |
+
"loss": 1.985,
|
5066 |
+
"step": 719
|
5067 |
+
},
|
5068 |
+
{
|
5069 |
+
"epoch": 0.14119031277576233,
|
5070 |
+
"grad_norm": 11.42885971069336,
|
5071 |
+
"learning_rate": 9.533145465306667e-05,
|
5072 |
+
"loss": 1.2503,
|
5073 |
+
"step": 720
|
5074 |
+
},
|
5075 |
+
{
|
5076 |
+
"epoch": 0.14138641043239533,
|
5077 |
+
"grad_norm": 6.561046600341797,
|
5078 |
+
"learning_rate": 9.510389864104069e-05,
|
5079 |
+
"loss": 1.207,
|
5080 |
+
"step": 721
|
5081 |
+
},
|
5082 |
+
{
|
5083 |
+
"epoch": 0.14158250808902834,
|
5084 |
+
"grad_norm": 5.668408393859863,
|
5085 |
+
"learning_rate": 9.487636803999538e-05,
|
5086 |
+
"loss": 1.5323,
|
5087 |
+
"step": 722
|
5088 |
+
},
|
5089 |
+
{
|
5090 |
+
"epoch": 0.14177860574566134,
|
5091 |
+
"grad_norm": 5.94567346572876,
|
5092 |
+
"learning_rate": 9.464886403082445e-05,
|
5093 |
+
"loss": 1.6155,
|
5094 |
+
"step": 723
|
5095 |
+
},
|
5096 |
+
{
|
5097 |
+
"epoch": 0.14197470340229434,
|
5098 |
+
"grad_norm": 5.67940092086792,
|
5099 |
+
"learning_rate": 9.442138779428376e-05,
|
5100 |
+
"loss": 1.1803,
|
5101 |
+
"step": 724
|
5102 |
+
},
|
5103 |
+
{
|
5104 |
+
"epoch": 0.14217080105892735,
|
5105 |
+
"grad_norm": 5.852919101715088,
|
5106 |
+
"learning_rate": 9.419394051098489e-05,
|
5107 |
+
"loss": 2.2628,
|
5108 |
+
"step": 725
|
5109 |
+
},
|
5110 |
+
{
|
5111 |
+
"epoch": 0.14236689871556035,
|
5112 |
+
"grad_norm": 5.615955352783203,
|
5113 |
+
"learning_rate": 9.396652336138922e-05,
|
5114 |
+
"loss": 2.6917,
|
5115 |
+
"step": 726
|
5116 |
+
},
|
5117 |
+
{
|
5118 |
+
"epoch": 0.14256299637219336,
|
5119 |
+
"grad_norm": 14.781683921813965,
|
5120 |
+
"learning_rate": 9.373913752580175e-05,
|
5121 |
+
"loss": 1.8918,
|
5122 |
+
"step": 727
|
5123 |
+
},
|
5124 |
+
{
|
5125 |
+
"epoch": 0.14275909402882636,
|
5126 |
+
"grad_norm": 6.98385763168335,
|
5127 |
+
"learning_rate": 9.351178418436488e-05,
|
5128 |
+
"loss": 0.8915,
|
5129 |
+
"step": 728
|
5130 |
+
},
|
5131 |
+
{
|
5132 |
+
"epoch": 0.14295519168545937,
|
5133 |
+
"grad_norm": 11.97396469116211,
|
5134 |
+
"learning_rate": 9.328446451705249e-05,
|
5135 |
+
"loss": 1.3453,
|
5136 |
+
"step": 729
|
5137 |
+
},
|
5138 |
+
{
|
5139 |
+
"epoch": 0.14315128934209237,
|
5140 |
+
"grad_norm": 7.939299583435059,
|
5141 |
+
"learning_rate": 9.305717970366358e-05,
|
5142 |
+
"loss": 1.7643,
|
5143 |
+
"step": 730
|
5144 |
+
},
|
5145 |
+
{
|
5146 |
+
"epoch": 0.14334738699872537,
|
5147 |
+
"grad_norm": 6.75924015045166,
|
5148 |
+
"learning_rate": 9.282993092381625e-05,
|
5149 |
+
"loss": 0.9198,
|
5150 |
+
"step": 731
|
5151 |
+
},
|
5152 |
+
{
|
5153 |
+
"epoch": 0.14354348465535838,
|
5154 |
+
"grad_norm": 7.962716579437256,
|
5155 |
+
"learning_rate": 9.260271935694168e-05,
|
5156 |
+
"loss": 1.5201,
|
5157 |
+
"step": 732
|
5158 |
+
},
|
5159 |
+
{
|
5160 |
+
"epoch": 0.14373958231199138,
|
5161 |
+
"grad_norm": 6.077849864959717,
|
5162 |
+
"learning_rate": 9.237554618227785e-05,
|
5163 |
+
"loss": 1.1027,
|
5164 |
+
"step": 733
|
5165 |
+
},
|
5166 |
+
{
|
5167 |
+
"epoch": 0.14393567996862439,
|
5168 |
+
"grad_norm": 5.715548992156982,
|
5169 |
+
"learning_rate": 9.214841257886349e-05,
|
5170 |
+
"loss": 0.9475,
|
5171 |
+
"step": 734
|
5172 |
+
},
|
5173 |
+
{
|
5174 |
+
"epoch": 0.1441317776252574,
|
5175 |
+
"grad_norm": 8.250515937805176,
|
5176 |
+
"learning_rate": 9.192131972553191e-05,
|
5177 |
+
"loss": 1.5568,
|
5178 |
+
"step": 735
|
5179 |
+
},
|
5180 |
+
{
|
5181 |
+
"epoch": 0.1443278752818904,
|
5182 |
+
"grad_norm": 4.075748443603516,
|
5183 |
+
"learning_rate": 9.16942688009051e-05,
|
5184 |
+
"loss": 1.1728,
|
5185 |
+
"step": 736
|
5186 |
+
},
|
5187 |
+
{
|
5188 |
+
"epoch": 0.1445239729385234,
|
5189 |
+
"grad_norm": 6.287231922149658,
|
5190 |
+
"learning_rate": 9.146726098338719e-05,
|
5191 |
+
"loss": 0.9493,
|
5192 |
+
"step": 737
|
5193 |
+
},
|
5194 |
+
{
|
5195 |
+
"epoch": 0.14472007059515637,
|
5196 |
+
"grad_norm": 7.561791896820068,
|
5197 |
+
"learning_rate": 9.12402974511587e-05,
|
5198 |
+
"loss": 1.2801,
|
5199 |
+
"step": 738
|
5200 |
+
},
|
5201 |
+
{
|
5202 |
+
"epoch": 0.14491616825178938,
|
5203 |
+
"grad_norm": 8.35126781463623,
|
5204 |
+
"learning_rate": 9.101337938217038e-05,
|
5205 |
+
"loss": 1.4908,
|
5206 |
+
"step": 739
|
5207 |
+
},
|
5208 |
+
{
|
5209 |
+
"epoch": 0.14511226590842238,
|
5210 |
+
"grad_norm": 7.66392707824707,
|
5211 |
+
"learning_rate": 9.078650795413692e-05,
|
5212 |
+
"loss": 2.3328,
|
5213 |
+
"step": 740
|
5214 |
+
},
|
5215 |
+
{
|
5216 |
+
"epoch": 0.1453083635650554,
|
5217 |
+
"grad_norm": 8.209086418151855,
|
5218 |
+
"learning_rate": 9.055968434453097e-05,
|
5219 |
+
"loss": 1.1856,
|
5220 |
+
"step": 741
|
5221 |
+
},
|
5222 |
+
{
|
5223 |
+
"epoch": 0.1455044612216884,
|
5224 |
+
"grad_norm": 6.804962635040283,
|
5225 |
+
"learning_rate": 9.0332909730577e-05,
|
5226 |
+
"loss": 1.8494,
|
5227 |
+
"step": 742
|
5228 |
+
},
|
5229 |
+
{
|
5230 |
+
"epoch": 0.1457005588783214,
|
5231 |
+
"grad_norm": 8.357268333435059,
|
5232 |
+
"learning_rate": 9.01061852892452e-05,
|
5233 |
+
"loss": 2.0554,
|
5234 |
+
"step": 743
|
5235 |
+
},
|
5236 |
+
{
|
5237 |
+
"epoch": 0.1458966565349544,
|
5238 |
+
"grad_norm": 9.071808815002441,
|
5239 |
+
"learning_rate": 8.98795121972454e-05,
|
5240 |
+
"loss": 1.455,
|
5241 |
+
"step": 744
|
5242 |
+
},
|
5243 |
+
{
|
5244 |
+
"epoch": 0.1460927541915874,
|
5245 |
+
"grad_norm": 6.041928768157959,
|
5246 |
+
"learning_rate": 8.965289163102078e-05,
|
5247 |
+
"loss": 1.386,
|
5248 |
+
"step": 745
|
5249 |
+
},
|
5250 |
+
{
|
5251 |
+
"epoch": 0.1462888518482204,
|
5252 |
+
"grad_norm": 9.787776947021484,
|
5253 |
+
"learning_rate": 8.942632476674211e-05,
|
5254 |
+
"loss": 1.4516,
|
5255 |
+
"step": 746
|
5256 |
+
},
|
5257 |
+
{
|
5258 |
+
"epoch": 0.1464849495048534,
|
5259 |
+
"grad_norm": 11.039609909057617,
|
5260 |
+
"learning_rate": 8.919981278030133e-05,
|
5261 |
+
"loss": 2.5638,
|
5262 |
+
"step": 747
|
5263 |
+
},
|
5264 |
+
{
|
5265 |
+
"epoch": 0.14668104716148642,
|
5266 |
+
"grad_norm": 7.651856899261475,
|
5267 |
+
"learning_rate": 8.89733568473055e-05,
|
5268 |
+
"loss": 2.2337,
|
5269 |
+
"step": 748
|
5270 |
+
},
|
5271 |
+
{
|
5272 |
+
"epoch": 0.14687714481811942,
|
5273 |
+
"grad_norm": 5.052631855010986,
|
5274 |
+
"learning_rate": 8.874695814307094e-05,
|
5275 |
+
"loss": 2.1706,
|
5276 |
+
"step": 749
|
5277 |
+
},
|
5278 |
+
{
|
5279 |
+
"epoch": 0.14707324247475242,
|
5280 |
+
"grad_norm": 4.033751964569092,
|
5281 |
+
"learning_rate": 8.852061784261678e-05,
|
5282 |
+
"loss": 0.7492,
|
5283 |
+
"step": 750
|
5284 |
+
},
|
5285 |
+
{
|
5286 |
+
"epoch": 0.14726934013138543,
|
5287 |
+
"grad_norm": 7.171450138092041,
|
5288 |
+
"learning_rate": 8.829433712065914e-05,
|
5289 |
+
"loss": 1.1281,
|
5290 |
+
"step": 751
|
5291 |
+
},
|
5292 |
+
{
|
5293 |
+
"epoch": 0.14746543778801843,
|
5294 |
+
"grad_norm": 4.533056735992432,
|
5295 |
+
"learning_rate": 8.806811715160485e-05,
|
5296 |
+
"loss": 1.1391,
|
5297 |
+
"step": 752
|
5298 |
+
},
|
5299 |
+
{
|
5300 |
+
"epoch": 0.14766153544465144,
|
5301 |
+
"grad_norm": 6.083869934082031,
|
5302 |
+
"learning_rate": 8.784195910954545e-05,
|
5303 |
+
"loss": 1.0986,
|
5304 |
+
"step": 753
|
5305 |
+
},
|
5306 |
+
{
|
5307 |
+
"epoch": 0.14785763310128444,
|
5308 |
+
"grad_norm": 8.265074729919434,
|
5309 |
+
"learning_rate": 8.761586416825117e-05,
|
5310 |
+
"loss": 2.8251,
|
5311 |
+
"step": 754
|
5312 |
+
},
|
5313 |
+
{
|
5314 |
+
"epoch": 0.14805373075791745,
|
5315 |
+
"grad_norm": 4.621420860290527,
|
5316 |
+
"learning_rate": 8.738983350116454e-05,
|
5317 |
+
"loss": 1.3493,
|
5318 |
+
"step": 755
|
5319 |
+
},
|
5320 |
+
{
|
5321 |
+
"epoch": 0.14824982841455045,
|
5322 |
+
"grad_norm": 9.803274154663086,
|
5323 |
+
"learning_rate": 8.716386828139477e-05,
|
5324 |
+
"loss": 1.6575,
|
5325 |
+
"step": 756
|
5326 |
+
},
|
5327 |
+
{
|
5328 |
+
"epoch": 0.14844592607118345,
|
5329 |
+
"grad_norm": 8.033260345458984,
|
5330 |
+
"learning_rate": 8.693796968171113e-05,
|
5331 |
+
"loss": 1.4908,
|
5332 |
+
"step": 757
|
5333 |
+
},
|
5334 |
+
{
|
5335 |
+
"epoch": 0.14864202372781646,
|
5336 |
+
"grad_norm": 10.312782287597656,
|
5337 |
+
"learning_rate": 8.671213887453735e-05,
|
5338 |
+
"loss": 1.0033,
|
5339 |
+
"step": 758
|
5340 |
+
},
|
5341 |
+
{
|
5342 |
+
"epoch": 0.14883812138444946,
|
5343 |
+
"grad_norm": 4.6735615730285645,
|
5344 |
+
"learning_rate": 8.648637703194516e-05,
|
5345 |
+
"loss": 2.4385,
|
5346 |
+
"step": 759
|
5347 |
+
},
|
5348 |
+
{
|
5349 |
+
"epoch": 0.14903421904108247,
|
5350 |
+
"grad_norm": 7.040375709533691,
|
5351 |
+
"learning_rate": 8.62606853256484e-05,
|
5352 |
+
