Training in progress, step 1457, 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 5752
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2cb705bad3827adf6c1423dafc1f8940ea34f89a84b64a1e65074910132bcf21
|
3 |
size 5752
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 15814
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3995ef4ab870a0e3cdf582cac2f5facad5a36676c21cba770efdb6b986267d0b
|
3 |
size 15814
|
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:b3a48f87c9cac922053b78fd0ccd147e85517cc792eccbd42f59dd8dca3ffb45
|
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:fdec083de80dd151eeee42c99f49e367aef5a0e8ae79a65712b53ce9f3959ae6
|
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": 365,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -7704,6 +7704,2540 @@
|
|
7704 |
"eval_samples_per_second": 175.707,
|
7705 |
"eval_steps_per_second": 87.888,
|
7706 |
"step": 1095
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7707 |
}
|
7708 |
],
|
7709 |
"logging_steps": 1,
|
@@ -7718,12 +10252,12 @@
|
|
7718 |
"should_evaluate": false,
|
7719 |
"should_log": false,
|
7720 |
"should_save": true,
|
7721 |
-
"should_training_stop":
|
7722 |
},
|
7723 |
"attributes": {}
|
7724 |
}
|
7725 |
},
|
7726 |
-
"total_flos":
|
7727 |
"train_batch_size": 2,
|
7728 |
"trial_name": null,
|
7729 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.2458760494452179,
|
5 |
"eval_steps": 365,
|
6 |
+
"global_step": 1457,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
7704 |
"eval_samples_per_second": 175.707,
|
7705 |
"eval_steps_per_second": 87.888,
|
7706 |
"step": 1095
|
7707 |
+
},
|
7708 |
+
{
|
7709 |
+
"epoch": 0.18495549086613508,
|
7710 |
+
"grad_norm": 0.0018527479842305183,
|
7711 |
+
"learning_rate": 2.917427530910154e-05,
|
7712 |
+
"loss": 46.0,
|
7713 |
+
"step": 1096
|
7714 |
+
},
|
7715 |
+
{
|
7716 |
+
"epoch": 0.18512424587604945,
|
7717 |
+
"grad_norm": 0.000325352099025622,
|
7718 |
+
"learning_rate": 2.9021172052347113e-05,
|
7719 |
+
"loss": 46.0,
|
7720 |
+
"step": 1097
|
7721 |
+
},
|
7722 |
+
{
|
7723 |
+
"epoch": 0.18529300088596382,
|
7724 |
+
"grad_norm": 0.0008033345802687109,
|
7725 |
+
"learning_rate": 2.886840336893606e-05,
|
7726 |
+
"loss": 46.0,
|
7727 |
+
"step": 1098
|
7728 |
+
},
|
7729 |
+
{
|
7730 |
+
"epoch": 0.18546175589587816,
|
7731 |
+
"grad_norm": 0.003062557429075241,
|
7732 |
+
"learning_rate": 2.871596997897509e-05,
|
7733 |
+
"loss": 46.0,
|
7734 |
+
"step": 1099
|
7735 |
+
},
|
7736 |
+
{
|
7737 |
+
"epoch": 0.18563051090579252,
|
7738 |
+
"grad_norm": 0.0005990050849504769,
|
7739 |
+
"learning_rate": 2.8563872600990394e-05,
|
7740 |
+
"loss": 46.0,
|
7741 |
+
"step": 1100
|
7742 |
+
},
|
7743 |
+
{
|
7744 |
+
"epoch": 0.18579926591570686,
|
7745 |
+
"grad_norm": 0.00030623353086411953,
|
7746 |
+
"learning_rate": 2.841211195192436e-05,
|
7747 |
+
"loss": 46.0,
|
7748 |
+
"step": 1101
|
7749 |
+
},
|
7750 |
+
{
|
7751 |
+
"epoch": 0.18596802092562123,
|
7752 |
+
"grad_norm": 0.0005539836129173636,
|
7753 |
+
"learning_rate": 2.826068874713208e-05,
|
7754 |
+
"loss": 46.0,
|
7755 |
+
"step": 1102
|
7756 |
+
},
|
7757 |
+
{
|
7758 |
+
"epoch": 0.1861367759355356,
|
7759 |
+
"grad_norm": 0.0013511950382962823,
|
7760 |
+
"learning_rate": 2.8109603700378074e-05,
|
7761 |
+
"loss": 46.0,
|
7762 |
+
"step": 1103
|
7763 |
+
},
|
7764 |
+
{
|
7765 |
+
"epoch": 0.18630553094544994,
|
7766 |
+
"grad_norm": 0.0009734542109072208,
|
7767 |
+
"learning_rate": 2.795885752383284e-05,
|
7768 |
+
"loss": 46.0,
|
7769 |
+
"step": 1104
|
7770 |
+
},
|
7771 |
+
{
|
7772 |
+
"epoch": 0.1864742859553643,
|
7773 |
+
"grad_norm": 0.000757011875975877,
|
7774 |
+
"learning_rate": 2.780845092806964e-05,
|
7775 |
+
"loss": 46.0,
|
7776 |
+
"step": 1105
|
7777 |
+
},
|
7778 |
+
{
|
7779 |
+
"epoch": 0.18664304096527864,
|
7780 |
+
"grad_norm": 0.0009687045239843428,
|
7781 |
+
"learning_rate": 2.7658384622060873e-05,
|
7782 |
+
"loss": 46.0,
|
7783 |
+
"step": 1106
|
7784 |
+
},
|
7785 |
+
{
|
7786 |
+
"epoch": 0.186811795975193,
|
7787 |
+
"grad_norm": 0.0008338790503330529,
|
7788 |
+
"learning_rate": 2.7508659313175112e-05,
|
7789 |
+
"loss": 46.0,
|
7790 |
+
"step": 1107
|
7791 |
+
},
|
7792 |
+
{
|
7793 |
+
"epoch": 0.18698055098510738,
|
7794 |
+
"grad_norm": 0.000879534287378192,
|
7795 |
+
"learning_rate": 2.735927570717348e-05,
|
7796 |
+
"loss": 46.0,
|
7797 |
+
"step": 1108
|
7798 |
+
},
|
7799 |
+
{
|
7800 |
+
"epoch": 0.18714930599502172,
|
7801 |
+
"grad_norm": 0.000895138131454587,
|
7802 |
+
"learning_rate": 2.7210234508206313e-05,
|
7803 |
+
"loss": 46.0,
|
7804 |
+
"step": 1109
|
7805 |
+
},
|
7806 |
+
{
|
7807 |
+
"epoch": 0.1873180610049361,
|
7808 |
+
"grad_norm": 0.0016813467955216765,
|
7809 |
+
"learning_rate": 2.706153641881013e-05,
|
7810 |
+
"loss": 46.0,
|
7811 |
+
"step": 1110
|
7812 |
+
},
|
7813 |
+
{
|
7814 |
+
"epoch": 0.18748681601485043,
|
7815 |
+
"grad_norm": 0.00046861334703862667,
|
7816 |
+
"learning_rate": 2.6913182139904014e-05,
|
7817 |
+
"loss": 46.0,
|
7818 |
+
"step": 1111
|
7819 |
+
},
|
7820 |
+
{
|
7821 |
+
"epoch": 0.1876555710247648,
|
7822 |
+
"grad_norm": 0.0008181455777958035,
|
7823 |
+
"learning_rate": 2.6765172370786463e-05,
|
7824 |
+
"loss": 46.0,
|
7825 |
+
"step": 1112
|
7826 |
+
},
|
7827 |
+
{
|
7828 |
+
"epoch": 0.18782432603467916,
|
7829 |
+
"grad_norm": 0.000511069200001657,
|
7830 |
+
"learning_rate": 2.6617507809132038e-05,
|
7831 |
+
"loss": 46.0,
|
7832 |
+
"step": 1113
|
7833 |
+
},
|
7834 |
+
{
|
7835 |
+
"epoch": 0.1879930810445935,
|
7836 |
+
"grad_norm": 0.0009120728354901075,
|
7837 |
+
"learning_rate": 2.647018915098809e-05,
|
7838 |
+
"loss": 46.0,
|
7839 |
+
"step": 1114
|
7840 |
+
},
|
7841 |
+
{
|
7842 |
+
"epoch": 0.18816183605450787,
|
7843 |
+
"grad_norm": 0.0011518978280946612,
|
7844 |
+
"learning_rate": 2.6323217090771512e-05,