"loss": 1.4345,
|
5353 |
+
"step": 760
|
5354 |
+
},
|
5355 |
+
{
|
5356 |
+
"epoch": 0.14923031669771547,
|
5357 |
+
"grad_norm": 5.844077110290527,
|
5358 |
+
"learning_rate": 8.603506492699699e-05,
|
5359 |
+
"loss": 1.9578,
|
5360 |
+
"step": 761
|
5361 |
+
},
|
5362 |
+
{
|
5363 |
+
"epoch": 0.14942641435434847,
|
5364 |
+
"grad_norm": 7.8477606773376465,
|
5365 |
+
"learning_rate": 8.58095170069706e-05,
|
5366 |
+
"loss": 1.915,
|
5367 |
+
"step": 762
|
5368 |
+
},
|
5369 |
+
{
|
5370 |
+
"epoch": 0.14962251201098148,
|
5371 |
+
"grad_norm": 6.696018218994141,
|
5372 |
+
"learning_rate": 8.55840427361729e-05,
|
5373 |
+
"loss": 2.6032,
|
5374 |
+
"step": 763
|
5375 |
+
},
|
5376 |
+
{
|
5377 |
+
"epoch": 0.14981860966761448,
|
5378 |
+
"grad_norm": 6.304353713989258,
|
5379 |
+
"learning_rate": 8.535864328482523e-05,
|
5380 |
+
"loss": 1.0302,
|
5381 |
+
"step": 764
|
5382 |
+
},
|
5383 |
+
{
|
5384 |
+
"epoch": 0.1500147073242475,
|
5385 |
+
"grad_norm": 4.32615852355957,
|
5386 |
+
"learning_rate": 8.513331982276053e-05,
|
5387 |
+
"loss": 1.253,
|
5388 |
+
"step": 765
|
5389 |
+
},
|
5390 |
+
{
|
5391 |
+
"epoch": 0.1502108049808805,
|
5392 |
+
"grad_norm": 5.056833744049072,
|
5393 |
+
"learning_rate": 8.490807351941753e-05,
|
5394 |
+
"loss": 0.9107,
|
5395 |
+
"step": 766
|
5396 |
+
},
|
5397 |
+
{
|
5398 |
+
"epoch": 0.1504069026375135,
|
5399 |
+
"grad_norm": 11.254875183105469,
|
5400 |
+
"learning_rate": 8.468290554383436e-05,
|
5401 |
+
"loss": 2.1077,
|
5402 |
+
"step": 767
|
5403 |
+
},
|
5404 |
+
{
|
5405 |
+
"epoch": 0.15060300029414647,
|
5406 |
+
"grad_norm": 5.245373249053955,
|
5407 |
+
"learning_rate": 8.445781706464277e-05,
|
5408 |
+
"loss": 0.7334,
|
5409 |
+
"step": 768
|
5410 |
+
},
|
5411 |
+
{
|
5412 |
+
"epoch": 0.15079909795077948,
|
5413 |
+
"grad_norm": 6.346343994140625,
|
5414 |
+
"learning_rate": 8.423280925006178e-05,
|
5415 |
+
"loss": 2.2026,
|
5416 |
+
"step": 769
|
5417 |
+
},
|
5418 |
+
{
|
5419 |
+
"epoch": 0.15099519560741248,
|
5420 |
+
"grad_norm": 9.745604515075684,
|
5421 |
+
"learning_rate": 8.400788326789175e-05,
|
5422 |
+
"loss": 0.8139,
|
5423 |
+
"step": 770
|
5424 |
+
},
|
5425 |
+
{
|
5426 |
+
"epoch": 0.15119129326404548,
|
5427 |
+
"grad_norm": 4.820820331573486,
|
5428 |
+
"learning_rate": 8.378304028550848e-05,
|
5429 |
+
"loss": 1.2438,
|
5430 |
+
"step": 771
|
5431 |
+
},
|
5432 |
+
{
|
5433 |
+
"epoch": 0.1513873909206785,
|
5434 |
+
"grad_norm": 7.597200393676758,
|
5435 |
+
"learning_rate": 8.355828146985684e-05,
|
5436 |
+
"loss": 2.3756,
|
5437 |
+
"step": 772
|
5438 |
+
},
|
5439 |
+
{
|
5440 |
+
"epoch": 0.1515834885773115,
|
5441 |
+
"grad_norm": 4.522254943847656,
|
5442 |
+
"learning_rate": 8.333360798744496e-05,
|
5443 |
+
"loss": 1.9562,
|
5444 |
+
"step": 773
|
5445 |
+
},
|
5446 |
+
{
|
5447 |
+
"epoch": 0.1517795862339445,
|
5448 |
+
"grad_norm": 11.164361000061035,
|
5449 |
+
"learning_rate": 8.310902100433809e-05,
|
5450 |
+
"loss": 2.1888,
|
5451 |
+
"step": 774
|
5452 |
+
},
|
5453 |
+
{
|
5454 |
+
"epoch": 0.1519756838905775,
|
5455 |
+
"grad_norm": 8.457398414611816,
|
5456 |
+
"learning_rate": 8.288452168615242e-05,
|
5457 |
+
"loss": 0.7737,
|
5458 |
+
"step": 775
|
5459 |
+
},
|
5460 |
+
{
|
5461 |
+
"epoch": 0.1521717815472105,
|
5462 |
+
"grad_norm": 10.31162166595459,
|
5463 |
+
"learning_rate": 8.266011119804936e-05,
|
5464 |
+
"loss": 1.1301,
|
5465 |
+
"step": 776
|
5466 |
+
},
|
5467 |
+
{
|
5468 |
+
"epoch": 0.1523678792038435,
|
5469 |
+
"grad_norm": 6.120646953582764,
|
5470 |
+
"learning_rate": 8.243579070472909e-05,
|
5471 |
+
"loss": 0.6059,
|
5472 |
+
"step": 777
|
5473 |
+
},
|
5474 |
+
{
|
5475 |
+
"epoch": 0.1525639768604765,
|
5476 |
+
"grad_norm": 7.931275844573975,
|
5477 |
+
"learning_rate": 8.221156137042489e-05,
|
5478 |
+
"loss": 2.2832,
|
5479 |
+
"step": 778
|
5480 |
+
},
|
5481 |
+
{
|
5482 |
+
"epoch": 0.15276007451710952,
|
5483 |
+
"grad_norm": 8.665715217590332,
|
5484 |
+
"learning_rate": 8.198742435889674e-05,
|
5485 |
+
"loss": 1.5639,
|
5486 |
+
"step": 779
|
5487 |
+
},
|
5488 |
+
{
|
5489 |
+
"epoch": 0.15295617217374252,
|
5490 |
+
"grad_norm": 6.210934638977051,
|
5491 |
+
"learning_rate": 8.176338083342561e-05,
|
5492 |
+
"loss": 0.5992,
|
5493 |
+
"step": 780
|
5494 |
+
},
|
5495 |
+
{
|
5496 |
+
"epoch": 0.15315226983037553,
|
5497 |
+
"grad_norm": 4.350475311279297,
|
5498 |
+
"learning_rate": 8.153943195680723e-05,
|
5499 |
+
"loss": 1.9462,
|
5500 |
+
"step": 781
|
5501 |
+
},
|
5502 |
+
{
|
5503 |
+
"epoch": 0.15334836748700853,
|
5504 |
+
"grad_norm": 10.62283992767334,
|
5505 |
+
"learning_rate": 8.131557889134602e-05,
|
5506 |
+
"loss": 2.5713,
|
5507 |
+
"step": 782
|
5508 |
+
},
|
5509 |
+
{
|
5510 |
+
"epoch": 0.15354446514364153,
|
5511 |
+
"grad_norm": 19.049753189086914,
|
5512 |
+
"learning_rate": 8.109182279884928e-05,
|
5513 |
+
"loss": 2.0163,
|
5514 |
+
"step": 783
|
5515 |
+
},
|
5516 |
+
{
|
5517 |
+
"epoch": 0.15374056280027454,
|
5518 |
+
"grad_norm": 6.098108768463135,
|
5519 |
+
"learning_rate": 8.086816484062085e-05,
|
5520 |
+
"loss": 0.9044,
|
5521 |
+
"step": 784
|
5522 |
+
},
|
5523 |
+
{
|
5524 |
+
"epoch": 0.15393666045690754,
|
5525 |
+
"grad_norm": 4.7711567878723145,
|
5526 |
+
"learning_rate": 8.064460617745542e-05,
|
5527 |
+
"loss": 1.8375,
|
5528 |
+
"step": 785
|
5529 |
+
},
|
5530 |
+
{
|
5531 |
+
"epoch": 0.15413275811354055,
|
5532 |
+
"grad_norm": 6.954524993896484,
|
5533 |
+
"learning_rate": 8.042114796963219e-05,
|
5534 |
+
"loss": 1.4588,
|
5535 |
+
"step": 786
|
5536 |
+
},
|
5537 |
+
{
|
5538 |
+
"epoch": 0.15432885577017355,
|
5539 |
+
"grad_norm": 4.3450422286987305,
|
5540 |
+
"learning_rate": 8.019779137690906e-05,
|
5541 |
+
"loss": 1.9173,
|
5542 |
+
"step": 787
|
5543 |
+
},
|
5544 |
+
{
|
5545 |
+
"epoch": 0.15452495342680655,
|
5546 |
+
"grad_norm": 6.4524688720703125,
|
5547 |
+
"learning_rate": 7.997453755851658e-05,
|
5548 |
+
"loss": 0.9777,
|
5549 |
+
"step": 788
|
5550 |
+
},
|
5551 |
+
{
|
5552 |
+
"epoch": 0.15472105108343956,
|
5553 |
+
"grad_norm": 4.7262139320373535,
|
5554 |
+
"learning_rate": 7.975138767315178e-05,
|
5555 |
+
"loss": 2.0938,
|
5556 |
+
"step": 789
|
5557 |
+
},
|
5558 |
+
{
|
5559 |
+
"epoch": 0.15491714874007256,
|
5560 |
+
"grad_norm": 6.220945835113525,
|
5561 |
+
"learning_rate": 7.952834287897242e-05,
|
5562 |
+
"loss": 1.1835,
|
5563 |
+
"step": 790
|
5564 |
+
},
|
5565 |
+
{
|
5566 |
+
"epoch": 0.15511324639670557,
|
5567 |
+
"grad_norm": 12.306231498718262,
|
5568 |
+
"learning_rate": 7.930540433359071e-05,
|
5569 |
+
"loss": 1.2498,
|
5570 |
+
"step": 791
|
5571 |
+
},
|
5572 |
+
{
|
5573 |
+
"epoch": 0.15530934405333857,
|
5574 |
+
"grad_norm": 5.665555953979492,
|
5575 |
+
"learning_rate": 7.908257319406747e-05,
|
5576 |
+
"loss": 1.2846,
|
5577 |
+
"step": 792
|
5578 |
+
},
|
5579 |
+
{
|
5580 |
+
"epoch": 0.15550544170997158,
|
5581 |
+
"grad_norm": 5.636224269866943,
|
5582 |
+
"learning_rate": 7.88598506169061e-05,
|
5583 |
+
"loss": 1.1359,
|
5584 |
+
"step": 793
|
5585 |
+
},
|
5586 |
+
{
|
5587 |
+
"epoch": 0.15570153936660458,
|
5588 |
+
"grad_norm": 7.562717914581299,
|
5589 |
+
"learning_rate": 7.863723775804651e-05,
|
5590 |
+
"loss": 1.3149,
|
5591 |
+
"step": 794
|
5592 |
+
},
|
5593 |
+
{
|
5594 |
+
"epoch": 0.15589763702323758,
|
5595 |
+
"grad_norm": 6.797637462615967,
|
5596 |
+
"learning_rate": 7.841473577285925e-05,
|
5597 |
+
"loss": 1.2934,
|
5598 |
+
"step": 795
|
5599 |
+
},
|
5600 |
+
{
|
5601 |
+
"epoch": 0.1560937346798706,
|
5602 |
+
"grad_norm": 9.185527801513672,
|
5603 |
+
"learning_rate": 7.819234581613934e-05,
|
5604 |
+
"loss": 2.0876,
|
5605 |
+
"step": 796
|
5606 |
+
},
|
5607 |
+
{
|
5608 |
+
"epoch": 0.1562898323365036,
|
5609 |
+
"grad_norm": 5.824263095855713,
|
5610 |
+
"learning_rate": 7.797006904210035e-05,
|
5611 |
+
"loss": 1.3984,
|
5612 |
+
"step": 797
|
5613 |
+
},
|
5614 |
+
{
|
5615 |
+
"epoch": 0.15648592999313657,
|
5616 |
+
"grad_norm": 8.152481079101562,
|
5617 |
+
"learning_rate": 7.774790660436858e-05,
|
5618 |
+
"loss": 1.9248,
|
5619 |
+
"step": 798
|
5620 |
+
},
|
5621 |
+
{
|
5622 |
+
"epoch": 0.15668202764976957,
|
5623 |
+
"grad_norm": 5.90303897857666,
|
5624 |
+
"learning_rate": 7.752585965597673e-05,
|
5625 |
+
"loss": 1.3629,
|
5626 |
+
"step": 799
|
5627 |
+
},
|
5628 |
+
{
|
5629 |
+
"epoch": 0.15687812530640258,
|
5630 |
+
"grad_norm": 7.13554573059082,
|
5631 |
+
"learning_rate": 7.730392934935825e-05,
|
5632 |
+
"loss": 1.6686,
|
5633 |
+
"step": 800
|
5634 |
+
},
|
5635 |
+
{
|
5636 |
+
"epoch": 0.15707422296303558,
|
5637 |
+
"grad_norm": 14.757647514343262,
|
5638 |
+
"learning_rate": 7.708211683634112e-05,
|
5639 |
+