|
7845 |
+
"loss": 46.0,
|
7846 |
+
"step": 1115
|
7847 |
+
},
|
7848 |
+
{
|
7849 |
+
"epoch": 0.18833059106442224,
|
7850 |
+
"grad_norm": 0.001370842568576336,
|
7851 |
+
"learning_rate": 2.6176592321265425e-05,
|
7852 |
+
"loss": 46.0,
|
7853 |
+
"step": 1116
|
7854 |
+
},
|
7855 |
+
{
|
7856 |
+
"epoch": 0.18849934607433658,
|
7857 |
+
"grad_norm": 0.002501879585906863,
|
7858 |
+
"learning_rate": 2.6030315533615878e-05,
|
7859 |
+
"loss": 46.0,
|
7860 |
+
"step": 1117
|
7861 |
+
},
|
7862 |
+
{
|
7863 |
+
"epoch": 0.18866810108425094,
|
7864 |
+
"grad_norm": 0.0005397344939410686,
|
7865 |
+
"learning_rate": 2.588438741732876e-05,
|
7866 |
+
"loss": 46.0,
|
7867 |
+
"step": 1118
|
7868 |
+
},
|
7869 |
+
{
|
7870 |
+
"epoch": 0.18883685609416528,
|
7871 |
+
"grad_norm": 0.0013044921215623617,
|
7872 |
+
"learning_rate": 2.5738808660266235e-05,
|
7873 |
+
"loss": 46.0,
|
7874 |
+
"step": 1119
|
7875 |
+
},
|
7876 |
+
{
|
7877 |
+
"epoch": 0.18900561110407965,
|
7878 |
+
"grad_norm": 0.0007338232826441526,
|
7879 |
+
"learning_rate": 2.5593579948643874e-05,
|
7880 |
+
"loss": 46.0,
|
7881 |
+
"step": 1120
|
7882 |
+
},
|
7883 |
+
{
|
7884 |
+
"epoch": 0.18917436611399402,
|
7885 |
+
"grad_norm": 0.0011662261094897985,
|
7886 |
+
"learning_rate": 2.5448701967027167e-05,
|
7887 |
+
"loss": 46.0,
|
7888 |
+
"step": 1121
|
7889 |
+
},
|
7890 |
+
{
|
7891 |
+
"epoch": 0.18934312112390836,
|
7892 |
+
"grad_norm": 0.00048239415627904236,
|
7893 |
+
"learning_rate": 2.5304175398328278e-05,
|
7894 |
+
"loss": 46.0,
|
7895 |
+
"step": 1122
|
7896 |
+
},
|
7897 |
+
{
|
7898 |
+
"epoch": 0.18951187613382273,
|
7899 |
+
"grad_norm": 0.002095632255077362,
|
7900 |
+
"learning_rate": 2.5160000923803063e-05,
|
7901 |
+
"loss": 46.0,
|
7902 |
+
"step": 1123
|
7903 |
+
},
|
7904 |
+
{
|
7905 |
+
"epoch": 0.18968063114373707,
|
7906 |
+
"grad_norm": 0.00038515665801241994,
|
7907 |
+
"learning_rate": 2.501617922304763e-05,
|
7908 |
+
"loss": 46.0,
|
7909 |
+
"step": 1124
|
7910 |
+
},
|
7911 |
+
{
|
7912 |
+
"epoch": 0.18984938615365143,
|
7913 |
+
"grad_norm": 0.0008104884182102978,
|
7914 |
+
"learning_rate": 2.4872710973995204e-05,
|
7915 |
+
"loss": 46.0,
|
7916 |
+
"step": 1125
|
7917 |
+
},
|
7918 |
+
{
|
7919 |
+
"epoch": 0.1900181411635658,
|
7920 |
+
"grad_norm": 0.0006576053565368056,
|
7921 |
+
"learning_rate": 2.4729596852912996e-05,
|
7922 |
+
"loss": 46.0,
|
7923 |
+
"step": 1126
|
7924 |
+
},
|
7925 |
+
{
|
7926 |
+
"epoch": 0.19018689617348014,
|
7927 |
+
"grad_norm": 0.0006934543489478528,
|
7928 |
+
"learning_rate": 2.4586837534398898e-05,
|
7929 |
+
"loss": 46.0,
|
7930 |
+
"step": 1127
|
7931 |
+
},
|
7932 |
+
{
|
7933 |
+
"epoch": 0.1903556511833945,
|
7934 |
+
"grad_norm": 0.0016504209488630295,
|
7935 |
+
"learning_rate": 2.444443369137841e-05,
|
7936 |
+
"loss": 46.0,
|
7937 |
+
"step": 1128
|
7938 |
+
},
|
7939 |
+
{
|
7940 |
+
"epoch": 0.19052440619330888,
|
7941 |
+
"grad_norm": 0.0016103170346468687,
|
7942 |
+
"learning_rate": 2.4302385995101417e-05,
|
7943 |
+
"loss": 46.0,
|
7944 |
+
"step": 1129
|
7945 |
+
},
|
7946 |
+
{
|
7947 |
+
"epoch": 0.19069316120322322,
|
7948 |
+
"grad_norm": 0.0009479466243647039,
|
7949 |
+
"learning_rate": 2.4160695115138998e-05,
|
7950 |
+
"loss": 46.0,
|
7951 |
+
"step": 1130
|
7952 |
+
},
|
7953 |
+
{
|
7954 |
+
"epoch": 0.19086191621313758,
|
7955 |
+
"grad_norm": 0.0006660729413852096,
|
7956 |
+
"learning_rate": 2.401936171938043e-05,
|
7957 |
+
"loss": 46.0,
|
7958 |
+
"step": 1131
|
7959 |
+
},
|
7960 |
+
{
|
7961 |
+
"epoch": 0.19103067122305192,
|
7962 |
+
"grad_norm": 0.002802348928526044,
|
7963 |
+
"learning_rate": 2.3878386474029724e-05,
|
7964 |
+
"loss": 46.0,
|
7965 |
+
"step": 1132
|
7966 |
+
},
|
7967 |
+
{
|
7968 |
+
"epoch": 0.1911994262329663,
|
7969 |
+
"grad_norm": 0.0021046160254627466,
|
7970 |
+
"learning_rate": 2.373777004360278e-05,
|
7971 |
+
"loss": 46.0,
|
7972 |
+
"step": 1133
|
7973 |
+
},
|
7974 |
+
{
|
7975 |
+
"epoch": 0.19136818124288066,
|
7976 |
+
"grad_norm": 0.0005733026191592216,
|
7977 |
+
"learning_rate": 2.3597513090924228e-05,
|
7978 |
+
"loss": 46.0,
|
7979 |
+
"step": 1134
|
7980 |
+
},
|
7981 |
+
{
|
7982 |
+
"epoch": 0.191536936252795,
|
7983 |
+
"grad_norm": 0.001300279633142054,
|
7984 |
+
"learning_rate": 2.3457616277124018e-05,
|
7985 |
+
"loss": 46.0,
|
7986 |
+
"step": 1135
|
7987 |
+
},
|
7988 |
+
{
|
7989 |
+
"epoch": 0.19170569126270937,
|
7990 |
+
"grad_norm": 0.0020821443758904934,
|
7991 |
+
"learning_rate": 2.3318080261634723e-05,
|
7992 |
+
"loss": 46.0,
|
7993 |
+
"step": 1136
|
7994 |
+
},
|
7995 |
+
{
|
7996 |
+
"epoch": 0.1918744462726237,
|
7997 |
+
"grad_norm": 0.0010038736509159207,
|
7998 |
+
"learning_rate": 2.3178905702188124e-05,
|
7999 |
+
"loss": 46.0,
|
8000 |
+
"step": 1137
|
8001 |
+
},
|
8002 |
+
{
|
8003 |
+
"epoch": 0.19204320128253807,
|
8004 |
+
"grad_norm": 0.0026241031009703875,
|
8005 |
+
"learning_rate": 2.3040093254812123e-05,
|
8006 |
+
"loss": 46.0,
|
8007 |
+
"step": 1138
|
8008 |
+
},
|
8009 |
+
{
|
8010 |
+
"epoch": 0.19221195629245244,
|
8011 |
+
"grad_norm": 0.0008017763611860573,
|
8012 |
+
"learning_rate": 2.2901643573827914e-05,
|
8013 |
+
"loss": 46.0,
|
8014 |
+
"step": 1139
|
8015 |
+
},
|
8016 |
+
{
|
8017 |
+
"epoch": 0.19238071130236678,
|
8018 |
+
"grad_norm": 0.0006083068437874317,
|
8019 |
+
"learning_rate": 2.2763557311846583e-05,
|
8020 |
+
"loss": 46.0,
|
8021 |
+
"step": 1140
|
8022 |
+
},
|
8023 |
+
{
|
8024 |
+
"epoch": 0.19254946631228115,
|
8025 |
+
"grad_norm": 0.0014171472284942865,
|
8026 |
+
"learning_rate": 2.2625835119766202e-05,
|
8027 |
+
"loss": 46.0,
|
8028 |
+
"step": 1141
|
8029 |
+
},
|
8030 |
+
{
|
8031 |
+
"epoch": 0.19271822132219552,
|
8032 |
+
"grad_norm": 0.00046038886648602784,
|
8033 |
+
"learning_rate": 2.2488477646768724e-05,
|
8034 |
+
"loss": 46.0,