"loss": 3.0637,
|
5640 |
+
"step": 801
|
5641 |
+
},
|
5642 |
+
{
|
5643 |
+
"epoch": 0.15727032061966859,
|
5644 |
+
"grad_norm": 6.754796504974365,
|
5645 |
+
"learning_rate": 7.686042326814205e-05,
|
5646 |
+
"loss": 1.6339,
|
5647 |
+
"step": 802
|
5648 |
+
},
|
5649 |
+
{
|
5650 |
+
"epoch": 0.1574664182763016,
|
5651 |
+
"grad_norm": 4.844980716705322,
|
5652 |
+
"learning_rate": 7.663884979536035e-05,
|
5653 |
+
"loss": 1.799,
|
5654 |
+
"step": 803
|
5655 |
+
},
|
5656 |
+
{
|
5657 |
+
"epoch": 0.1576625159329346,
|
5658 |
+
"grad_norm": 7.784121990203857,
|
5659 |
+
"learning_rate": 7.641739756797202e-05,
|
5660 |
+
"loss": 1.4723,
|
5661 |
+
"step": 804
|
5662 |
+
},
|
5663 |
+
{
|
5664 |
+
"epoch": 0.1578586135895676,
|
5665 |
+
"grad_norm": 9.167228698730469,
|
5666 |
+
"learning_rate": 7.619606773532386e-05,
|
5667 |
+
"loss": 1.3234,
|
5668 |
+
"step": 805
|
5669 |
+
},
|
5670 |
+
{
|
5671 |
+
"epoch": 0.1580547112462006,
|
5672 |
+
"grad_norm": 4.63117790222168,
|
5673 |
+
"learning_rate": 7.59748614461274e-05,
|
5674 |
+
"loss": 1.2373,
|
5675 |
+
"step": 806
|
5676 |
+
},
|
5677 |
+
{
|
5678 |
+
"epoch": 0.1582508089028336,
|
5679 |
+
"grad_norm": 6.5301690101623535,
|
5680 |
+
"learning_rate": 7.5753779848453e-05,
|
5681 |
+
"loss": 2.5172,
|
5682 |
+
"step": 807
|
5683 |
+
},
|
5684 |
+
{
|
5685 |
+
"epoch": 0.1584469065594666,
|
5686 |
+
"grad_norm": 6.791456699371338,
|
5687 |
+
"learning_rate": 7.553282408972382e-05,
|
5688 |
+
"loss": 1.563,
|
5689 |
+
"step": 808
|
5690 |
+
},
|
5691 |
+
{
|
5692 |
+
"epoch": 0.15864300421609961,
|
5693 |
+
"grad_norm": 5.697129249572754,
|
5694 |
+
"learning_rate": 7.531199531670988e-05,
|
5695 |
+
"loss": 1.6731,
|
5696 |
+
"step": 809
|
5697 |
+
},
|
5698 |
+
{
|
5699 |
+
"epoch": 0.15883910187273262,
|
5700 |
+
"grad_norm": 6.511412143707275,
|
5701 |
+
"learning_rate": 7.50912946755223e-05,
|
5702 |
+
"loss": 1.2208,
|
5703 |
+
"step": 810
|
5704 |
+
},
|
5705 |
+
{
|
5706 |
+
"epoch": 0.15903519952936562,
|
5707 |
+
"grad_norm": 4.849149703979492,
|
5708 |
+
"learning_rate": 7.487072331160696e-05,
|
5709 |
+
"loss": 1.9081,
|
5710 |
+
"step": 811
|
5711 |
+
},
|
5712 |
+
{
|
5713 |
+
"epoch": 0.15923129718599863,
|
5714 |
+
"grad_norm": 2.844851016998291,
|
5715 |
+
"learning_rate": 7.465028236973897e-05,
|
5716 |
+
"loss": 1.2562,
|
5717 |
+
"step": 812
|
5718 |
+
},
|
5719 |
+
{
|
5720 |
+
"epoch": 0.15942739484263163,
|
5721 |
+
"grad_norm": 4.782979965209961,
|
5722 |
+
"learning_rate": 7.442997299401652e-05,
|
5723 |
+
"loss": 0.8038,
|
5724 |
+
"step": 813
|
5725 |
+
},
|
5726 |
+
{
|
5727 |
+
"epoch": 0.15962349249926464,
|
5728 |
+
"grad_norm": 13.652055740356445,
|
5729 |
+
"learning_rate": 7.420979632785483e-05,
|
5730 |
+
"loss": 1.3842,
|
5731 |
+
"step": 814
|
5732 |
+
},
|
5733 |
+
{
|
5734 |
+
"epoch": 0.15981959015589764,
|
5735 |
+
"grad_norm": 9.692163467407227,
|
5736 |
+
"learning_rate": 7.398975351398053e-05,
|
5737 |
+
"loss": 1.3089,
|
5738 |
+
"step": 815
|
5739 |
+
},
|
5740 |
+
{
|
5741 |
+
"epoch": 0.16001568781253064,
|
5742 |
+
"grad_norm": 6.529005527496338,
|
5743 |
+
"learning_rate": 7.37698456944254e-05,
|
5744 |
+
"loss": 1.8166,
|
5745 |
+
"step": 816
|
5746 |
+
},
|
5747 |
+
{
|
5748 |
+
"epoch": 0.16021178546916365,
|
5749 |
+
"grad_norm": 6.1307454109191895,
|
5750 |
+
"learning_rate": 7.355007401052072e-05,
|
5751 |
+
"loss": 1.6805,
|
5752 |
+
"step": 817
|
5753 |
+
},
|
5754 |
+
{
|
5755 |
+
"epoch": 0.16040788312579665,
|
5756 |
+
"grad_norm": 23.4924259185791,
|
5757 |
+
"learning_rate": 7.333043960289113e-05,
|
5758 |
+
"loss": 1.8122,
|
5759 |
+
"step": 818
|
5760 |
+
},
|
5761 |
+
{
|
5762 |
+
"epoch": 0.16060398078242966,
|
5763 |
+
"grad_norm": 5.680008411407471,
|
5764 |
+
"learning_rate": 7.311094361144881e-05,
|
5765 |
+
"loss": 1.7381,
|
5766 |
+
"step": 819
|
5767 |
+
},
|
5768 |
+
{
|
5769 |
+
"epoch": 0.16080007843906266,
|
5770 |
+
"grad_norm": 6.790538787841797,
|
5771 |
+
"learning_rate": 7.289158717538765e-05,
|
5772 |
+
"loss": 1.9142,
|
5773 |
+
"step": 820
|
5774 |
+
},
|
5775 |
+
{
|
5776 |
+
"epoch": 0.16099617609569566,
|
5777 |
+
"grad_norm": 7.620028495788574,
|
5778 |
+
"learning_rate": 7.267237143317706e-05,
|
5779 |
+
"loss": 1.8411,
|
5780 |
+
"step": 821
|
5781 |
+
},
|
5782 |
+
{
|
5783 |
+
"epoch": 0.16119227375232867,
|
5784 |
+
"grad_norm": 5.989513397216797,
|
5785 |
+
"learning_rate": 7.245329752255647e-05,
|
5786 |
+
"loss": 1.7845,
|
5787 |
+
"step": 822
|
5788 |
+
},
|
5789 |
+
{
|
5790 |
+
"epoch": 0.16138837140896167,
|
5791 |
+
"grad_norm": 5.91652774810791,
|
5792 |
+
"learning_rate": 7.223436658052898e-05,
|
5793 |
+
"loss": 1.7267,
|
5794 |
+
"step": 823
|
5795 |
+
},
|
5796 |
+
{
|
5797 |
+
"epoch": 0.16158446906559468,
|
5798 |
+
"grad_norm": 6.713134765625,
|
5799 |
+
"learning_rate": 7.201557974335583e-05,
|
5800 |
+
"loss": 1.6799,
|
5801 |
+
"step": 824
|
5802 |
+
},
|
5803 |
+
{
|
5804 |
+
"epoch": 0.16178056672222768,
|
5805 |
+
"grad_norm": 5.243997097015381,
|
5806 |
+
"learning_rate": 7.179693814655026e-05,
|
5807 |
+
"loss": 1.7857,
|
5808 |
+
"step": 825
|
5809 |
+
},
|
5810 |
+
{
|
5811 |
+
"epoch": 0.16197666437886069,
|
5812 |
+
"grad_norm": 4.856655120849609,
|
5813 |
+
"learning_rate": 7.157844292487173e-05,
|
5814 |
+
"loss": 1.0967,
|
5815 |
+
"step": 826
|
5816 |
+
},
|
5817 |
+
{
|
5818 |
+
"epoch": 0.1621727620354937,
|
5819 |
+
"grad_norm": 4.94920015335083,
|
5820 |
+
"learning_rate": 7.13600952123201e-05,
|
5821 |
+
"loss": 1.4571,
|
5822 |
+
"step": 827
|
5823 |
+
},
|
5824 |
+
{
|
5825 |
+
"epoch": 0.16236885969212667,
|
5826 |
+
"grad_norm": 5.146086692810059,
|
5827 |
+
"learning_rate": 7.114189614212944e-05,
|
5828 |
+
"loss": 0.6922,
|
5829 |
+
"step": 828
|
5830 |
+
},
|
5831 |
+
{
|
5832 |
+
"epoch": 0.16256495734875967,
|
5833 |
+
"grad_norm": 8.812291145324707,
|
5834 |
+
"learning_rate": 7.092384684676262e-05,
|
5835 |
+
"loss": 1.2531,
|
5836 |
+
"step": 829
|
5837 |
+
},
|
5838 |
+
{
|
5839 |
+
"epoch": 0.16276105500539267,
|
5840 |
+
"grad_norm": 6.924501419067383,
|
5841 |
+
"learning_rate": 7.070594845790497e-05,
|
5842 |
+
"loss": 1.3011,
|
5843 |
+
"step": 830
|
5844 |
+
},
|
5845 |
+
{
|
5846 |
+
"epoch": 0.16295715266202568,
|
5847 |
+
"grad_norm": 10.452120780944824,
|
5848 |
+
"learning_rate": 7.048820210645862e-05,
|
5849 |
+
"loss": 1.1114,
|
5850 |
+
"step": 831
|
5851 |
+
},
|
5852 |
+
{
|
5853 |
+
"epoch": 0.16315325031865868,
|
5854 |
+
"grad_norm": 8.212396621704102,
|
5855 |
+
"learning_rate": 7.027060892253679e-05,
|
5856 |
+
"loss": 0.9207,
|
5857 |
+
"step": 832
|
5858 |
+
},
|
5859 |
+
{
|
5860 |
+
"epoch": 0.1633493479752917,
|
5861 |
+
"grad_norm": 17.70586585998535,
|
5862 |
+
"learning_rate": 7.005317003545754e-05,
|
5863 |
+
"loss": 1.3019,
|
5864 |
+
"step": 833
|
5865 |
+
},
|
5866 |
+
{
|
5867 |
+
"epoch": 0.1635454456319247,
|
5868 |
+
"grad_norm": 12.174421310424805,
|
5869 |
+
"learning_rate": 6.983588657373833e-05,
|
5870 |
+
"loss": 2.1275,
|
5871 |
+
"step": 834
|
5872 |
+
},
|
5873 |
+
{
|
5874 |
+
"epoch": 0.1637415432885577,
|
5875 |
+
"grad_norm": 3.9426016807556152,
|
5876 |
+
"learning_rate": 6.96187596650898e-05,
|
5877 |
+
"loss": 2.4098,
|
5878 |
+
"step": 835
|
5879 |
+
},
|
5880 |
+
{
|
5881 |
+
"epoch": 0.1639376409451907,
|
5882 |
+
"grad_norm": 6.121764183044434,
|
5883 |
+
"learning_rate": 6.940179043641005e-05,
|
5884 |
+
"loss": 1.0717,
|
5885 |
+
"step": 836
|
5886 |
+
},
|
5887 |
+
{
|
5888 |
+
"epoch": 0.1641337386018237,
|
5889 |
+
"grad_norm": 4.13683557510376,
|
5890 |
+
"learning_rate": 6.918498001377901e-05,
|
5891 |
+
"loss": 1.4934,
|
5892 |
+
"step": 837
|
5893 |
+
},
|
5894 |
+
{
|
5895 |
+
"epoch": 0.1643298362584567,
|
5896 |
+
"grad_norm": 4.907169818878174,
|
5897 |
+
"learning_rate": 6.896832952245217e-05,
|
5898 |
+
"loss": 1.9002,
|
5899 |
+
"step": 838
|
5900 |
+
},
|
5901 |
+
{
|
5902 |
+
"epoch": 0.1645259339150897,
|
5903 |
+
"grad_norm": 8.611984252929688,
|
5904 |
+
"learning_rate": 6.875184008685514e-05,
|
5905 |
+
"loss": 1.1805,
|
5906 |
+
"step": 839
|
5907 |
+
},
|
5908 |
+
{
|
5909 |
+
"epoch": 0.16472203157172272,
|
5910 |
+
"grad_norm": 8.626840591430664,
|
5911 |
+
"learning_rate": 6.853551283057757e-05,
|
5912 |
+
"loss": 1.1437,
|
5913 |
+
"step": 840
|
5914 |
+
},
|
5915 |
+
{
|
5916 |
+
"epoch": 0.16491812922835572,
|
5917 |
+
"grad_norm": 6.1718878746032715,
|
5918 |
+
"learning_rate": 6.831934887636738e-05,
|
5919 |
+
"loss": 1.5465,
|
5920 |
+
"step": 841
|
5921 |
+
},
|
5922 |
+
{
|
5923 |
+
"epoch": 0.16511422688498872,
|
5924 |
+
"grad_norm": 6.6393961906433105,
|
5925 |
+
"learning_rate": 6.8103349346125e-05,
|
5926 |
+
"loss": 1.0682,