|
8035 |
+
"step": 1142
|
8036 |
+
},
|
8037 |
+
{
|
8038 |
+
"epoch": 0.19288697633210986,
|
8039 |
+
"grad_norm": 0.0006552223348990083,
|
8040 |
+
"learning_rate": 2.235148554031694e-05,
|
8041 |
+
"loss": 46.0,
|
8042 |
+
"step": 1143
|
8043 |
+
},
|
8044 |
+
{
|
8045 |
+
"epoch": 0.19305573134202422,
|
8046 |
+
"grad_norm": 0.0009186275419779122,
|
8047 |
+
"learning_rate": 2.221485944615137e-05,
|
8048 |
+
"loss": 46.0,
|
8049 |
+
"step": 1144
|
8050 |
+
},
|
8051 |
+
{
|
8052 |
+
"epoch": 0.19322448635193856,
|
8053 |
+
"grad_norm": 0.0005543860606849194,
|
8054 |
+
"learning_rate": 2.207860000828731e-05,
|
8055 |
+
"loss": 46.0,
|
8056 |
+
"step": 1145
|
8057 |
+
},
|
8058 |
+
{
|
8059 |
+
"epoch": 0.19339324136185293,
|
8060 |
+
"grad_norm": 0.0007307975320145488,
|
8061 |
+
"learning_rate": 2.1942707869011682e-05,
|
8062 |
+
"loss": 46.0,
|
8063 |
+
"step": 1146
|
8064 |
+
},
|
8065 |
+
{
|
8066 |
+
"epoch": 0.1935619963717673,
|
8067 |
+
"grad_norm": 0.0009457955020479858,
|
8068 |
+
"learning_rate": 2.1807183668880194e-05,
|
8069 |
+
"loss": 46.0,
|
8070 |
+
"step": 1147
|
8071 |
+
},
|
8072 |
+
{
|
8073 |
+
"epoch": 0.19373075138168164,
|
8074 |
+
"grad_norm": 0.0009510859963484108,
|
8075 |
+
"learning_rate": 2.1672028046713998e-05,
|
8076 |
+
"loss": 46.0,
|
8077 |
+
"step": 1148
|
8078 |
+
},
|
8079 |
+
{
|
8080 |
+
"epoch": 0.193899506391596,
|
8081 |
+
"grad_norm": 0.0012068504001945257,
|
8082 |
+
"learning_rate": 2.1537241639597082e-05,
|
8083 |
+
"loss": 46.0,
|
8084 |
+
"step": 1149
|
8085 |
+
},
|
8086 |
+
{
|
8087 |
+
"epoch": 0.19406826140151034,
|
8088 |
+
"grad_norm": 0.0008088087779469788,
|
8089 |
+
"learning_rate": 2.1402825082872968e-05,
|
8090 |
+
"loss": 46.0,
|
8091 |
+
"step": 1150
|
8092 |
+
},
|
8093 |
+
{
|
8094 |
+
"epoch": 0.1942370164114247,
|
8095 |
+
"grad_norm": 0.00042066810419782996,
|
8096 |
+
"learning_rate": 2.1268779010141737e-05,
|
8097 |
+
"loss": 46.0,
|
8098 |
+
"step": 1151
|
8099 |
+
},
|
8100 |
+
{
|
8101 |
+
"epoch": 0.19440577142133908,
|
8102 |
+
"grad_norm": 0.0008168341009877622,
|
8103 |
+
"learning_rate": 2.1135104053257272e-05,
|
8104 |
+
"loss": 46.0,
|
8105 |
+
"step": 1152
|
8106 |
+
},
|
8107 |
+
{
|
8108 |
+
"epoch": 0.19457452643125342,
|
8109 |
+
"grad_norm": 0.0011002906830981374,
|
8110 |
+
"learning_rate": 2.1001800842324026e-05,
|
8111 |
+
"loss": 46.0,
|
8112 |
+
"step": 1153
|
8113 |
+
},
|
8114 |
+
{
|
8115 |
+
"epoch": 0.1947432814411678,
|
8116 |
+
"grad_norm": 0.0006414004019461572,
|
8117 |
+
"learning_rate": 2.0868870005694173e-05,
|
8118 |
+
"loss": 46.0,
|
8119 |
+
"step": 1154
|
8120 |
+
},
|
8121 |
+
{
|
8122 |
+
"epoch": 0.19491203645108215,
|
8123 |
+
"grad_norm": 0.0005531748756766319,
|
8124 |
+
"learning_rate": 2.0736312169964635e-05,
|
8125 |
+
"loss": 46.0,
|
8126 |
+
"step": 1155
|
8127 |
+
},
|
8128 |
+
{
|
8129 |
+
"epoch": 0.1950807914609965,
|
8130 |
+
"grad_norm": 0.000641561346128583,
|
8131 |
+
"learning_rate": 2.0604127959974107e-05,
|
8132 |
+
"loss": 46.0,
|
8133 |
+
"step": 1156
|
8134 |
+
},
|
8135 |
+
{
|
8136 |
+
"epoch": 0.19524954647091086,
|
8137 |
+
"grad_norm": 0.0006983129424043,
|
8138 |
+
"learning_rate": 2.0472317998800106e-05,
|
8139 |
+
"loss": 46.0,
|
8140 |
+
"step": 1157
|
8141 |
+
},
|
8142 |
+
{
|
8143 |
+
"epoch": 0.1954183014808252,
|
8144 |
+
"grad_norm": 0.0006936094141565263,
|
8145 |
+
"learning_rate": 2.0340882907756076e-05,
|
8146 |
+
"loss": 46.0,
|
8147 |
+
"step": 1158
|
8148 |
+
},
|
8149 |
+
{
|
8150 |
+
"epoch": 0.19558705649073957,
|
8151 |
+
"grad_norm": 0.0006084066699258983,
|
8152 |
+
"learning_rate": 2.020982330638841e-05,
|
8153 |
+
"loss": 46.0,
|
8154 |
+
"step": 1159
|
8155 |
+
},
|
8156 |
+
{
|
8157 |
+
"epoch": 0.19575581150065394,
|
8158 |
+
"grad_norm": 0.004117981996387243,
|
8159 |
+
"learning_rate": 2.0079139812473636e-05,
|
8160 |
+
"loss": 46.0,
|
8161 |
+
"step": 1160
|
8162 |
+
},
|
8163 |
+
{
|
8164 |
+
"epoch": 0.19592456651056828,
|
8165 |
+
"grad_norm": 0.0016725576715543866,
|
8166 |
+
"learning_rate": 1.994883304201527e-05,
|
8167 |
+
"loss": 46.0,
|
8168 |
+
"step": 1161
|
8169 |
+
},
|
8170 |
+
{
|
8171 |
+
"epoch": 0.19609332152048264,
|
8172 |
+
"grad_norm": 0.000729929655790329,
|
8173 |
+
"learning_rate": 1.981890360924116e-05,
|
8174 |
+
"loss": 46.0,
|
8175 |
+
"step": 1162
|
8176 |
+
},
|
8177 |
+
{
|
8178 |
+
"epoch": 0.19626207653039698,
|
8179 |
+
"grad_norm": 0.0005720785120502114,
|
8180 |
+
"learning_rate": 1.9689352126600536e-05,
|
8181 |
+
"loss": 46.0,
|
8182 |
+
"step": 1163
|
8183 |
+
},
|
8184 |
+
{
|
8185 |
+
"epoch": 0.19643083154031135,
|
8186 |
+
"grad_norm": 0.0008306180825456977,
|
8187 |
+
"learning_rate": 1.9560179204760954e-05,
|
8188 |
+
"loss": 46.0,
|
8189 |
+
"step": 1164
|
8190 |
+
},
|
8191 |
+
{
|
8192 |
+
"epoch": 0.19659958655022572,
|
8193 |
+
"grad_norm": 0.0006977242301218212,
|
8194 |
+
"learning_rate": 1.943138545260569e-05,
|
8195 |
+
"loss": 46.0,
|
8196 |
+
"step": 1165
|
8197 |
+
},
|
8198 |
+
{
|
8199 |
+
"epoch": 0.19676834156014006,
|
8200 |
+
"grad_norm": 0.0005357645568437874,
|
8201 |
+
"learning_rate": 1.9302971477230613e-05,
|
8202 |
+
"loss": 46.0,
|
8203 |
+
"step": 1166
|
8204 |
+
},
|
8205 |
+
{
|
8206 |
+
"epoch": 0.19693709657005443,
|
8207 |
+
"grad_norm": 0.0009179019252769649,
|
8208 |
+
"learning_rate": 1.9174937883941503e-05,
|
8209 |
+
"loss": 46.0,
|
8210 |
+
"step": 1167
|
8211 |
+
},
|
8212 |
+
{
|
8213 |
+
"epoch": 0.1971058515799688,
|
8214 |
+
"grad_norm": 0.000713715679012239,
|
8215 |
+
"learning_rate": 1.904728527625108e-05,
|
8216 |
+
"loss": 46.0,
|
8217 |
+
"step": 1168
|
8218 |
+
},
|
8219 |
+
{
|
8220 |
+
"epoch": 0.19727460658988313,
|
8221 |
+
"grad_norm": 0.0006716207135468721,
|
8222 |
+
"learning_rate": 1.8920014255876263e-05,
|
8223 |
+
"loss": 46.0,
|
8224 |
+
"step": 1169
|
8225 |
+
},
|
8226 |
+
{
|
8227 |
+
"epoch": 0.1974433615997975,
|
8228 |
+
"grad_norm": 0.0018655718304216862,
|
8229 |
+
"learning_rate": 1.8793125422735235e-05,
|
8230 |
+
"loss": 46.0,
|
8231 |
+
"step": 1170
|
8232 |
+
},
|
8233 |
+
{
|
8234 |
+
"epoch": 0.19761211660971184,
|
8235 |
+
"grad_norm": 0.0017995084635913372,
|
8236 |
+
"learning_rate": 1.8666619374944684e-05,
|
8237 |
+
"loss": 46.0,
|
8238 |
+
"step": 1171
|
8239 |
+
},
|
8240 |
+
{
|
8241 |
+
"epoch": 0.1977808716196262,
|
8242 |
+
"grad_norm": 0.001270498731173575,
|
8243 |
+
"learning_rate": 1.8540496708816935e-05,
|
8244 |
+
"loss": 46.0,
|
8245 |
+
"step": 1172
|
8246 |
+
},
|
8247 |
+
{
|
8248 |
+
"epoch": 0.19794962662954058,
|
8249 |
+
"grad_norm": 0.0005029829917475581,
|
8250 |
+
"learning_rate": 1.8414758018857248e-05,
|
8251 |
+
"loss": 46.0,
|
8252 |
+
"step": 1173
|
8253 |
+
},
|
8254 |
+
{
|
8255 |
+
"epoch": 0.19811838163945492,
|
8256 |
+
"grad_norm": 0.0007416327716782689,
|
8257 |
+
"learning_rate": 1.8289403897760815e-05,
|
8258 |
+
"loss": 46.0,
|
8259 |
+
"step": 1174
|
8260 |
+
},
|
8261 |
+
{
|
8262 |
+
"epoch": 0.19828713664936928,
|
8263 |
+
"grad_norm": 0.0004362264589872211,
|
8264 |
+
"learning_rate": 1.8164434936410114e-05,
|
8265 |
+
"loss": 46.0,
|
8266 |
+
"step": 1175
|
8267 |
+
},
|
8268 |
+
{
|
8269 |
+
"epoch": 0.19845589165928362,
|
8270 |
+
"grad_norm": 0.0005498372483998537,
|
8271 |
+
"learning_rate": 1.8039851723872204e-05,
|
8272 |
+
"loss": 46.0,
|
8273 |
+
"step": 1176
|
8274 |
+
},
|
8275 |
+
{
|
8276 |
+
"epoch": 0.198624646669198,
|
8277 |
+
"grad_norm": 0.0005490055773407221,
|
8278 |
+
"learning_rate": 1.7915654847395646e-05,
|
8279 |
+
"loss": 46.0,
|
8280 |
+
"step": 1177
|
8281 |
+
},
|
8282 |
+
{
|
8283 |
+
"epoch": 0.19879340167911236,
|
8284 |
+
"grad_norm": 0.0007399431779049337,
|
8285 |
+
"learning_rate": 1.7791844892408104e-05,
|
8286 |
+
"loss": 46.0,
|
8287 |
+
"step": 1178
|
8288 |
+
},
|
8289 |
+
{
|
8290 |
+
"epoch": 0.1989621566890267,
|
8291 |
+
"grad_norm": 0.00065398751758039,
|
8292 |
+
"learning_rate": 1.766842244251332e-05,
|
8293 |
+
"loss": 46.0,
|
8294 |
+
"step": 1179
|
8295 |
+
},
|
8296 |
+
{
|
8297 |
+
"epoch": 0.19913091169894107,
|
8298 |
+
"grad_norm": 0.0003902087628375739,
|
8299 |
+
"learning_rate": 1.7545388079488455e-05,
|
8300 |
+
"loss": 46.0,
|
8301 |
+
"step": 1180
|
8302 |
+
},
|
8303 |
+
{
|
8304 |
+
"epoch": 0.19929966670885543,
|
8305 |
+
"grad_norm": 0.0006421052967198193,
|
8306 |
+
"learning_rate": 1.7422742383281355e-05,
|
8307 |
+
"loss": 46.0,
|
8308 |
+
"step": 1181
|
8309 |
+
},
|
8310 |
+
{
|
8311 |
+
"epoch": 0.19946842171876977,
|
8312 |
+
"grad_norm": 0.000821156136225909,
|
8313 |
+
"learning_rate": 1.7300485932007794e-05,
|
8314 |
+
"loss": 46.0,
|
8315 |
+
"step": 1182
|
8316 |
+
},
|
8317 |
+
{
|
8318 |
+
"epoch": 0.19963717672868414,
|
8319 |
+
"grad_norm": 0.000910441973246634,
|
8320 |
+
"learning_rate": 1.7178619301948774e-05,
|
8321 |
+
"loss": 46.0,
|
8322 |
+
"step": 1183
|
8323 |
+
},
|
8324 |
+
{
|
8325 |
+
"epoch": 0.19980593173859848,
|
8326 |
+
"grad_norm": 0.0008174055255949497,
|
8327 |
+
"learning_rate": 1.7057143067547788e-05,
|
8328 |
+
"loss": 46.0,
|
8329 |
+
"step": 1184
|
8330 |
+
},
|
8331 |
+
{
|
8332 |
+
"epoch": 0.19997468674851285,
|
8333 |
+
"grad_norm": 0.0007186917937360704,
|
8334 |
+
"learning_rate": 1.69360578014081e-05,
|
8335 |
+
"loss": 46.0,
|
8336 |
+
"step": 1185
|
8337 |
+
},
|
8338 |
+
{
|
8339 |
+
"epoch": 0.20014344175842722,
|
8340 |
+
"grad_norm": 0.000985615304671228,
|
8341 |
+
"learning_rate": 1.6815364074290153e-05,
|
8342 |
+
"loss": 46.0,
|
8343 |
+
"step": 1186
|
8344 |
+
},
|
8345 |
+
{
|
8346 |
+
"epoch": 0.20031219676834155,
|
8347 |
+
"grad_norm": 0.0004897791077382863,
|
8348 |
+
"learning_rate": 1.6695062455108646e-05,
|
8349 |
+
"loss": 46.0,
|
8350 |
+
"step": 1187
|
8351 |
+
},
|
8352 |
+
{
|
8353 |
+
"epoch": 0.20048095177825592,
|
8354 |
+
"grad_norm": 0.0006414767703972757,
|
8355 |
+
"learning_rate": 1.6575153510930065e-05,
|
8356 |
+
"loss": 46.0,
|
8357 |
+
"step": 1188
|
8358 |
+
},
|
8359 |
+
{
|
8360 |
+
"epoch": 0.20064970678817026,
|
8361 |
+
"grad_norm": 0.00031621320522390306,
|
8362 |
+
"learning_rate": 1.6455637806970027e-05,
|
8363 |
+
"loss": 46.0,
|
8364 |
+
"step": 1189
|
8365 |
+
},
|
8366 |
+
{
|
8367 |
+
"epoch": 0.20081846179808463,
|
8368 |
+
"grad_norm": 0.004970474634319544,
|
8369 |
+
"learning_rate": 1.6336515906590354e-05,
|
8370 |
+
"loss": 46.0,
|
8371 |
+
"step": 1190
|
8372 |
+
},
|
8373 |
+
{
|
8374 |
+
"epoch": 0.200987216807999,
|
8375 |
+
"grad_norm": 0.0018685179529711604,
|
8376 |
+
"learning_rate": 1.621778837129676e-05,
|
8377 |
+
"loss": 46.0,
|
8378 |
+
"step": 1191
|
8379 |
+
},
|
8380 |
+
{
|
8381 |
+
"epoch": 0.20115597181791334,
|
8382 |
+
"grad_norm": 0.0012024708557873964,
|
8383 |
+
"learning_rate": 1.6099455760735937e-05,
|
8384 |
+
"loss": 46.0,
|
8385 |
+
"step": 1192
|
8386 |
+
},
|
8387 |
+
{
|
8388 |
+
"epoch": 0.2013247268278277,
|
8389 |
+
"grad_norm": 0.0004908541450276971,
|
8390 |
+
"learning_rate": 1.598151863269306e-05,
|
8391 |
+
"loss": 46.0,
|
8392 |
+
"step": 1193
|
8393 |
+
},
|
8394 |
+
{
|
8395 |
+
"epoch": 0.20149348183774204,
|
8396 |
+
"grad_norm": 0.0007760179578326643,
|
8397 |
+
"learning_rate": 1.5863977543089092e-05,
|
8398 |
+
"loss": 46.0,
|
8399 |
+
"step": 1194
|
8400 |
+
},
|
8401 |
+
{
|
8402 |
+
"epoch": 0.2016622368476564,
|
8403 |
+
"grad_norm": 0.00047840975457802415,
|
8404 |
+
"learning_rate": 1.5746833045978216e-05,
|
8405 |
+
"loss": 46.0,
|
8406 |
+
"step": 1195
|
8407 |
+
},
|
8408 |
+
{
|
8409 |
+
"epoch": 0.20183099185757078,
|
8410 |
+
"grad_norm": 0.0011836383491754532,
|
8411 |
+
"learning_rate": 1.563008569354516e-05,
|
8412 |
+
"loss": 46.0,
|
8413 |
+
"step": 1196
|
8414 |
+
},
|
8415 |
+
{
|
8416 |
+
"epoch": 0.20199974686748512,