|
5927 |
+
"step": 842
|
5928 |
+
},
|
5929 |
+
{
|
5930 |
+
"epoch": 0.16531032454162173,
|
5931 |
+
"grad_norm": 5.233375072479248,
|
5932 |
+
"learning_rate": 6.788751536089739e-05,
|
5933 |
+
"loss": 1.8218,
|
5934 |
+
"step": 843
|
5935 |
+
},
|
5936 |
+
{
|
5937 |
+
"epoch": 0.16550642219825473,
|
5938 |
+
"grad_norm": 9.505565643310547,
|
5939 |
+
"learning_rate": 6.767184804087248e-05,
|
5940 |
+
"loss": 1.3462,
|
5941 |
+
"step": 844
|
5942 |
+
},
|
5943 |
+
{
|
5944 |
+
"epoch": 0.16570251985488774,
|
5945 |
+
"grad_norm": 5.554503440856934,
|
5946 |
+
"learning_rate": 6.745634850537302e-05,
|
5947 |
+
"loss": 0.7193,
|
5948 |
+
"step": 845
|
5949 |
+
},
|
5950 |
+
{
|
5951 |
+
"epoch": 0.16589861751152074,
|
5952 |
+
"grad_norm": 6.603649139404297,
|
5953 |
+
"learning_rate": 6.724101787285114e-05,
|
5954 |
+
"loss": 1.3164,
|
5955 |
+
"step": 846
|
5956 |
+
},
|
5957 |
+
{
|
5958 |
+
"epoch": 0.16609471516815374,
|
5959 |
+
"grad_norm": 10.690621376037598,
|
5960 |
+
"learning_rate": 6.702585726088222e-05,
|
5961 |
+
"loss": 3.3145,
|
5962 |
+
"step": 847
|
5963 |
+
},
|
5964 |
+
{
|
5965 |
+
"epoch": 0.16629081282478675,
|
5966 |
+
"grad_norm": 17.521020889282227,
|
5967 |
+
"learning_rate": 6.681086778615922e-05,
|
5968 |
+
"loss": 2.3221,
|
5969 |
+
"step": 848
|
5970 |
+
},
|
5971 |
+
{
|
5972 |
+
"epoch": 0.16648691048141975,
|
5973 |
+
"grad_norm": 11.0389986038208,
|
5974 |
+
"learning_rate": 6.659605056448702e-05,
|
5975 |
+
"loss": 1.2903,
|
5976 |
+
"step": 849
|
5977 |
+
},
|
5978 |
+
{
|
5979 |
+
"epoch": 0.16668300813805276,
|
5980 |
+
"grad_norm": 5.6815900802612305,
|
5981 |
+
"learning_rate": 6.638140671077633e-05,
|
5982 |
+
"loss": 1.1651,
|
5983 |
+
"step": 850
|
5984 |
+
},
|
5985 |
+
{
|
5986 |
+
"epoch": 0.16687910579468576,
|
5987 |
+
"grad_norm": 14.61434555053711,
|
5988 |
+
"learning_rate": 6.616693733903823e-05,
|
5989 |
+
"loss": 2.5921,
|
5990 |
+
"step": 851
|
5991 |
+
},
|
5992 |
+
{
|
5993 |
+
"epoch": 0.16707520345131877,
|
5994 |
+
"grad_norm": 6.983891487121582,
|
5995 |
+
"learning_rate": 6.595264356237812e-05,
|
5996 |
+
"loss": 1.2114,
|
5997 |
+
"step": 852
|
5998 |
+
},
|
5999 |
+
{
|
6000 |
+
"epoch": 0.16727130110795177,
|
6001 |
+
"grad_norm": 4.762997627258301,
|
6002 |
+
"learning_rate": 6.573852649299009e-05,
|
6003 |
+
"loss": 0.8655,
|
6004 |
+
"step": 853
|
6005 |
+
},
|
6006 |
+
{
|
6007 |
+
"epoch": 0.16746739876458477,
|
6008 |
+
"grad_norm": 8.16657829284668,
|
6009 |
+
"learning_rate": 6.552458724215114e-05,
|
6010 |
+
"loss": 1.584,
|
6011 |
+
"step": 854
|
6012 |
+
},
|
6013 |
+
{
|
6014 |
+
"epoch": 0.16766349642121778,
|
6015 |
+
"grad_norm": 3.753600597381592,
|
6016 |
+
"learning_rate": 6.531082692021532e-05,
|
6017 |
+
"loss": 0.7213,
|
6018 |
+
"step": 855
|
6019 |
+
},
|
6020 |
+
{
|
6021 |
+
"epoch": 0.16785959407785078,
|
6022 |
+
"grad_norm": 20.525659561157227,
|
6023 |
+
"learning_rate": 6.509724663660813e-05,
|
6024 |
+
"loss": 3.2134,
|
6025 |
+
"step": 856
|
6026 |
+
},
|
6027 |
+
{
|
6028 |
+
"epoch": 0.1680556917344838,
|
6029 |
+
"grad_norm": 4.803929805755615,
|
6030 |
+
"learning_rate": 6.488384749982053e-05,
|
6031 |
+
"loss": 1.0283,
|
6032 |
+
"step": 857
|
6033 |
+
},
|
6034 |
+
{
|
6035 |
+
"epoch": 0.16825178939111676,
|
6036 |
+
"grad_norm": 4.031068325042725,
|
6037 |
+
"learning_rate": 6.467063061740345e-05,
|
6038 |
+
"loss": 0.86,
|
6039 |
+
"step": 858
|
6040 |
+
},
|
6041 |
+
{
|
6042 |
+
"epoch": 0.16844788704774977,
|
6043 |
+
"grad_norm": 8.304176330566406,
|
6044 |
+
"learning_rate": 6.445759709596182e-05,
|
6045 |
+
"loss": 0.7242,
|
6046 |
+
"step": 859
|
6047 |
+
},
|
6048 |
+
{
|
6049 |
+
"epoch": 0.16864398470438277,
|
6050 |
+
"grad_norm": 5.160337448120117,
|
6051 |
+
"learning_rate": 6.424474804114895e-05,
|
6052 |
+
"loss": 1.7592,
|
6053 |
+
"step": 860
|
6054 |
+
},
|
6055 |
+
{
|
6056 |
+
"epoch": 0.16884008236101578,
|
6057 |
+
"grad_norm": 6.138307571411133,
|
6058 |
+
"learning_rate": 6.403208455766081e-05,
|
6059 |
+
"loss": 2.6926,
|
6060 |
+
"step": 861
|
6061 |
+
},
|
6062 |
+
{
|
6063 |
+
"epoch": 0.16903618001764878,
|
6064 |
+
"grad_norm": 3.878741979598999,
|
6065 |
+
"learning_rate": 6.381960774923017e-05,
|
6066 |
+
"loss": 1.2294,
|
6067 |
+
"step": 862
|
6068 |
+
},
|
6069 |
+
{
|
6070 |
+
"epoch": 0.16923227767428178,
|
6071 |
+
"grad_norm": 4.720643043518066,
|
6072 |
+
"learning_rate": 6.360731871862093e-05,
|
6073 |
+
"loss": 1.7562,
|
6074 |
+
"step": 863
|
6075 |
+
},
|
6076 |
+
{
|
6077 |
+
"epoch": 0.1694283753309148,
|
6078 |
+
"grad_norm": 4.653330326080322,
|
6079 |
+
"learning_rate": 6.339521856762254e-05,
|
6080 |
+
"loss": 1.2522,
|
6081 |
+
"step": 864
|
6082 |
+
},
|
6083 |
+
{
|
6084 |
+
"epoch": 0.1696244729875478,
|
6085 |
+
"grad_norm": 6.003101348876953,
|
6086 |
+
"learning_rate": 6.318330839704405e-05,
|
6087 |
+
"loss": 1.4006,
|
6088 |
+
"step": 865
|
6089 |
+
},
|
6090 |
+
{
|
6091 |
+
"epoch": 0.1698205706441808,
|
6092 |
+
"grad_norm": 5.075412273406982,
|
6093 |
+
"learning_rate": 6.297158930670852e-05,
|
6094 |
+
"loss": 0.7885,
|
6095 |
+
"step": 866
|
6096 |
+
},
|
6097 |
+
{
|
6098 |
+
"epoch": 0.1700166683008138,
|
6099 |
+
"grad_norm": 4.062684059143066,
|
6100 |
+
"learning_rate": 6.276006239544729e-05,
|
6101 |
+
"loss": 1.0201,
|
6102 |
+
"step": 867
|
6103 |
+
},
|
6104 |
+
{
|
6105 |
+
"epoch": 0.1702127659574468,
|
6106 |
+
"grad_norm": 5.678539752960205,
|
6107 |
+
"learning_rate": 6.254872876109438e-05,
|
6108 |
+
"loss": 1.1822,
|
6109 |
+
"step": 868
|
6110 |
+
},
|
6111 |
+
{
|
6112 |
+
"epoch": 0.1704088636140798,
|
6113 |
+
"grad_norm": 4.655857563018799,
|
6114 |
+
"learning_rate": 6.233758950048056e-05,
|
6115 |
+
"loss": 1.4193,
|
6116 |
+
"step": 869
|
6117 |
+
},
|
6118 |
+
{
|
6119 |
+
"epoch": 0.1706049612707128,
|
6120 |
+
"grad_norm": 18.205698013305664,
|
6121 |
+
"learning_rate": 6.21266457094278e-05,
|
6122 |
+
"loss": 0.7067,
|
6123 |
+
"step": 870
|
6124 |
+
},
|
6125 |
+
{
|
6126 |
+
"epoch": 0.17080105892734582,
|
6127 |
+
"grad_norm": 8.73768424987793,
|
6128 |
+
"learning_rate": 6.191589848274368e-05,
|
6129 |
+
"loss": 1.9048,
|
6130 |
+
"step": 871
|
6131 |
+
},
|
6132 |
+
{
|
6133 |
+
"epoch": 0.17099715658397882,
|
6134 |
+
"grad_norm": 5.723482131958008,
|
6135 |
+
"learning_rate": 6.170534891421556e-05,
|
6136 |
+
"loss": 0.8121,
|
6137 |
+
"step": 872
|
6138 |
+
},
|
6139 |
+
{
|
6140 |
+
"epoch": 0.17119325424061183,
|
6141 |
+
"grad_norm": 6.16123104095459,
|
6142 |
+
"learning_rate": 6.149499809660488e-05,
|
6143 |
+
"loss": 2.5358,
|
6144 |
+
"step": 873
|
6145 |
+
},
|
6146 |
+
{
|
6147 |
+
"epoch": 0.17138935189724483,
|
6148 |
+
"grad_norm": 6.115005016326904,
|
6149 |
+
"learning_rate": 6.128484712164164e-05,
|
6150 |
+
"loss": 1.6489,
|
6151 |
+
"step": 874
|
6152 |
+
},
|
6153 |
+
{
|
6154 |
+
"epoch": 0.17158544955387783,
|
6155 |
+
"grad_norm": 9.149282455444336,
|
6156 |
+
"learning_rate": 6.107489708001855e-05,
|
6157 |
+
"loss": 1.7977,
|
6158 |
+
"step": 875
|
6159 |
+
},
|
6160 |
+
{
|
6161 |
+
"epoch": 0.17178154721051084,
|
6162 |
+
"grad_norm": 7.190993309020996,
|
6163 |
+
"learning_rate": 6.086514906138563e-05,
|
6164 |
+
"loss": 1.3526,
|
6165 |
+
"step": 876
|
6166 |
+
},
|
6167 |
+
{
|
6168 |
+
"epoch": 0.17197764486714384,
|
6169 |
+
"grad_norm": 8.292972564697266,
|
6170 |
+
"learning_rate": 6.065560415434417e-05,
|
6171 |
+
"loss": 1.8477,
|
6172 |
+
"step": 877
|
6173 |
+
},
|
6174 |
+
{
|
6175 |
+
"epoch": 0.17217374252377685,
|
6176 |
+
"grad_norm": 8.858931541442871,
|
6177 |
+
"learning_rate": 6.044626344644151e-05,
|
6178 |
+
"loss": 1.3605,
|
6179 |
+
"step": 878
|
6180 |
+
},
|
6181 |
+
{
|
6182 |
+
"epoch": 0.17236984018040985,
|
6183 |
+
"grad_norm": 6.186944961547852,
|
6184 |
+
"learning_rate": 6.023712802416508e-05,
|
6185 |
+
"loss": 1.0484,
|
6186 |
+
"step": 879
|
6187 |
+
},
|
6188 |
+
{
|
6189 |
+
"epoch": 0.17256593783704285,
|
6190 |
+
"grad_norm": 9.702549934387207,
|
6191 |
+
"learning_rate": 6.0028198972936836e-05,
|
6192 |
+
"loss": 1.3224,
|
6193 |
+
"step": 880
|
6194 |
+
},
|
6195 |
+
{
|
6196 |
+
"epoch": 0.17276203549367586,
|
6197 |
+
"grad_norm": 12.195408821105957,
|
6198 |
+
"learning_rate": 5.981947737710779e-05,
|
6199 |
+
"loss": 1.8586,
|
6200 |
+
"step": 881
|
6201 |
+
},
|
6202 |
+
{
|
6203 |
+
"epoch": 0.17295813315030886,
|
6204 |
+
"grad_norm": 6.8144025802612305,
|
6205 |
+
"learning_rate": 5.9610964319952104e-05,
|
6206 |
+
"loss": 1.5967,
|
6207 |
+
"step": 882
|
6208 |
+
},
|
6209 |
+
{
|
6210 |
+
"epoch": 0.17315423080694187,
|
6211 |
+
"grad_norm": 10.37683391571045,
|
6212 |
+
"learning_rate": 5.940266088366173e-05,
|
6213 |
+
"loss": 1.7278,
|
6214 |
+
"step": 883
|
6215 |
+
},
|
6216 |
+
{
|
6217 |
+
"epoch": 0.17335032846357487,
|
6218 |
+
"grad_norm": 5.85037899017334,
|
6219 |
+
"learning_rate": 5.919456814934061e-05,
|
6220 |
+
"loss": 1.3165,
|
6221 |
+
"step": 884
|
6222 |
+
},
|
6223 |
+
{
|
6224 |
+
"epoch": 0.17354642612020788,
|
6225 |
+
"grad_norm": 7.210351943969727,
|
6226 |
+
"learning_rate": 5.8986687196999135e-05,
|
6227 |
+
"loss": 1.1156,
|
6228 |
+
"step": 885
|
6229 |
+
},
|
6230 |
+
{
|
6231 |
+
"epoch": 0.17374252377684088,
|
6232 |
+
"grad_norm": 5.840338706970215,
|
6233 |
+
"learning_rate": 5.877901910554862e-05,
|
6234 |
+
"loss": 1.4444,
|
6235 |
+
"step": 886
|
6236 |
+
},
|
6237 |
+
{
|
6238 |
+
"epoch": 0.17393862143347388,
|
6239 |
+
"grad_norm": 3.731435775756836,
|
6240 |
+
"learning_rate": 5.8571564952795475e-05,
|
6241 |
+
"loss": 0.9172,
|
6242 |
+
"step": 887
|
6243 |
+
},
|
6244 |
+
{
|
6245 |
+
"epoch": 0.17413471909010686,
|
6246 |
+
"grad_norm": 13.559743881225586,
|
6247 |
+
"learning_rate": 5.8364325815435916e-05,
|
6248 |
+
"loss": 2.1312,
|
6249 |
+
"step": 888
|
6250 |
+
},
|
6251 |
+
{
|
6252 |
+
"epoch": 0.17433081674673986,
|
6253 |
+
"grad_norm": 5.5318193435668945,
|
6254 |
+
"learning_rate": 5.815730276905014e-05,
|
6255 |
+
"loss": 1.735,
|
6256 |
+
"step": 889
|
6257 |
+
},
|
6258 |
+
{
|
6259 |
+
"epoch": 0.17452691440337287,
|
6260 |
+
"grad_norm": 7.40202522277832,
|
6261 |
+
"learning_rate": 5.7950496888096795e-05,
|
6262 |
+
"loss": 1.42,
|
6263 |
+
"step": 890
|
6264 |
+
},
|
6265 |
+
{
|
6266 |
+
"epoch": 0.17472301206000587,
|
6267 |
+
"grad_norm": 3.6923038959503174,
|
6268 |
+
"learning_rate": 5.7743909245907535e-05,
|
6269 |
+
"loss": 1.095,
|
6270 |
+
"step": 891
|
6271 |
+
},
|
6272 |
+
{
|
6273 |
+
"epoch": 0.17491910971663888,
|
6274 |
+
"grad_norm": 5.6912713050842285,
|
6275 |
+
"learning_rate": 5.753754091468115e-05,
|
6276 |
+
"loss": 2.2399,
|
6277 |
+
"step": 892
|
6278 |
+
},
|
6279 |
+
{
|
6280 |
+
"epoch": 0.17511520737327188,
|
6281 |
+
"grad_norm": 3.8538315296173096,
|
6282 |
+
"learning_rate": 5.7331392965478493e-05,
|
6283 |
+
"loss": 1.2043,
|
6284 |
+
"step": 893
|
6285 |
+
},
|
6286 |
+
{
|
6287 |
+
"epoch": 0.17531130502990488,
|
6288 |
+
"grad_norm": 5.220914840698242,
|
6289 |
+
"learning_rate": 5.712546646821627e-05,
|
6290 |
+
"loss": 1.0669,
|
6291 |
+
"step": 894
|
6292 |
+
},
|
6293 |
+
{
|
6294 |
+
"epoch": 0.1755074026865379,
|
6295 |
+
"grad_norm": 5.886853218078613,
|
6296 |
+
"learning_rate": 5.6919762491662164e-05,
|
6297 |
+
"loss": 1.3425,
|
6298 |
+
"step": 895
|
6299 |
+
},
|
6300 |
+
{
|
6301 |
+
"epoch": 0.1757035003431709,
|
6302 |
+
"grad_norm": 7.820699214935303,
|
6303 |
+
"learning_rate": 5.671428210342884e-05,
|
6304 |
+
"loss": 1.6338,
|
6305 |
+
"step": 896
|
6306 |
+
},
|
6307 |
+
{
|
6308 |
+
"epoch": 0.1758995979998039,
|
6309 |
+
"grad_norm": 9.36957836151123,
|
6310 |
+
"learning_rate": 5.650902636996837e-05,
|
6311 |
+
"loss": 1.1894,
|
6312 |
+
"step": 897
|
6313 |
+
},
|
6314 |
+
{
|
6315 |
+
"epoch": 0.1760956956564369,
|
6316 |
+
"grad_norm": 3.9427287578582764,
|
6317 |
+
"learning_rate": 5.6303996356567177e-05,
|
6318 |
+
"loss": 1.179,
|
6319 |
+
"step": 898
|
6320 |
+
},
|
6321 |
+
{
|
6322 |
+
"epoch": 0.1762917933130699,
|
6323 |
+
"grad_norm": 8.343499183654785,
|
6324 |
+
"learning_rate": 5.6099193127339864e-05,
|
6325 |
+
"loss": 1.9635,
|
6326 |
+
"step": 899
|
6327 |
+
},
|
6328 |
+
{
|
6329 |
+
"epoch": 0.1764878909697029,
|
6330 |
+
"grad_norm": 7.9742279052734375,
|
6331 |
+
"learning_rate": 5.589461774522433e-05,
|
6332 |
+
"loss": 1.6162,
|
6333 |
+
"step": 900
|
6334 |
+
},
|
6335 |
+
{
|
6336 |
+
"epoch": 0.1766839886263359,
|
6337 |
+
"grad_norm": 4.933746337890625,
|
6338 |
+
"learning_rate": 5.5690271271975644e-05,
|
6339 |
+
"loss": 1.0064,
|
6340 |
+
"step": 901
|
6341 |
+
},
|
6342 |
+
{
|
6343 |
+
"epoch": 0.17688008628296892,
|
6344 |
+
"grad_norm": 6.986325740814209,
|
6345 |
+
"learning_rate": 5.548615476816097e-05,
|
6346 |
+
"loss": 0.7052,
|
6347 |
+
"step": 902
|
6348 |
+
},
|
6349 |
+
{
|
6350 |
+
"epoch": 0.17707618393960192,
|
6351 |
+
"grad_norm": 5.324771881103516,
|
6352 |
+
"learning_rate": 5.528226929315401e-05,
|
6353 |
+
"loss": 1.5658,
|
6354 |
+
"step": 903
|
6355 |
+
},
|
6356 |
+
{
|
6357 |
+
"epoch": 0.17727228159623493,
|
6358 |
+
"grad_norm": 3.9526147842407227,
|
6359 |
+
"learning_rate": 5.507861590512916e-05,
|
6360 |
+
"loss": 1.0183,
|
6361 |
+
"step": 904
|
6362 |
+
},
|
6363 |
+
{
|
6364 |
+
"epoch": 0.17746837925286793,
|
6365 |
+
"grad_norm": 8.242249488830566,
|
6366 |
+
"learning_rate": 5.48751956610566e-05,
|
6367 |
+
"loss": 1.9991,
|
6368 |
+
"step": 905
|
6369 |
+
},
|
6370 |
+
{
|
6371 |
+
"epoch": 0.17766447690950093,
|
6372 |
+
"grad_norm": 6.519176483154297,
|
6373 |
+
"learning_rate": 5.467200961669619e-05,
|
6374 |
+
"loss": 1.2312,
|
6375 |
+
"step": 906
|
6376 |
+
},
|
6377 |
+
{
|
6378 |
+
"epoch": 0.17786057456613394,
|
6379 |
+
"grad_norm": 3.49365496635437,
|
6380 |
+
"learning_rate": 5.446905882659243e-05,
|
6381 |
+
"loss": 0.5029,
|
6382 |
+
"step": 907
|
6383 |
+
},
|
6384 |
+
{
|
6385 |
+
"epoch": 0.17805667222276694,
|
6386 |
+
"grad_norm": 5.808468818664551,
|
6387 |
+
"learning_rate": 5.426634434406883e-05,
|
6388 |
+
"loss": 1.1932,
|
6389 |
+
"step": 908
|
6390 |
+
},
|
6391 |
+
{
|
6392 |
+
"epoch": 0.17825276987939995,
|
6393 |
+
"grad_norm": 6.1657328605651855,
|
6394 |
+
"learning_rate": 5.40638672212224e-05,
|
6395 |
+
"loss": 1.0127,
|
6396 |
+
"step": 909
|
6397 |
+
},
|
6398 |
+
{
|
6399 |
+
"epoch": 0.17844886753603295,
|
6400 |
+
"grad_norm": 8.090532302856445,
|
6401 |
+
"learning_rate": 5.3861628508918384e-05,
|
6402 |
+
"loss": 1.0638,
|
6403 |
+
"step": 910
|
6404 |
+
},
|
6405 |
+
{
|
6406 |
+
"epoch": 0.17864496519266596,
|
6407 |
+
"grad_norm": 6.026912689208984,
|
6408 |
+
"learning_rate": 5.3659629256784424e-05,
|
6409 |
+
"loss": 1.1598,
|
6410 |
+
"step": 911
|
6411 |
+
},
|
6412 |
+
{
|
6413 |
+
"epoch": 0.17884106284929896,
|
6414 |
+
"grad_norm": 7.103152751922607,
|
6415 |
+
"learning_rate": 5.345787051320564e-05,
|
6416 |
+
"loss": 1.9001,
|
6417 |
+
"step": 912
|
6418 |
+
},
|
6419 |
+
{
|
6420 |
+
"epoch": 0.17903716050593196,
|
6421 |
+
"grad_norm": 4.5661163330078125,
|
6422 |
+
"learning_rate": 5.325635332531864e-05,
|
6423 |
+
"loss": 1.2711,
|
6424 |
+
"step": 913
|
6425 |
+
},
|
6426 |
+
{
|
6427 |
+
"epoch": 0.17923325816256497,
|
6428 |
+
"grad_norm": 6.035799980163574,
|
6429 |
+
"learning_rate": 5.305507873900649e-05,
|
6430 |
+
"loss": 2.3649,
|
6431 |
+
"step": 914
|
6432 |
+
},
|
6433 |
+
{
|
6434 |
+
"epoch": 0.17942935581919797,
|
6435 |
+
"grad_norm": 6.231631278991699,
|
6436 |
+
"learning_rate": 5.2854047798893125e-05,
|
6437 |
+
"loss": 2.7058,
|
6438 |
+
"step": 915
|
6439 |
+
},
|
6440 |
+
{
|
6441 |
+
"epoch": 0.17962545347583098,
|
6442 |
+
"grad_norm": 9.146753311157227,
|
6443 |
+
"learning_rate": 5.26532615483379e-05,
|
6444 |
+
"loss": 1.1388,
|
6445 |
+
"step": 916
|
6446 |
+
},
|
6447 |
+
{
|
6448 |
+
"epoch": 0.17982155113246398,
|
6449 |
+
"grad_norm": 8.76220417022705,
|
6450 |
+
"learning_rate": 5.245272102943034e-05,
|
6451 |
+
"loss": 1.0848,
|
6452 |
+
"step": 917
|
6453 |
+
},
|
6454 |
+
{
|
6455 |
+
"epoch": 0.18001764878909696,
|
6456 |
+
"grad_norm": 6.711489200592041,
|
6457 |
+
"learning_rate": 5.225242728298445e-05,
|
6458 |
+
"loss": 1.0747,
|
6459 |
+
"step": 918
|
6460 |
+
},
|
6461 |
+
{
|
6462 |
+
"epoch": 0.18021374644572996,
|
6463 |
+
"grad_norm": 4.66898775100708,
|
6464 |
+
"learning_rate": 5.2052381348533564e-05,
|
6465 |
+
"loss": 0.6995,
|
6466 |
+
"step": 919
|
6467 |
+
},
|
6468 |
+
{
|
6469 |
+
"epoch": 0.18040984410236297,
|
6470 |
+
"grad_norm": 6.504391193389893,
|
6471 |
+
"learning_rate": 5.1852584264324866e-05,
|
6472 |
+
"loss": 1.5352,
|
6473 |
+
"step": 920
|
6474 |
+
},
|
6475 |
+
{
|
6476 |
+
"epoch": 0.18060594175899597,
|
6477 |
+
"grad_norm": 6.724686145782471,
|
6478 |
+
"learning_rate": 5.165303706731397e-05,
|
6479 |
+
"loss": 1.2562,
|
6480 |
+
"step": 921
|
6481 |
+
},
|
6482 |
+
{
|
6483 |
+
"epoch": 0.18080203941562897,
|
6484 |
+
"grad_norm": 7.192296981811523,
|
6485 |
+
"learning_rate": 5.1453740793159586e-05,
|
6486 |
+
"loss": 1.0158,
|
6487 |
+
"step": 922
|
6488 |
+
},
|
6489 |
+
{
|
6490 |
+
"epoch": 0.18099813707226198,
|
6491 |
+
"grad_norm": 5.173651218414307,
|
6492 |
+
"learning_rate": 5.12546964762181e-05,
|
6493 |
+
"loss": 1.3406,
|
6494 |
+
"step": 923
|
6495 |
+
},
|
6496 |
+
{
|
6497 |
+
"epoch": 0.18119423472889498,
|
6498 |
+
"grad_norm": 8.315585136413574,
|
6499 |
+
"learning_rate": 5.105590514953824e-05,