|
8417 |
+
"grad_norm": 0.000663977290969342,
|
8418 |
+
"learning_rate": 1.5513736036102644e-05,
|
8419 |
+
"loss": 46.0,
|
8420 |
+
"step": 1197
|
8421 |
+
},
|
8422 |
+
{
|
8423 |
+
"epoch": 0.2021685018773995,
|
8424 |
+
"grad_norm": 0.0010960629442706704,
|
8425 |
+
"learning_rate": 1.5397784622088772e-05,
|
8426 |
+
"loss": 46.0,
|
8427 |
+
"step": 1198
|
8428 |
+
},
|
8429 |
+
{
|
8430 |
+
"epoch": 0.20233725688731385,
|
8431 |
+
"grad_norm": 0.000916484568733722,
|
8432 |
+
"learning_rate": 1.52822319980645e-05,
|
8433 |
+
"loss": 46.0,
|
8434 |
+
"step": 1199
|
8435 |
+
},
|
8436 |
+
{
|
8437 |
+
"epoch": 0.2025060118972282,
|
8438 |
+
"grad_norm": 0.0016858198214322329,
|
8439 |
+
"learning_rate": 1.516707870871089e-05,
|
8440 |
+
"loss": 46.0,
|
8441 |
+
"step": 1200
|
8442 |
+
},
|
8443 |
+
{
|
8444 |
+
"epoch": 0.20267476690714256,
|
8445 |
+
"grad_norm": 0.0004934574826620519,
|
8446 |
+
"learning_rate": 1.5052325296826708e-05,
|
8447 |
+
"loss": 46.0,
|
8448 |
+
"step": 1201
|
8449 |
+
},
|
8450 |
+
{
|
8451 |
+
"epoch": 0.2028435219170569,
|
8452 |
+
"grad_norm": 0.0013131732121109962,
|
8453 |
+
"learning_rate": 1.4937972303325909e-05,
|
8454 |
+
"loss": 46.0,
|
8455 |
+
"step": 1202
|
8456 |
+
},
|
8457 |
+
{
|
8458 |
+
"epoch": 0.20301227692697127,
|
8459 |
+
"grad_norm": 0.0024639791809022427,
|
8460 |
+
"learning_rate": 1.4824020267234828e-05,
|
8461 |
+
"loss": 46.0,
|
8462 |
+
"step": 1203
|
8463 |
+
},
|
8464 |
+
{
|
8465 |
+
"epoch": 0.20318103193688564,
|
8466 |
+
"grad_norm": 0.001286798040382564,
|
8467 |
+
"learning_rate": 1.471046972568989e-05,
|
8468 |
+
"loss": 46.0,
|
8469 |
+
"step": 1204
|
8470 |
+
},
|
8471 |
+
{
|
8472 |
+
"epoch": 0.20334978694679998,
|
8473 |
+
"grad_norm": 0.0017689433880150318,
|
8474 |
+
"learning_rate": 1.4597321213935045e-05,
|
8475 |
+
"loss": 46.0,
|
8476 |
+
"step": 1205
|
8477 |
+
},
|
8478 |
+
{
|
8479 |
+
"epoch": 0.20351854195671434,
|
8480 |
+
"grad_norm": 0.0009689538856036961,
|
8481 |
+
"learning_rate": 1.4484575265319112e-05,
|
8482 |
+
"loss": 46.0,
|
8483 |
+
"step": 1206
|
8484 |
+
},
|
8485 |
+
{
|
8486 |
+
"epoch": 0.20368729696662868,
|
8487 |
+
"grad_norm": 0.0009614164009690285,
|
8488 |
+
"learning_rate": 1.4372232411293373e-05,
|
8489 |
+
"loss": 46.0,
|
8490 |
+
"step": 1207
|
8491 |
+
},
|
8492 |
+
{
|
8493 |
+
"epoch": 0.20385605197654305,
|
8494 |
+
"grad_norm": 0.0008921355474740267,
|
8495 |
+
"learning_rate": 1.4260293181409023e-05,
|
8496 |
+
"loss": 46.0,
|
8497 |
+
"step": 1208
|
8498 |
+
},
|
8499 |
+
{
|
8500 |
+
"epoch": 0.20402480698645742,
|
8501 |
+
"grad_norm": 0.0005731172277592123,
|
8502 |
+
"learning_rate": 1.414875810331473e-05,
|
8503 |
+
"loss": 46.0,
|
8504 |
+
"step": 1209
|
8505 |
+
},
|
8506 |
+
{
|
8507 |
+
"epoch": 0.20419356199637176,
|
8508 |
+
"grad_norm": 0.0005115303792990744,
|
8509 |
+
"learning_rate": 1.4037627702754064e-05,
|
8510 |
+
"loss": 46.0,
|
8511 |
+
"step": 1210
|
8512 |
+
},
|
8513 |
+
{
|
8514 |
+
"epoch": 0.20436231700628613,
|
8515 |
+
"grad_norm": 0.0010284315794706345,
|
8516 |
+
"learning_rate": 1.3926902503563099e-05,
|
8517 |
+
"loss": 46.0,
|
8518 |
+
"step": 1211
|
8519 |
+
},
|
8520 |
+
{
|
8521 |
+
"epoch": 0.2045310720162005,
|
8522 |
+
"grad_norm": 0.0005095731467008591,
|
8523 |
+
"learning_rate": 1.3816583027667895e-05,
|
8524 |
+
"loss": 46.0,
|
8525 |
+
"step": 1212
|
8526 |
+
},
|
8527 |
+
{
|
8528 |
+
"epoch": 0.20469982702611483,
|
8529 |
+
"grad_norm": 0.0006683744722977281,
|
8530 |
+
"learning_rate": 1.370666979508206e-05,
|
8531 |
+
"loss": 46.0,
|
8532 |
+
"step": 1213
|
8533 |
+
},
|
8534 |
+
{
|
8535 |
+
"epoch": 0.2048685820360292,
|
8536 |
+
"grad_norm": 0.0013482188805937767,
|
8537 |
+
"learning_rate": 1.3597163323904238e-05,
|
8538 |
+
"loss": 46.0,
|
8539 |
+
"step": 1214
|
8540 |
+
},
|
8541 |
+
{
|
8542 |
+
"epoch": 0.20503733704594354,
|
8543 |
+
"grad_norm": 0.0006872873054817319,
|
8544 |
+
"learning_rate": 1.3488064130315858e-05,
|
8545 |
+
"loss": 46.0,
|
8546 |
+
"step": 1215
|
8547 |
+
},
|
8548 |
+
{
|
8549 |
+
"epoch": 0.2052060920558579,
|
8550 |
+
"grad_norm": 0.000505985866766423,
|
8551 |
+
"learning_rate": 1.3379372728578387e-05,
|
8552 |
+
"loss": 46.0,
|
8553 |
+
"step": 1216
|
8554 |
+
},
|
8555 |
+
{
|
8556 |
+
"epoch": 0.20537484706577228,
|
8557 |
+
"grad_norm": 0.0005020827520638704,
|
8558 |
+
"learning_rate": 1.3271089631031152e-05,
|
8559 |
+
"loss": 46.0,
|
8560 |
+
"step": 1217
|
8561 |
+
},
|
8562 |
+
{
|
8563 |
+
"epoch": 0.20554360207568662,
|
8564 |
+
"grad_norm": 0.0010777136776596308,
|
8565 |
+
"learning_rate": 1.316321534808893e-05,
|
8566 |
+
"loss": 46.0,
|
8567 |
+
"step": 1218
|
8568 |
+
},
|
8569 |
+
{
|
8570 |
+
"epoch": 0.20571235708560098,
|
8571 |
+
"grad_norm": 0.0018740965751931071,
|
8572 |
+
"learning_rate": 1.3055750388239374e-05,
|
8573 |
+
"loss": 46.0,
|
8574 |
+
"step": 1219
|
8575 |
+
},
|
8576 |
+
{
|
8577 |
+
"epoch": 0.20588111209551532,
|
8578 |
+
"grad_norm": 0.0017112314235419035,
|
8579 |
+
"learning_rate": 1.2948695258040734e-05,
|
8580 |
+
"loss": 46.0,
|
8581 |
+
"step": 1220
|
8582 |
+
},
|
8583 |
+
{
|
8584 |
+
"epoch": 0.2060498671054297,
|
8585 |
+
"grad_norm": 0.0021086453925818205,
|
8586 |
+
"learning_rate": 1.2842050462119426e-05,
|
8587 |
+
"loss": 46.0,
|
8588 |
+
"step": 1221
|
8589 |
+
},
|
8590 |
+
{
|
8591 |
+
"epoch": 0.20621862211534406,
|
8592 |
+
"grad_norm": 0.0006901904125697911,
|
8593 |
+
"learning_rate": 1.2735816503167708e-05,
|
8594 |
+
"loss": 46.0,
|
8595 |
+
"step": 1222
|
8596 |
+
},
|
8597 |
+
{
|
8598 |
+
"epoch": 0.2063873771252584,
|
8599 |
+
"grad_norm": 0.0009343160782009363,
|
8600 |
+
"learning_rate": 1.2629993881941249e-05,
|
8601 |
+
"loss": 46.0,
|
8602 |
+
"step": 1223
|
8603 |
+
},
|
8604 |
+
{
|
8605 |
+
"epoch": 0.20655613213517277,
|
8606 |
+
"grad_norm": 0.0008830896113067865,