|
6500 |
+
"loss": 2.0057,
|
6501 |
+
"step": 924
|
6502 |
+
},
|
6503 |
+
{
|
6504 |
+
"epoch": 0.18139033238552799,
|
6505 |
+
"grad_norm": 4.940390110015869,
|
6506 |
+
"learning_rate": 5.085736784485571e-05,
|
6507 |
+
"loss": 1.6468,
|
6508 |
+
"step": 925
|
6509 |
+
},
|
6510 |
+
{
|
6511 |
+
"epoch": 0.181586430042161,
|
6512 |
+
"grad_norm": 6.491610527038574,
|
6513 |
+
"learning_rate": 5.065908559258782e-05,
|
6514 |
+
"loss": 2.6567,
|
6515 |
+
"step": 926
|
6516 |
+
},
|
6517 |
+
{
|
6518 |
+
"epoch": 0.181782527698794,
|
6519 |
+
"grad_norm": 8.9893798828125,
|
6520 |
+
"learning_rate": 5.046105942182815e-05,
|
6521 |
+
"loss": 1.2412,
|
6522 |
+
"step": 927
|
6523 |
+
},
|
6524 |
+
{
|
6525 |
+
"epoch": 0.181978625355427,
|
6526 |
+
"grad_norm": 8.103008270263672,
|
6527 |
+
"learning_rate": 5.026329036034119e-05,
|
6528 |
+
"loss": 2.1319,
|
6529 |
+
"step": 928
|
6530 |
+
},
|
6531 |
+
{
|
6532 |
+
"epoch": 0.18217472301206,
|
6533 |
+
"grad_norm": 8.022109985351562,
|
6534 |
+
"learning_rate": 5.006577943455706e-05,
|
6535 |
+
"loss": 1.4119,
|
6536 |
+
"step": 929
|
6537 |
+
},
|
6538 |
+
{
|
6539 |
+
"epoch": 0.182370820668693,
|
6540 |
+
"grad_norm": 12.814486503601074,
|
6541 |
+
"learning_rate": 4.9868527669566113e-05,
|
6542 |
+
"loss": 1.9388,
|
6543 |
+
"step": 930
|
6544 |
+
},
|
6545 |
+
{
|
6546 |
+
"epoch": 0.182566918325326,
|
6547 |
+
"grad_norm": 7.568852424621582,
|
6548 |
+
"learning_rate": 4.967153608911366e-05,
|
6549 |
+
"loss": 1.2114,
|
6550 |
+
"step": 931
|
6551 |
+
},
|
6552 |
+
{
|
6553 |
+
"epoch": 0.18276301598195901,
|
6554 |
+
"grad_norm": 3.7613790035247803,
|
6555 |
+
"learning_rate": 4.947480571559462e-05,
|
6556 |
+
"loss": 1.2221,
|
6557 |
+
"step": 932
|
6558 |
+
},
|
6559 |
+
{
|
6560 |
+
"epoch": 0.18295911363859202,
|
6561 |
+
"grad_norm": 8.301228523254395,
|
6562 |
+
"learning_rate": 4.927833757004826e-05,
|
6563 |
+
"loss": 1.2867,
|
6564 |
+
"step": 933
|
6565 |
+
},
|
6566 |
+
{
|
6567 |
+
"epoch": 0.18315521129522502,
|
6568 |
+
"grad_norm": 5.2592244148254395,
|
6569 |
+
"learning_rate": 4.908213267215287e-05,
|
6570 |
+
"loss": 2.0937,
|
6571 |
+
"step": 934
|
6572 |
+
},
|
6573 |
+
{
|
6574 |
+
"epoch": 0.18335130895185803,
|
6575 |
+
"grad_norm": 6.635796546936035,
|
6576 |
+
"learning_rate": 4.888619204022047e-05,
|
6577 |
+
"loss": 1.9856,
|
6578 |
+
"step": 935
|
6579 |
+
},
|
6580 |
+
{
|
6581 |
+
"epoch": 0.18354740660849103,
|
6582 |
+
"grad_norm": 5.507410049438477,
|
6583 |
+
"learning_rate": 4.869051669119154e-05,
|
6584 |
+
"loss": 1.2305,
|
6585 |
+
"step": 936
|
6586 |
+
},
|
6587 |
+
{
|
6588 |
+
"epoch": 0.18374350426512404,
|
6589 |
+
"grad_norm": 8.553701400756836,
|
6590 |
+
"learning_rate": 4.8495107640629675e-05,
|
6591 |
+
"loss": 1.6872,
|
6592 |
+
"step": 937
|
6593 |
+
},
|
6594 |
+
{
|
6595 |
+
"epoch": 0.18393960192175704,
|
6596 |
+
"grad_norm": 7.652828693389893,
|
6597 |
+
"learning_rate": 4.829996590271646e-05,
|
6598 |
+
"loss": 1.4061,
|
6599 |
+
"step": 938
|
6600 |
+
},
|
6601 |
+
{
|
6602 |
+
"epoch": 0.18413569957839004,
|
6603 |
+
"grad_norm": 6.595338821411133,
|
6604 |
+
"learning_rate": 4.810509249024604e-05,
|
6605 |
+
"loss": 1.2953,
|
6606 |
+
"step": 939
|
6607 |
+
},
|
6608 |
+
{
|
6609 |
+
"epoch": 0.18433179723502305,
|
6610 |
+
"grad_norm": 4.507275581359863,
|
6611 |
+
"learning_rate": 4.7910488414619967e-05,
|
6612 |
+
"loss": 0.95,
|
6613 |
+
"step": 940
|
6614 |
+
},
|
6615 |
+
{
|
6616 |
+
"epoch": 0.18452789489165605,
|
6617 |
+
"grad_norm": 7.35803747177124,
|
6618 |
+
"learning_rate": 4.7716154685841944e-05,
|
6619 |
+
"loss": 1.7958,
|
6620 |
+
"step": 941
|
6621 |
+
},
|
6622 |
+
{
|
6623 |
+
"epoch": 0.18472399254828906,
|
6624 |
+
"grad_norm": 5.53951358795166,
|
6625 |
+
"learning_rate": 4.752209231251251e-05,
|
6626 |
+
"loss": 2.8191,
|
6627 |
+
"step": 942
|
6628 |
+
},
|
6629 |
+
{
|
6630 |
+
"epoch": 0.18492009020492206,
|
6631 |
+
"grad_norm": 4.738945960998535,
|
6632 |
+
"learning_rate": 4.73283023018239e-05,
|
6633 |
+
"loss": 1.7728,
|
6634 |
+
"step": 943
|
6635 |
+
},
|
6636 |
+
{
|
6637 |
+
"epoch": 0.18511618786155506,
|
6638 |
+
"grad_norm": 5.192052364349365,
|
6639 |
+
"learning_rate": 4.713478565955478e-05,
|
6640 |
+
"loss": 0.8087,
|
6641 |
+
"step": 944
|
6642 |
+
},
|
6643 |
+
{
|
6644 |
+
"epoch": 0.18531228551818807,
|
6645 |
+
"grad_norm": 5.9963507652282715,
|
6646 |
+
"learning_rate": 4.694154339006501e-05,
|
6647 |
+
"loss": 1.1223,
|
6648 |
+
"step": 945
|
6649 |
+
},
|
6650 |
+
{
|
6651 |
+
"epoch": 0.18550838317482107,
|
6652 |
+
"grad_norm": 14.023632049560547,
|
6653 |
+
"learning_rate": 4.6748576496290356e-05,
|
6654 |
+
"loss": 3.0858,
|
6655 |
+
"step": 946
|
6656 |
+
},
|
6657 |
+
{
|
6658 |
+
"epoch": 0.18570448083145408,
|
6659 |
+
"grad_norm": 6.268606662750244,
|
6660 |
+
"learning_rate": 4.655588597973754e-05,
|
6661 |
+
"loss": 0.8988,
|
6662 |
+
"step": 947
|
6663 |
+
},
|
6664 |
+
{
|
6665 |
+
"epoch": 0.18590057848808705,
|
6666 |
+
"grad_norm": 7.719700336456299,
|
6667 |
+
"learning_rate": 4.636347284047877e-05,
|
6668 |
+
"loss": 1.5117,
|
6669 |
+
"step": 948
|
6670 |
+
},
|
6671 |
+
{
|
6672 |
+
"epoch": 0.18609667614472006,
|
6673 |
+
"grad_norm": 4.05161714553833,
|
6674 |
+
"learning_rate": 4.617133807714666e-05,
|
6675 |
+
"loss": 0.7625,
|
6676 |
+
"step": 949
|
6677 |
+
},
|
6678 |
+
{
|
6679 |
+
"epoch": 0.18629277380135306,
|
6680 |
+
"grad_norm": 5.103604793548584,
|
6681 |
+
"learning_rate": 4.59794826869291e-05,
|
6682 |
+
"loss": 0.8914,
|
6683 |
+
"step": 950
|
6684 |
+
},
|
6685 |
+
{
|
6686 |
+
"epoch": 0.18648887145798607,
|
6687 |
+
"grad_norm": 7.7601847648620605,
|
6688 |
+
"learning_rate": 4.578790766556386e-05,
|
6689 |
+
"loss": 0.8747,
|
6690 |
+
"step": 951
|
6691 |
+
},
|
6692 |
+
{
|
6693 |
+
"epoch": 0.18668496911461907,
|
6694 |
+
"grad_norm": 12.439096450805664,
|
6695 |
+
"learning_rate": 4.559661400733383e-05,
|
6696 |
+
"loss": 2.2439,
|
6697 |
+
"step": 952
|
6698 |
+
},
|
6699 |
+
{
|
6700 |
+
"epoch": 0.18688106677125207,
|
6701 |
+
"grad_norm": 7.002286911010742,
|
6702 |
+
"learning_rate": 4.5405602705061345e-05,
|
6703 |
+
"loss": 1.4917,
|
6704 |
+
"step": 953
|
6705 |
+
},
|
6706 |
+
{
|
6707 |
+
"epoch": 0.18707716442788508,
|
6708 |
+
"grad_norm": 5.178001880645752,
|
6709 |
+
"learning_rate": 4.521487475010354e-05,
|
6710 |
+
"loss": 1.5737,
|
6711 |
+
"step": 954
|
6712 |
+
},
|
6713 |
+
{
|
6714 |
+
"epoch": 0.18727326208451808,
|
6715 |
+
"grad_norm": 9.555988311767578,
|
6716 |
+
"learning_rate": 4.502443113234688e-05,
|
6717 |
+
"loss": 1.7495,
|
6718 |
+
"step": 955
|
6719 |
+
},
|
6720 |
+
{
|
6721 |
+
"epoch": 0.1874693597411511,
|
6722 |
+
"grad_norm": 8.907003402709961,
|
6723 |
+
"learning_rate": 4.483427284020194e-05,
|
6724 |
+
"loss": 1.6053,
|
6725 |
+
"step": 956
|
6726 |
+
},
|
6727 |
+
{
|
6728 |
+
"epoch": 0.1876654573977841,
|
6729 |
+
"grad_norm": 5.562190532684326,
|
6730 |
+
"learning_rate": 4.464440086059878e-05,
|
6731 |
+
"loss": 1.1854,
|
6732 |
+
"step": 957
|
6733 |
+
},
|
6734 |
+
{
|
6735 |
+
"epoch": 0.1878615550544171,
|
6736 |
+
"grad_norm": 6.078771591186523,
|
6737 |
+
"learning_rate": 4.4454816178981115e-05,
|
6738 |
+
"loss": 1.3774,
|
6739 |
+
"step": 958
|
6740 |
+
},
|
6741 |
+
{
|
6742 |
+
"epoch": 0.1880576527110501,
|
6743 |
+
"grad_norm": 12.021577835083008,
|
6744 |
+
"learning_rate": 4.426551977930191e-05,
|
6745 |
+
"loss": 1.6445,
|
6746 |
+
"step": 959
|
6747 |
+
},
|
6748 |
+
{
|
6749 |
+
"epoch": 0.1882537503676831,
|
6750 |
+
"grad_norm": 7.406763553619385,
|
6751 |
+
"learning_rate": 4.407651264401763e-05,
|
6752 |
+
"loss": 2.0784,
|
6753 |
+
"step": 960
|
6754 |
+
},
|
6755 |
+
{
|
6756 |
+
"epoch": 0.1884498480243161,
|
6757 |
+
"grad_norm": 3.961423635482788,
|
6758 |
+
"learning_rate": 4.38877957540837e-05,
|
6759 |
+
"loss": 1.5526,
|
6760 |
+
"step": 961
|
6761 |
+
},
|
6762 |
+
{
|
6763 |
+
"epoch": 0.1886459456809491,
|
6764 |
+
"grad_norm": 7.410959243774414,
|
6765 |
+
"learning_rate": 4.3699370088949066e-05,
|
6766 |
+
"loss": 2.0247,
|
6767 |
+
"step": 962
|
6768 |
+
},
|
6769 |
+
{
|
6770 |
+
"epoch": 0.18884204333758212,
|
6771 |
+
"grad_norm": 5.237401485443115,
|
6772 |
+
"learning_rate": 4.3511236626551047e-05,
|
6773 |
+
"loss": 2.199,
|
6774 |
+
"step": 963
|
6775 |
+
},
|
6776 |
+
{
|
6777 |
+
"epoch": 0.18903814099421512,
|
6778 |
+
"grad_norm": 4.068509578704834,
|
6779 |
+
"learning_rate": 4.3323396343310715e-05,
|
6780 |
+
"loss": 2.3318,
|
6781 |
+
"step": 964
|
6782 |
+
},
|
6783 |
+
{
|
6784 |
+
"epoch": 0.18923423865084812,
|
6785 |
+
"grad_norm": 5.523316383361816,
|
6786 |
+
"learning_rate": 4.313585021412724e-05,
|
6787 |
+
"loss": 1.3474,
|
6788 |
+
"step": 965
|
6789 |
+
},
|
6790 |
+
{
|
6791 |
+
"epoch": 0.18943033630748113,
|
6792 |
+
"grad_norm": 5.052401542663574,
|
6793 |
+
"learning_rate": 4.294859921237339e-05,
|
6794 |
+
"loss": 1.1039,
|
6795 |
+
"step": 966
|
6796 |
+
},
|
6797 |
+
{
|
6798 |
+
"epoch": 0.18962643396411413,
|
6799 |
+
"grad_norm": 8.468581199645996,
|
6800 |
+
"learning_rate": 4.2761644309889946e-05,
|
6801 |
+
"loss": 2.9943,
|
6802 |
+
"step": 967
|
6803 |
+
},
|
6804 |
+
{
|
6805 |
+
"epoch": 0.18982253162074714,
|
6806 |
+
"grad_norm": 4.66163444519043,
|
6807 |
+
"learning_rate": 4.257498647698107e-05,
|
6808 |
+
"loss": 2.0152,
|
6809 |
+
"step": 968
|
6810 |
+
},
|
6811 |
+
{
|
6812 |
+
"epoch": 0.19001862927738014,
|
6813 |
+
"grad_norm": 4.58842658996582,
|
6814 |
+
"learning_rate": 4.2388626682409194e-05,
|
6815 |
+
"loss": 1.2759,
|
6816 |
+
"step": 969
|
6817 |
+
},
|
6818 |
+
{
|
6819 |
+
"epoch": 0.19021472693401315,
|
6820 |
+
"grad_norm": 3.5642545223236084,
|
6821 |
+
"learning_rate": 4.220256589338968e-05,
|
6822 |
+
"loss": 0.9877,
|
6823 |
+
"step": 970
|
6824 |
+
},
|
6825 |
+
{
|
6826 |
+
"epoch": 0.19041082459064615,
|
6827 |
+
"grad_norm": 6.506111145019531,
|
6828 |
+
"learning_rate": 4.201680507558631e-05,
|
6829 |
+
"loss": 1.7171,
|
6830 |
+
"step": 971
|
6831 |
+
},
|
6832 |
+
{
|
6833 |
+
"epoch": 0.19060692224727915,
|
6834 |
+
"grad_norm": 4.93696928024292,
|
6835 |
+
"learning_rate": 4.183134519310576e-05,
|
6836 |
+
"loss": 0.8798,
|
6837 |
+
"step": 972
|
6838 |
+
},
|
6839 |
+
{
|
6840 |
+
"epoch": 0.19080301990391216,
|
6841 |
+
"grad_norm": 12.539702415466309,
|
6842 |
+
"learning_rate": 4.1646187208493005e-05,
|
6843 |
+
"loss": 3.0454,
|
6844 |
+
"step": 973
|
6845 |
+
},
|
6846 |
+
{
|
6847 |
+
"epoch": 0.19099911756054516,
|
6848 |
+
"grad_norm": 9.189221382141113,
|
6849 |
+
"learning_rate": 4.146133208272608e-05,
|
6850 |
+
"loss": 1.9269,
|
6851 |
+
"step": 974
|
6852 |
+
},
|
6853 |
+
{
|
6854 |
+
"epoch": 0.19119521521717817,
|
6855 |
+
"grad_norm": 5.362178802490234,
|
6856 |
+
"learning_rate": 4.1276780775211156e-05,
|
6857 |
+
"loss": 1.7086,
|
6858 |
+
"step": 975
|
6859 |
+
},
|
6860 |
+
{
|
6861 |
+
"epoch": 0.19139131287381117,
|
6862 |
+
"grad_norm": 4.587559700012207,
|
6863 |
+
"learning_rate": 4.109253424377772e-05,
|
6864 |
+
"loss": 0.7529,
|
6865 |
+
"step": 976
|
6866 |
+
},
|
6867 |
+
{
|
6868 |
+
"epoch": 0.19158741053044417,
|
6869 |
+
"grad_norm": 7.922691345214844,
|
6870 |
+
"learning_rate": 4.090859344467325e-05,
|
6871 |
+
"loss": 2.6641,
|
6872 |
+
"step": 977
|
6873 |
+
},
|
6874 |
+
{
|
6875 |
+
"epoch": 0.19178350818707715,
|
6876 |
+
"grad_norm": 6.759201526641846,
|
6877 |
+
"learning_rate": 4.072495933255857e-05,
|
6878 |
+
"loss": 1.3197,
|
6879 |
+
"step": 978
|
6880 |
+
},
|
6881 |
+
{
|
6882 |
+
"epoch": 0.19197960584371015,
|
6883 |
+
"grad_norm": 5.635506629943848,
|
6884 |
+
"learning_rate": 4.054163286050276e-05,
|
6885 |
+
"loss": 0.8712,
|
6886 |
+
"step": 979
|
6887 |
+
},
|
6888 |
+
{
|
6889 |
+
"epoch": 0.19217570350034316,
|
6890 |
+
"grad_norm": 4.1558637619018555,
|
6891 |
+
"learning_rate": 4.035861497997828e-05,
|
6892 |
+
"loss": 2.3436,
|
6893 |
+
"step": 980
|
6894 |
+
},
|
6895 |
+
{
|
6896 |
+
"epoch": 0.19237180115697616,
|
6897 |
+
"grad_norm": 9.821369171142578,
|
6898 |
+
"learning_rate": 4.017590664085593e-05,
|
6899 |
+
"loss": 1.6202,
|
6900 |
+
"step": 981
|
6901 |
+
},
|
6902 |
+
{
|
6903 |
+
"epoch": 0.19256789881360917,
|
6904 |
+
"grad_norm": 7.816912651062012,
|
6905 |
+
"learning_rate": 3.999350879139997e-05,
|
6906 |
+
"loss": 1.1312,
|
6907 |
+
"step": 982
|
6908 |
+
},
|
6909 |
+
{
|
6910 |
+
"epoch": 0.19276399647024217,
|
6911 |
+
"grad_norm": 5.585190296173096,
|
6912 |
+
"learning_rate": 3.981142237826332e-05,
|
6913 |
+
"loss": 1.9218,
|
6914 |
+
"step": 983
|
6915 |
+
},
|
6916 |
+
{
|
6917 |
+
"epoch": 0.19296009412687518,
|
6918 |
+
"grad_norm": 4.785935878753662,
|
6919 |
+
"learning_rate": 3.962964834648236e-05,
|
6920 |
+
"loss": 0.8693,
|
6921 |
+
"step": 984
|
6922 |
+
},
|
6923 |
+
{
|
6924 |
+
"epoch": 0.19315619178350818,
|
6925 |
+
"grad_norm": 5.3974385261535645,
|
6926 |
+
"learning_rate": 3.944818763947231e-05,
|
6927 |
+
"loss": 1.1065,
|
6928 |
+
"step": 985
|
6929 |
+
},
|
6930 |
+
{
|
6931 |
+
"epoch": 0.19335228944014118,
|
6932 |
+
"grad_norm": 9.581192016601562,
|
6933 |
+
"learning_rate": 3.926704119902219e-05,
|
6934 |
+
"loss": 0.867,
|
6935 |
+
"step": 986
|
6936 |
+
},
|
6937 |
+
{
|
6938 |
+
"epoch": 0.1935483870967742,
|
6939 |
+
"grad_norm": 10.229225158691406,
|
6940 |
+
"learning_rate": 3.9086209965289965e-05,
|
6941 |
+
"loss": 2.8435,
|
6942 |
+
"step": 987
|
6943 |
+
},
|
6944 |
+
{
|
6945 |
+
"epoch": 0.1937444847534072,
|
6946 |
+
"grad_norm": 7.257898807525635,
|
6947 |
+
"learning_rate": 3.890569487679766e-05,
|
6948 |
+
"loss": 2.4263,
|
6949 |
+
"step": 988
|
6950 |
+
},
|
6951 |
+
{
|
6952 |
+
"epoch": 0.1939405824100402,
|
6953 |
+
"grad_norm": 5.026792526245117,
|
6954 |
+
"learning_rate": 3.87254968704265e-05,
|
6955 |
+
"loss": 1.2831,
|
6956 |
+
"step": 989
|
6957 |
+
},
|
6958 |
+
{
|
6959 |
+
"epoch": 0.1941366800666732,
|
6960 |
+
"grad_norm": 5.230473041534424,
|
6961 |
+
"learning_rate": 3.854561688141205e-05,
|
6962 |
+
"loss": 3.3858,
|
6963 |
+
"step": 990
|
6964 |
+
},
|
6965 |
+
{
|
6966 |
+
"epoch": 0.1943327777233062,
|
6967 |
+
"grad_norm": 6.340061187744141,
|
6968 |
+
"learning_rate": 3.836605584333931e-05,
|
6969 |
+
"loss": 1.5421,
|
6970 |
+
"step": 991
|
6971 |
+
},
|
6972 |
+
{
|
6973 |
+
"epoch": 0.1945288753799392,
|
6974 |
+
"grad_norm": 7.890949726104736,
|
6975 |
+
"learning_rate": 3.818681468813794e-05,
|
6976 |
+
"loss": 1.7882,
|
6977 |
+
"step": 992
|
6978 |
+
},
|
6979 |
+
{
|
6980 |
+
"epoch": 0.1947249730365722,
|
6981 |
+
"grad_norm": 4.434299945831299,
|
6982 |
+
"learning_rate": 3.800789434607741e-05,
|
6983 |
+
"loss": 1.274,
|
6984 |
+
"step": 993
|
6985 |
+
},
|
6986 |
+
{
|
6987 |
+
"epoch": 0.19492107069320522,
|
6988 |
+
"grad_norm": 5.311686038970947,
|
6989 |
+
"learning_rate": 3.782929574576213e-05,
|
6990 |
+
"loss": 1.7605,
|
6991 |
+
"step": 994
|
6992 |
+
},
|
6993 |
+
{
|
6994 |
+
"epoch": 0.19511716834983822,
|
6995 |
+
"grad_norm": 22.297441482543945,
|
6996 |
+
"learning_rate": 3.7651019814126654e-05,
|
6997 |
+
"loss": 1.9094,
|
6998 |
+
"step": 995
|
6999 |
+
},
|
7000 |
+
{
|
7001 |
+
"epoch": 0.19531326600647123,
|
7002 |
+
"grad_norm": 4.7996826171875,
|
7003 |
+
"learning_rate": 3.747306747643089e-05,
|
7004 |
+
"loss": 1.7196,
|
7005 |
+
"step": 996
|
7006 |
+
},
|
7007 |
+
{
|
7008 |
+
"epoch": 0.19550936366310423,
|
7009 |
+
"grad_norm": 12.06876277923584,
|
7010 |
+
"learning_rate": 3.729543965625526e-05,
|
7011 |
+
"loss": 1.2351,
|
7012 |
+
"step": 997
|
7013 |
+
},
|
7014 |
+
{
|
7015 |
+
"epoch": 0.19570546131973723,
|
7016 |
+
"grad_norm": 5.477703094482422,
|
7017 |
+
"learning_rate": 3.711813727549594e-05,
|
7018 |
+
"loss": 1.3366,
|
7019 |
+
"step": 998
|
7020 |
+
},
|
7021 |
+
{
|
7022 |
+
"epoch": 0.19590155897637024,
|
7023 |
+
"grad_norm": 4.249327659606934,
|
7024 |
+
"learning_rate": 3.694116125436007e-05,
|
7025 |
+
"loss": 1.9066,
|
7026 |
+
"step": 999
|
7027 |
+
},
|
7028 |
+
{
|
7029 |
+
"epoch": 0.19609765663300324,
|
7030 |
+
"grad_norm": 7.70191764831543,
|
7031 |
+
"learning_rate": 3.6764512511360935e-05,
|
7032 |
+
"loss": 1.3611,
|
7033 |
+
"step": 1000
|
7034 |
+
},
|
7035 |
+
{
|
7036 |
+
"epoch": 0.19629375428963625,
|
7037 |
+
"grad_norm": 5.822709560394287,
|
7038 |
+
"learning_rate": 3.658819196331327e-05,
|
7039 |
+
"loss": 2.3203,
|
7040 |
+
"step": 1001
|
7041 |
+
},
|
7042 |
+
{
|
7043 |
+
"epoch": 0.19648985194626925,
|
7044 |
+
"grad_norm": 4.531265735626221,
|
7045 |
+
"learning_rate": 3.6412200525328435e-05,
|
7046 |
+
"loss": 0.7644,
|
7047 |
+
"step": 1002
|
7048 |
+
},
|
7049 |
+
{
|
7050 |
+
"epoch": 0.19668594960290225,
|
7051 |
+
"grad_norm": 4.714295387268066,
|
7052 |
+
"learning_rate": 3.623653911080971e-05,
|
7053 |
+
"loss": 1.6555,
|
7054 |
+
"step": 1003
|
7055 |
+
},
|
7056 |
+
{
|
7057 |
+
"epoch": 0.19688204725953526,
|
7058 |
+
"grad_norm": 7.089095592498779,
|
7059 |
+
"learning_rate": 3.606120863144753e-05,
|
7060 |
+
"loss": 1.6242,
|
7061 |
+
"step": 1004
|
7062 |
+
},
|
7063 |
+
{
|
7064 |
+
"epoch": 0.19707814491616826,
|
7065 |
+
"grad_norm": 2.268071174621582,
|
7066 |
+
"learning_rate": 3.588620999721477e-05,
|
7067 |
+
"loss": 0.3747,
|
7068 |
+
"step": 1005
|
7069 |
+
},
|
7070 |
+
{
|
7071 |
+
"epoch": 0.19727424257280127,
|
7072 |
+
"grad_norm": 9.244414329528809,
|
7073 |
+
"learning_rate": 3.571154411636203e-05,