|
8607 |
+
"learning_rate": 1.2524583097256793e-05,
|
8608 |
+
"loss": 46.0,
|
8609 |
+
"step": 1224
|
8610 |
+
},
|
8611 |
+
{
|
8612 |
+
"epoch": 0.20672488714508713,
|
8613 |
+
"grad_norm": 0.0018291313899680972,
|
8614 |
+
"learning_rate": 1.2419584645989823e-05,
|
8615 |
+
"loss": 46.0,
|
8616 |
+
"step": 1225
|
8617 |
+
},
|
8618 |
+
{
|
8619 |
+
"epoch": 0.20689364215500147,
|
8620 |
+
"grad_norm": 0.00036748184356838465,
|
8621 |
+
"learning_rate": 1.2314999023072182e-05,
|
8622 |
+
"loss": 46.0,
|
8623 |
+
"step": 1226
|
8624 |
+
},
|
8625 |
+
{
|
8626 |
+
"epoch": 0.20706239716491584,
|
8627 |
+
"grad_norm": 0.0010965716792270541,
|
8628 |
+
"learning_rate": 1.2210826721489765e-05,
|
8629 |
+
"loss": 46.0,
|
8630 |
+
"step": 1227
|
8631 |
+
},
|
8632 |
+
{
|
8633 |
+
"epoch": 0.20723115217483018,
|
8634 |
+
"grad_norm": 0.0008960228879004717,
|
8635 |
+
"learning_rate": 1.2107068232280238e-05,
|
8636 |
+
"loss": 46.0,
|
8637 |
+
"step": 1228
|
8638 |
+
},
|
8639 |
+
{
|
8640 |
+
"epoch": 0.20739990718474455,
|
8641 |
+
"grad_norm": 0.004184830002486706,
|
8642 |
+
"learning_rate": 1.2003724044530596e-05,
|
8643 |
+
"loss": 46.0,
|
8644 |
+
"step": 1229
|
8645 |
+
},
|
8646 |
+
{
|
8647 |
+
"epoch": 0.20756866219465891,
|
8648 |
+
"grad_norm": 0.0006089484668336809,
|
8649 |
+
"learning_rate": 1.190079464537498e-05,
|
8650 |
+
"loss": 46.0,
|
8651 |
+
"step": 1230
|
8652 |
+
},
|
8653 |
+
{
|
8654 |
+
"epoch": 0.20773741720457325,
|
8655 |
+
"grad_norm": 0.0005351770669221878,
|
8656 |
+
"learning_rate": 1.1798280519992366e-05,
|
8657 |
+
"loss": 46.0,
|
8658 |
+
"step": 1231
|
8659 |
+
},
|
8660 |
+
{
|
8661 |
+
"epoch": 0.20790617221448762,
|
8662 |
+
"grad_norm": 0.000453771062893793,
|
8663 |
+
"learning_rate": 1.169618215160424e-05,
|
8664 |
+
"loss": 46.0,
|
8665 |
+
"step": 1232
|
8666 |
+
},
|
8667 |
+
{
|
8668 |
+
"epoch": 0.20807492722440196,
|
8669 |
+
"grad_norm": 0.0010923264781013131,
|
8670 |
+
"learning_rate": 1.1594500021472243e-05,
|
8671 |
+
"loss": 46.0,
|
8672 |
+
"step": 1233
|
8673 |
+
},
|
8674 |
+
{
|
8675 |
+
"epoch": 0.20824368223431633,
|
8676 |
+
"grad_norm": 0.0011471729958429933,
|
8677 |
+
"learning_rate": 1.1493234608896141e-05,
|
8678 |
+
"loss": 46.0,
|
8679 |
+
"step": 1234
|
8680 |
+
},
|
8681 |
+
{
|
8682 |
+
"epoch": 0.2084124372442307,
|
8683 |
+
"grad_norm": 0.0009632937726564705,
|
8684 |
+
"learning_rate": 1.1392386391211307e-05,
|
8685 |
+
"loss": 46.0,
|
8686 |
+
"step": 1235
|
8687 |
+
},
|
8688 |
+
{
|
8689 |
+
"epoch": 0.20858119225414504,
|
8690 |
+
"grad_norm": 0.0008420674130320549,
|
8691 |
+
"learning_rate": 1.1291955843786617e-05,
|
8692 |
+
"loss": 46.0,
|
8693 |
+
"step": 1236
|
8694 |
+
},
|
8695 |
+
{
|
8696 |
+
"epoch": 0.2087499472640594,
|
8697 |
+
"grad_norm": 0.0013222956331446767,
|
8698 |
+
"learning_rate": 1.1191943440022179e-05,
|
8699 |
+
"loss": 46.0,
|
8700 |
+
"step": 1237
|
8701 |
+
},
|
8702 |
+
{
|
8703 |
+
"epoch": 0.20891870227397377,
|
8704 |
+
"grad_norm": 0.001955078449100256,
|
8705 |
+
"learning_rate": 1.1092349651347055e-05,
|
8706 |
+
"loss": 46.0,
|
8707 |
+
"step": 1238
|
8708 |
+
},
|
8709 |
+
{
|
8710 |
+
"epoch": 0.2090874572838881,
|
8711 |
+
"grad_norm": 0.0019749640487134457,
|
8712 |
+
"learning_rate": 1.0993174947217144e-05,
|
8713 |
+
"loss": 46.0,
|
8714 |
+
"step": 1239
|
8715 |
+
},
|
8716 |
+
{
|
8717 |
+
"epoch": 0.20925621229380248,
|
8718 |
+
"grad_norm": 0.0010516421170905232,
|
8719 |
+
"learning_rate": 1.089441979511282e-05,
|
8720 |
+
"loss": 46.0,
|
8721 |
+
"step": 1240
|
8722 |
+
},
|
8723 |
+
{
|
8724 |
+
"epoch": 0.20942496730371682,
|
8725 |
+
"grad_norm": 0.0008922962588258088,
|
8726 |
+
"learning_rate": 1.079608466053692e-05,
|
8727 |
+
"loss": 46.0,
|
8728 |
+
"step": 1241
|
8729 |
+
},
|
8730 |
+
{
|
8731 |
+
"epoch": 0.2095937223136312,
|
8732 |
+
"grad_norm": 0.0005941609852015972,
|
8733 |
+
"learning_rate": 1.0698170007012332e-05,
|
8734 |
+
"loss": 46.0,
|
8735 |
+
"step": 1242
|
8736 |
+
},
|
8737 |
+
{
|
8738 |
+
"epoch": 0.20976247732354555,
|
8739 |
+
"grad_norm": 0.0004414799332153052,
|
8740 |
+
"learning_rate": 1.060067629607996e-05,
|
8741 |
+
"loss": 46.0,
|
8742 |
+
"step": 1243
|
8743 |
+
},
|
8744 |
+
{
|
8745 |
+
"epoch": 0.2099312323334599,
|
8746 |
+
"grad_norm": 0.002375362440943718,
|
8747 |
+
"learning_rate": 1.0503603987296562e-05,
|
8748 |
+
"loss": 46.0,
|
8749 |
+
"step": 1244
|
8750 |
+
},
|
8751 |
+
{
|
8752 |
+
"epoch": 0.21009998734337426,
|
8753 |
+
"grad_norm": 0.0010825851932168007,
|
8754 |
+
"learning_rate": 1.0406953538232479e-05,
|
8755 |
+
"loss": 46.0,
|
8756 |
+
"step": 1245
|
8757 |
+
},
|
8758 |
+
{
|
8759 |
+
"epoch": 0.2102687423532886,
|
8760 |
+
"grad_norm": 0.0012761763064190745,
|
8761 |
+
"learning_rate": 1.0310725404469479e-05,
|
8762 |
+
"loss": 46.0,
|
8763 |
+
"step": 1246
|
8764 |
+
},
|
8765 |
+
{
|
8766 |
+
"epoch": 0.21043749736320297,
|
8767 |
+
"grad_norm": 0.0008807352278381586,
|
8768 |
+
"learning_rate": 1.0214920039598774e-05,
|
8769 |
+
"loss": 46.0,
|
8770 |
+
"step": 1247
|
8771 |
+
},
|
8772 |
+
{
|
8773 |
+
"epoch": 0.21060625237311734,
|
8774 |
+
"grad_norm": 0.0006370813935063779,
|
8775 |
+
"learning_rate": 1.011953789521869e-05,
|
8776 |
+
"loss": 46.0,
|
8777 |
+
"step": 1248
|
8778 |
+
},
|
8779 |
+
{
|
8780 |
+
"epoch": 0.21077500738303168,
|
8781 |
+
"grad_norm": 0.0006687640561722219,
|
8782 |
+
"learning_rate": 1.0024579420932633e-05,
|
8783 |
+
"loss": 46.0,
|
8784 |
+
"step": 1249
|
8785 |
+
},
|
8786 |
+
{
|
8787 |
+
"epoch": 0.21094376239294604,
|
8788 |
+
"grad_norm": 0.0018508404027670622,
|
8789 |
+
"learning_rate": 9.93004506434696e-06,
|
8790 |
+
"loss": 46.0,
|
8791 |
+
"step": 1250
|
8792 |
+
},
|
8793 |
+
{
|
8794 |
+
"epoch": 0.2111125174028604,
|
8795 |
+
"grad_norm": 0.00048345469986088574,
|
8796 |
+