|
7074 |
+
"loss": 1.8298,
|
7075 |
+
"step": 1006
|
7076 |
+
},
|
7077 |
+
{
|
7078 |
+
"epoch": 0.19747034022943427,
|
7079 |
+
"grad_norm": 4.8095808029174805,
|
7080 |
+
"learning_rate": 3.5537211895412846e-05,
|
7081 |
+
"loss": 0.8923,
|
7082 |
+
"step": 1007
|
7083 |
+
},
|
7084 |
+
{
|
7085 |
+
"epoch": 0.19766643788606725,
|
7086 |
+
"grad_norm": 7.700345516204834,
|
7087 |
+
"learning_rate": 3.536321423915913e-05,
|
7088 |
+
"loss": 0.691,
|
7089 |
+
"step": 1008
|
7090 |
+
},
|
7091 |
+
{
|
7092 |
+
"epoch": 0.19786253554270025,
|
7093 |
+
"grad_norm": 7.553009510040283,
|
7094 |
+
"learning_rate": 3.518955205065632e-05,
|
7095 |
+
"loss": 1.1634,
|
7096 |
+
"step": 1009
|
7097 |
+
},
|
7098 |
+
{
|
7099 |
+
"epoch": 0.19805863319933326,
|
7100 |
+
"grad_norm": 5.901485443115234,
|
7101 |
+
"learning_rate": 3.5016226231218774e-05,
|
7102 |
+
"loss": 2.3853,
|
7103 |
+
"step": 1010
|
7104 |
+
},
|
7105 |
+
{
|
7106 |
+
"epoch": 0.19825473085596626,
|
7107 |
+
"grad_norm": 5.342306137084961,
|
7108 |
+
"learning_rate": 3.4843237680415156e-05,
|
7109 |
+
"loss": 0.9747,
|
7110 |
+
"step": 1011
|
7111 |
+
},
|
7112 |
+
{
|
7113 |
+
"epoch": 0.19845082851259926,
|
7114 |
+
"grad_norm": 5.793070316314697,
|
7115 |
+
"learning_rate": 3.46705872960635e-05,
|
7116 |
+
"loss": 0.7877,
|
7117 |
+
"step": 1012
|
7118 |
+
},
|
7119 |
+
{
|
7120 |
+
"epoch": 0.19864692616923227,
|
7121 |
+
"grad_norm": 9.105026245117188,
|
7122 |
+
"learning_rate": 3.449827597422698e-05,
|
7123 |
+
"loss": 1.5736,
|
7124 |
+
"step": 1013
|
7125 |
+
},
|
7126 |
+
{
|
7127 |
+
"epoch": 0.19884302382586527,
|
7128 |
+
"grad_norm": 7.359753608703613,
|
7129 |
+
"learning_rate": 3.432630460920887e-05,
|
7130 |
+
"loss": 1.4213,
|
7131 |
+
"step": 1014
|
7132 |
+
},
|
7133 |
+
{
|
7134 |
+
"epoch": 0.19903912148249828,
|
7135 |
+
"grad_norm": 4.728425025939941,
|
7136 |
+
"learning_rate": 3.415467409354809e-05,
|
7137 |
+
"loss": 0.7005,
|
7138 |
+
"step": 1015
|
7139 |
+
},
|
7140 |
+
{
|
7141 |
+
"epoch": 0.19923521913913128,
|
7142 |
+
"grad_norm": 4.326775074005127,
|
7143 |
+
"learning_rate": 3.398338531801457e-05,
|
7144 |
+
"loss": 1.0779,
|
7145 |
+
"step": 1016
|
7146 |
+
},
|
7147 |
+
{
|
7148 |
+
"epoch": 0.19943131679576429,
|
7149 |
+
"grad_norm": 5.717733383178711,
|
7150 |
+
"learning_rate": 3.381243917160448e-05,
|
7151 |
+
"loss": 2.7467,
|
7152 |
+
"step": 1017
|
7153 |
+
},
|
7154 |
+
{
|
7155 |
+
"epoch": 0.1996274144523973,
|
7156 |
+
"grad_norm": 12.189995765686035,
|
7157 |
+
"learning_rate": 3.364183654153592e-05,
|
7158 |
+
"loss": 1.4596,
|
7159 |
+
"step": 1018
|
7160 |
+
},
|
7161 |
+
{
|
7162 |
+
"epoch": 0.1998235121090303,
|
7163 |
+
"grad_norm": 5.9200239181518555,
|
7164 |
+
"learning_rate": 3.3471578313243903e-05,
|
7165 |
+
"loss": 1.8764,
|
7166 |
+
"step": 1019
|
7167 |
+
},
|
7168 |
+
{
|
7169 |
+
"epoch": 0.2000196097656633,
|
7170 |
+
"grad_norm": 11.550288200378418,
|
7171 |
+
"learning_rate": 3.330166537037618e-05,
|
7172 |
+
"loss": 1.3655,
|
7173 |
+
"step": 1020
|
7174 |
+
},
|
7175 |
+
{
|
7176 |
+
"epoch": 0.2002157074222963,
|
7177 |
+
"grad_norm": 4.954752445220947,
|
7178 |
+
"learning_rate": 3.313209859478839e-05,
|
7179 |
+
"loss": 1.2426,
|
7180 |
+
"step": 1021
|
7181 |
+
},
|
7182 |
+
{
|
7183 |
+
"epoch": 0.2004118050789293,
|
7184 |
+
"grad_norm": 11.455079078674316,
|
7185 |
+
"learning_rate": 3.296287886653941e-05,
|
7186 |
+
"loss": 2.2854,
|
7187 |
+
"step": 1022
|
7188 |
+
},
|
7189 |
+
{
|
7190 |
+
"epoch": 0.2006079027355623,
|
7191 |
+
"grad_norm": 6.088212966918945,
|
7192 |
+
"learning_rate": 3.2794007063887186e-05,
|
7193 |
+
"loss": 1.3675,
|
7194 |
+
"step": 1023
|
7195 |
+
},
|
7196 |
+
{
|
7197 |
+
"epoch": 0.20080400039219531,
|
7198 |
+
"grad_norm": 7.098971366882324,
|
7199 |
+
"learning_rate": 3.262548406328365e-05,
|
7200 |
+
"loss": 1.0983,
|
7201 |
+
"step": 1024
|
7202 |
+
},
|
7203 |
+
{
|
7204 |
+
"epoch": 0.20100009804882832,
|
7205 |
+
"grad_norm": 3.8312485218048096,
|
7206 |
+
"learning_rate": 3.245731073937068e-05,
|
7207 |
+
"loss": 1.7243,
|
7208 |
+
"step": 1025
|
7209 |
+
},
|
7210 |
+
{
|
7211 |
+
"epoch": 0.20119619570546132,
|
7212 |
+
"grad_norm": 3.807969570159912,
|
7213 |
+
"learning_rate": 3.2289487964975076e-05,
|
7214 |
+
"loss": 1.0291,
|
7215 |
+
"step": 1026
|
7216 |
+
},
|
7217 |
+
{
|
7218 |
+
"epoch": 0.20139229336209433,
|
7219 |
+
"grad_norm": 4.957508087158203,
|
7220 |
+
"learning_rate": 3.212201661110449e-05,
|
7221 |
+
"loss": 1.1555,
|
7222 |
+
"step": 1027
|
7223 |
+
},
|
7224 |
+
{
|
7225 |
+
"epoch": 0.20158839101872733,
|
7226 |
+
"grad_norm": 10.736882209777832,
|
7227 |
+
"learning_rate": 3.1954897546942584e-05,
|
7228 |
+
"loss": 1.9095,
|
7229 |
+
"step": 1028
|
7230 |
+
},
|
7231 |
+
{
|
7232 |
+
"epoch": 0.20178448867536034,
|
7233 |
+
"grad_norm": 9.901065826416016,
|
7234 |
+
"learning_rate": 3.1788131639844534e-05,
|
7235 |
+
"loss": 2.2914,
|
7236 |
+
"step": 1029
|
7237 |
+
},
|
7238 |
+
{
|
7239 |
+
"epoch": 0.20198058633199334,
|
7240 |
+
"grad_norm": 5.642101764678955,
|
7241 |
+
"learning_rate": 3.162171975533282e-05,
|
7242 |
+
"loss": 1.3295,
|
7243 |
+
"step": 1030
|
7244 |
+
},
|
7245 |
+
{
|
7246 |
+
"epoch": 0.20217668398862634,
|
7247 |
+
"grad_norm": 6.677104949951172,
|
7248 |
+
"learning_rate": 3.1455662757092306e-05,
|
7249 |
+
"loss": 1.3554,
|
7250 |
+
"step": 1031
|
7251 |
+
},
|
7252 |
+
{
|
7253 |
+
"epoch": 0.20237278164525935,
|
7254 |
+
"grad_norm": 5.27210807800293,
|
7255 |
+
"learning_rate": 3.1289961506966214e-05,
|
7256 |
+
"loss": 1.0698,
|
7257 |
+
"step": 1032
|
7258 |
+
},
|
7259 |
+
{
|
7260 |
+
"epoch": 0.20256887930189235,
|
7261 |
+
"grad_norm": 6.362321853637695,
|
7262 |
+
"learning_rate": 3.11246168649512e-05,
|
7263 |
+
"loss": 1.135,
|
7264 |
+
"step": 1033
|
7265 |
+
},
|
7266 |
+
{
|
7267 |
+
"epoch": 0.20276497695852536,
|
7268 |
+
"grad_norm": 4.513217449188232,
|
7269 |
+
"learning_rate": 3.095962968919319e-05,
|
7270 |
+
"loss": 1.9138,
|
7271 |
+
"step": 1034
|
7272 |
+
},
|
7273 |
+
{
|
7274 |
+
"epoch": 0.20296107461515836,
|
7275 |
+
"grad_norm": 16.53858184814453,
|
7276 |
+
"learning_rate": 3.079500083598297e-05,
|
7277 |
+
"loss": 2.4364,
|
7278 |
+
"step": 1035
|
7279 |
+
},
|
7280 |
+
{
|
7281 |
+
"epoch": 0.20315717227179136,
|
7282 |
+
"grad_norm": 5.629922389984131,
|
7283 |
+
"learning_rate": 3.063073115975136e-05,
|
7284 |
+
"loss": 1.7899,
|
7285 |
+
"step": 1036
|
7286 |
+
},
|
7287 |
+
{
|
7288 |
+
"epoch": 0.20335326992842437,
|
7289 |
+
"grad_norm": 9.979622840881348,
|
7290 |
+
"learning_rate": 3.0466821513065314e-05,
|
7291 |
+
"loss": 1.7134,
|
7292 |
+
"step": 1037
|
7293 |
+
},
|
7294 |
+
{
|
7295 |
+
"epoch": 0.20354936758505734,
|
7296 |
+
"grad_norm": 8.400696754455566,
|
7297 |
+
"learning_rate": 3.030327274662298e-05,
|
7298 |
+
"loss": 1.6557,
|
7299 |
+
"step": 1038
|
7300 |
+
},
|
7301 |
+
{
|
7302 |
+
"epoch": 0.20374546524169035,
|
7303 |
+
"grad_norm": 9.457781791687012,
|
7304 |
+
"learning_rate": 3.0140085709249667e-05,
|
7305 |
+
"loss": 2.3906,
|
7306 |
+
"step": 1039
|
7307 |
+
},
|
7308 |
+
{
|
7309 |
+
"epoch": 0.20394156289832335,
|
7310 |
+
"grad_norm": 4.818408489227295,
|
7311 |
+
"learning_rate": 2.997726124789324e-05,
|
7312 |
+
"loss": 1.0997,
|
7313 |
+
"step": 1040
|
7314 |
+
},
|
7315 |
+
{
|
7316 |
+
"epoch": 0.20413766055495636,
|
7317 |
+
"grad_norm": 7.07374382019043,
|
7318 |
+
"learning_rate": 2.9814800207619774e-05,
|
7319 |
+
"loss": 1.441,
|
7320 |
+
"step": 1041
|
7321 |
+
},
|
7322 |
+
{
|
7323 |
+
"epoch": 0.20433375821158936,
|
7324 |
+
"grad_norm": 9.976203918457031,
|
7325 |
+
"learning_rate": 2.9652703431609263e-05,
|
7326 |
+
"loss": 2.2031,
|
7327 |
+
"step": 1042
|
7328 |
+
},
|
7329 |
+
{
|
7330 |
+
"epoch": 0.20452985586822237,
|
7331 |
+
"grad_norm": 8.051301002502441,
|
7332 |
+
"learning_rate": 2.9490971761151e-05,
|
7333 |
+
"loss": 1.2837,
|
7334 |
+
"step": 1043
|
7335 |
+
},
|
7336 |
+
{
|
7337 |
+
"epoch": 0.20472595352485537,
|
7338 |
+
"grad_norm": 3.8618202209472656,
|
7339 |
+
"learning_rate": 2.9329606035639458e-05,
|
7340 |
+
"loss": 1.5356,
|
7341 |
+
"step": 1044
|
7342 |
+
},
|
7343 |
+
{
|
7344 |
+
"epoch": 0.20472595352485537,
|
7345 |
+
"eval_loss": 0.3870697021484375,
|
7346 |
+
"eval_runtime": 78.8825,
|
7347 |
+
"eval_samples_per_second": 27.23,
|
7348 |
+
"eval_steps_per_second": 13.615,
|
7349 |
+
"step": 1044
|
7350 |
}
|
7351 |
],
|
7352 |
"logging_steps": 1,
|
|
|
7366 |
"attributes": {}
|
7367 |
}
|
7368 |
},
|
7369 |
+
"total_flos": 2.242105399144612e+17,
|
7370 |
"train_batch_size": 2,
|
7371 |
"trial_name": null,
|
7372 |
"trial_params": null
|