"learning_rate": 9.835935271068842e-06,
|
8797 |
+
"loss": 46.0,
|
8798 |
+
"step": 1251
|
8799 |
+
},
|
8800 |
+
{
|
8801 |
+
"epoch": 0.21128127241277475,
|
8802 |
+
"grad_norm": 0.0006890186341479421,
|
8803 |
+
"learning_rate": 9.74225048470422e-06,
|
8804 |
+
"loss": 46.0,
|
8805 |
+
"step": 1252
|
8806 |
+
},
|
8807 |
+
{
|
8808 |
+
"epoch": 0.21145002742268912,
|
8809 |
+
"grad_norm": 0.0008637924329377711,
|
8810 |
+
"learning_rate": 9.648991146855636e-06,
|
8811 |
+
"loss": 46.0,
|
8812 |
+
"step": 1253
|
8813 |
+
},
|
8814 |
+
{
|
8815 |
+
"epoch": 0.21161878243260346,
|
8816 |
+
"grad_norm": 0.0007931495201773942,
|
8817 |
+
"learning_rate": 9.556157697120215e-06,
|
8818 |
+
"loss": 46.0,
|
8819 |
+
"step": 1254
|
8820 |
+
},
|
8821 |
+
{
|
8822 |
+
"epoch": 0.21178753744251783,
|
8823 |
+
"grad_norm": 0.0005852460162714124,
|
8824 |
+
"learning_rate": 9.463750573087571e-06,
|
8825 |
+
"loss": 46.0,
|
8826 |
+
"step": 1255
|
8827 |
+
},
|
8828 |
+
{
|
8829 |
+
"epoch": 0.2119562924524322,
|
8830 |
+
"grad_norm": 0.0008610963704995811,
|
8831 |
+
"learning_rate": 9.371770210337727e-06,
|
8832 |
+
"loss": 46.0,
|
8833 |
+
"step": 1256
|
8834 |
+
},
|
8835 |
+
{
|
8836 |
+
"epoch": 0.21212504746234653,
|
8837 |
+
"grad_norm": 0.000635376840364188,
|
8838 |
+
"learning_rate": 9.280217042439122e-06,
|
8839 |
+
"loss": 46.0,
|
8840 |
+
"step": 1257
|
8841 |
+
},
|
8842 |
+
{
|
8843 |
+
"epoch": 0.2122938024722609,
|
8844 |
+
"grad_norm": 0.0008849898586049676,
|
8845 |
+
"learning_rate": 9.18909150094649e-06,
|
8846 |
+
"loss": 46.0,
|
8847 |
+
"step": 1258
|
8848 |
+
},
|
8849 |
+
{
|
8850 |
+
"epoch": 0.21246255748217524,
|
8851 |
+
"grad_norm": 0.0008719837060198188,
|
8852 |
+
"learning_rate": 9.098394015398814e-06,
|
8853 |
+
"loss": 46.0,
|
8854 |
+
"step": 1259
|
8855 |
+
},
|
8856 |
+
{
|
8857 |
+
"epoch": 0.2126313124920896,
|
8858 |
+
"grad_norm": 0.0021515190601348877,
|
8859 |
+
"learning_rate": 9.008125013317448e-06,
|
8860 |
+
"loss": 46.0,
|
8861 |
+
"step": 1260
|
8862 |
+
},
|
8863 |
+
{
|
8864 |
+
"epoch": 0.21280006750200398,
|
8865 |
+
"grad_norm": 0.00268219574354589,
|
8866 |
+
"learning_rate": 8.918284920203934e-06,
|
8867 |
+
"loss": 46.0,
|
8868 |
+
"step": 1261
|
8869 |
+
},
|
8870 |
+
{
|
8871 |
+
"epoch": 0.21296882251191832,
|
8872 |
+
"grad_norm": 0.0009798301616683602,
|
8873 |
+
"learning_rate": 8.828874159538091e-06,
|
8874 |
+
"loss": 46.0,
|
8875 |
+
"step": 1262
|
8876 |
+
},
|
8877 |
+
{
|
8878 |
+
"epoch": 0.21313757752183268,
|
8879 |
+
"grad_norm": 0.000541876011993736,
|
8880 |
+
"learning_rate": 8.739893152775958e-06,
|
8881 |
+
"loss": 46.0,
|
8882 |
+
"step": 1263
|
8883 |
+
},
|
8884 |
+
{
|
8885 |
+
"epoch": 0.21330633253174702,
|
8886 |
+
"grad_norm": 0.00041087419958785176,
|
8887 |
+
"learning_rate": 8.651342319347867e-06,
|
8888 |
+
"loss": 46.0,
|
8889 |
+
"step": 1264
|
8890 |
+
},
|
8891 |
+
{
|
8892 |
+
"epoch": 0.2134750875416614,
|
8893 |
+
"grad_norm": 0.0012815961381420493,
|
8894 |
+
"learning_rate": 8.563222076656429e-06,
|
8895 |
+
"loss": 46.0,
|
8896 |
+
"step": 1265
|
8897 |
+
},
|
8898 |
+
{
|
8899 |
+
"epoch": 0.21364384255157576,
|
8900 |
+
"grad_norm": 0.000593251024838537,
|
8901 |
+
"learning_rate": 8.475532840074585e-06,
|
8902 |
+
"loss": 46.0,
|
8903 |
+
"step": 1266
|
8904 |
+
},
|
8905 |
+
{
|
8906 |
+
"epoch": 0.2138125975614901,
|
8907 |
+
"grad_norm": 0.00048146312474273145,
|
8908 |
+
"learning_rate": 8.388275022943648e-06,
|
8909 |
+
"loss": 46.0,
|
8910 |
+
"step": 1267
|
8911 |
+
},
|
8912 |
+
{
|
8913 |
+
"epoch": 0.21398135257140447,
|
8914 |
+
"grad_norm": 0.0008700843318365514,
|
8915 |
+
"learning_rate": 8.301449036571319e-06,
|
8916 |
+
"loss": 46.0,
|
8917 |
+
"step": 1268
|
8918 |
+
},
|
8919 |
+
{
|
8920 |
+
"epoch": 0.21415010758131883,
|
8921 |
+
"grad_norm": 0.0005789704155176878,
|
8922 |
+
"learning_rate": 8.215055290229779e-06,
|
8923 |
+
"loss": 46.0,
|
8924 |
+
"step": 1269
|
8925 |
+
},
|
8926 |
+
{
|
8927 |
+
"epoch": 0.21431886259123317,
|
8928 |
+
"grad_norm": 0.0015106883365660906,
|
8929 |
+
"learning_rate": 8.129094191153808e-06,
|
8930 |
+
"loss": 46.0,
|
8931 |
+
"step": 1270
|
8932 |
+
},
|
8933 |
+
{
|
8934 |
+
"epoch": 0.21448761760114754,
|
8935 |
+
"grad_norm": 0.0004934304743073881,
|
8936 |
+
"learning_rate": 8.04356614453874e-06,
|
8937 |
+
"loss": 46.0,
|
8938 |
+
"step": 1271
|
8939 |
+
},
|
8940 |
+
{
|
8941 |
+
"epoch": 0.21465637261106188,
|
8942 |
+
"grad_norm": 0.0019693290814757347,
|
8943 |
+
"learning_rate": 7.95847155353865e-06,
|
8944 |
+
"loss": 46.0,
|
8945 |
+
"step": 1272
|
8946 |
+
},
|
8947 |
+
{
|
8948 |
+
"epoch": 0.21482512762097625,
|
8949 |
+
"grad_norm": 0.002005916088819504,
|
8950 |
+
"learning_rate": 7.873810819264483e-06,
|
8951 |
+
"loss": 46.0,
|
8952 |
+
"step": 1273
|
8953 |
+
},
|
8954 |
+
{
|
8955 |
+
"epoch": 0.21499388263089061,
|
8956 |
+
"grad_norm": 0.0012562709162011743,
|
8957 |
+
"learning_rate": 7.789584340782075e-06,
|
8958 |
+
"loss": 46.0,
|
8959 |
+
"step": 1274
|
8960 |
+
},
|
8961 |
+
{
|
8962 |
+
"epoch": 0.21516263764080495,
|
8963 |
+
"grad_norm": 0.0009011728689074516,
|
8964 |
+
"learning_rate": 7.705792515110232e-06,
|
8965 |
+
"loss": 46.0,
|
8966 |
+
"step": 1275
|
8967 |
+
},
|
8968 |
+
{
|
8969 |
+
"epoch": 0.21533139265071932,
|
8970 |
+
"grad_norm": 0.0014264247147366405,
|
8971 |
+
"learning_rate": 7.622435737219069e-06,
|
8972 |
+
"loss": 46.0,
|
8973 |
+
"step": 1276
|
8974 |
+
},
|
8975 |
+
{
|
8976 |
+
"epoch": 0.21550014766063366,
|
8977 |
+
"grad_norm": 0.0006953159463591874,
|
8978 |
+
"learning_rate": 7.539514400027925e-06,
|
8979 |
+
"loss": 46.0,
|
8980 |
+
"step": 1277
|
8981 |
+
},
|
8982 |
+
{
|
8983 |
+
"epoch": 0.21566890267054803,
|
8984 |
+
"grad_norm": 0.001039333757944405,
|
8985 |
+
"learning_rate": 7.457028894403628e-06,
|
8986 |
+
"loss": 46.0,
|
8987 |
+
"step": 1278
|
8988 |
+
},
|
8989 |
+
{
|
8990 |
+
"epoch": 0.2158376576804624,
|
8991 |
+
"grad_norm": 0.000679376011248678,
|
8992 |
+
"learning_rate": 7.374979609158617e-06,
|
8993 |
+
"loss": 46.0,
|
8994 |
+
"step": 1279
|
8995 |
+
},
|
8996 |
+
{
|
8997 |
+
"epoch": 0.21600641269037674,
|
8998 |
+
"grad_norm": 0.000941453967243433,
|
8999 |
+
"learning_rate": 7.293366931049128e-06,
|
9000 |
+
"loss": 46.0,
|
9001 |
+
"step": 1280
|
9002 |
+
},
|
9003 |
+
{
|
9004 |
+
"epoch": 0.2161751677002911,
|
9005 |
+
"grad_norm": 0.0007799931336194277,
|
9006 |
+
"learning_rate": 7.212191244773336e-06,
|
9007 |
+
"loss": 46.0,
|
9008 |
+
"step": 1281
|
9009 |
+
},
|
9010 |
+
{
|
9011 |
+
"epoch": 0.21634392271020547,
|
9012 |
+
"grad_norm": 0.0007033965666778386,
|
9013 |
+
"learning_rate": 7.131452932969595e-06,
|
9014 |
+
"loss": 46.0,
|
9015 |
+
"step": 1282
|
9016 |
+
},
|
9017 |
+
{
|
9018 |
+
"epoch": 0.2165126777201198,
|
9019 |
+
"grad_norm": 0.001040134928189218,
|
9020 |
+
"learning_rate": 7.051152376214565e-06,
|
9021 |
+
"loss": 46.0,
|
9022 |
+
"step": 1283
|
9023 |
+
},
|
9024 |
+
{
|
9025 |
+
"epoch": 0.21668143273003418,
|
9026 |
+
"grad_norm": 0.0005914743524044752,
|
9027 |
+
"learning_rate": 6.9712899530215095e-06,
|
9028 |
+
"loss": 46.0,
|
9029 |
+
"step": 1284
|
9030 |
+
},
|
9031 |
+
{
|
9032 |
+
"epoch": 0.21685018773994852,
|
9033 |
+
"grad_norm": 0.0005753316800110042,
|
9034 |
+
"learning_rate": 6.891866039838391e-06,
|
9035 |
+
"loss": 46.0,
|
9036 |
+
"step": 1285
|
9037 |
+
},
|
9038 |
+
{
|
9039 |
+
"epoch": 0.2170189427498629,
|
9040 |
+
"grad_norm": 0.00033438560785725713,
|
9041 |
+
"learning_rate": 6.812881011046246e-06,
|
9042 |
+
"loss": 46.0,
|
9043 |
+
"step": 1286
|
9044 |
+
},
|
9045 |
+
{
|
9046 |
+
"epoch": 0.21718769775977725,
|
9047 |
+
"grad_norm": 0.000527498428709805,
|
9048 |
+
"learning_rate": 6.734335238957301e-06,
|
9049 |
+
"loss": 46.0,
|
9050 |
+
"step": 1287
|
9051 |
+
},
|
9052 |
+
{
|
9053 |
+
"epoch": 0.2173564527696916,
|
9054 |
+
"grad_norm": 0.0015326968859881163,
|
9055 |
+
"learning_rate": 6.656229093813182e-06,
|
9056 |
+
"loss": 46.0,
|
9057 |
+
"step": 1288
|
9058 |
+
},
|
9059 |
+
{
|
9060 |
+
"epoch": 0.21752520777960596,
|
9061 |
+
"grad_norm": 0.0006215130561031401,
|
9062 |
+
"learning_rate": 6.578562943783384e-06,
|
9063 |
+
"loss": 46.0,
|
9064 |
+
"step": 1289
|
9065 |
+
},
|
9066 |
+
{
|
9067 |
+
"epoch": 0.2176939627895203,
|
9068 |
+
"grad_norm": 0.0009670006693340838,
|
9069 |
+
"learning_rate": 6.501337154963305e-06,
|
9070 |
+
"loss": 46.0,
|
9071 |
+
"step": 1290
|
9072 |
+
},
|
9073 |
+
{
|
9074 |
+
"epoch": 0.21786271779943467,
|
9075 |
+
"grad_norm": 0.0005261996411718428,
|
9076 |
+
"learning_rate": 6.424552091372604e-06,
|
9077 |
+
"loss": 46.0,
|
9078 |
+
"step": 1291
|
9079 |
+
},
|
9080 |
+
{
|
9081 |
+
"epoch": 0.21803147280934904,
|
9082 |
+
"grad_norm": 0.000595112273003906,
|
9083 |
+
"learning_rate": 6.3482081149535355e-06,
|
9084 |
+
"loss": 46.0,
|
9085 |
+
"step": 1292
|
9086 |
+
},
|
9087 |
+
{
|
9088 |
+
"epoch": 0.21820022781926338,
|
9089 |
+
"grad_norm": 0.000555274193175137,
|
9090 |
+
"learning_rate": 6.272305585569172e-06,
|
9091 |
+
"loss": 46.0,
|
9092 |
+
"step": 1293
|
9093 |
+
},
|
9094 |
+
{
|
9095 |
+
"epoch": 0.21836898282917774,
|
9096 |
+
"grad_norm": 0.0012713369214907289,
|
9097 |
+
"learning_rate": 6.196844861001727e-06,
|
9098 |
+
"loss": 46.0,
|
9099 |
+
"step": 1294
|
9100 |
+
},
|
9101 |
+
{
|
9102 |
+
"epoch": 0.2185377378390921,
|
9103 |
+
"grad_norm": 0.0006671750452369452,
|
9104 |
+
"learning_rate": 6.1218262969509055e-06,
|
9105 |
+
"loss": 46.0,
|
9106 |
+
"step": 1295
|
9107 |
+
},
|
9108 |
+
{
|
9109 |
+
"epoch": 0.21870649284900645,
|
9110 |
+
"grad_norm": 0.001461333711631596,
|
9111 |
+
"learning_rate": 6.047250247032165e-06,
|
9112 |
+
"loss": 46.0,
|
9113 |
+
"step": 1296
|
9114 |
+
},
|
9115 |
+
{
|
9116 |
+
"epoch": 0.21887524785892082,
|
9117 |
+
"grad_norm": 0.000863353256136179,
|
9118 |
+
"learning_rate": 5.973117062775113e-06,
|
9119 |
+
"loss": 46.0,
|
9120 |
+
"step": 1297
|
9121 |
+
},
|
9122 |
+
{
|
9123 |
+
"epoch": 0.21904400286883516,
|
9124 |
+
"grad_norm": 0.0012815648224204779,
|
9125 |
+
"learning_rate": 5.899427093621768e-06,
|
9126 |
+
"loss": 46.0,
|
9127 |
+
"step": 1298
|
9128 |
+
},
|
9129 |
+
{
|
9130 |
+
"epoch": 0.21921275787874953,
|
9131 |
+
"grad_norm": 0.0009904132457450032,
|
9132 |
+
"learning_rate": 5.826180686925031e-06,
|
9133 |
+
"loss": 46.0,
|
9134 |
+
"step": 1299
|
9135 |
+
},
|
9136 |
+
{
|
9137 |
+
"epoch": 0.2193815128886639,
|
9138 |
+
"grad_norm": 0.0006559500470757484,
|
9139 |
+
"learning_rate": 5.753378187946967e-06,
|
9140 |
+
"loss": 46.0,
|
9141 |
+
"step": 1300
|
9142 |
+
},
|
9143 |
+
{
|
9144 |
+
"epoch": 0.21955026789857823,
|
9145 |
+
"grad_norm": 0.001320411916822195,
|
9146 |
+
"learning_rate": 5.6810199398570905e-06,
|
9147 |
+
"loss": 46.0,
|
9148 |
+
"step": 1301
|
9149 |
+
},
|
9150 |
+
{
|
9151 |
+
"epoch": 0.2197190229084926,
|
9152 |
+
"grad_norm": 0.00033066104515455663,
|
9153 |
+
"learning_rate": 5.609106283730991e-06,
|
9154 |
+
"loss": 46.0,
|
9155 |
+
"step": 1302
|
9156 |
+
},
|
9157 |
+
{
|
9158 |
+
"epoch": 0.21988777791840694,
|
9159 |
+
"grad_norm": 0.0012177408207207918,
|
9160 |
+
"learning_rate": 5.537637558548525e-06,
|
9161 |
+
"loss": 46.0,
|
9162 |
+
"step": 1303
|
9163 |
+
},
|
9164 |
+
{
|
9165 |
+
"epoch": 0.2200565329283213,
|
9166 |
+
"grad_norm": 0.00042670281254686415,
|
9167 |
+
"learning_rate": 5.4666141011922025e-06,
|
9168 |
+
"loss": 46.0,
|
9169 |
+
"step": 1304
|
9170 |
+
},
|
9171 |
+
{
|
9172 |
+
"epoch": 0.22022528793823568,
|
9173 |
+
"grad_norm": 0.0011261850595474243,
|
9174 |
+
"learning_rate": 5.39603624644579e-06,
|
9175 |
+
"loss": 46.0,
|
9176 |
+
"step": 1305
|
9177 |
+
},
|
9178 |
+
{
|
9179 |
+
"epoch": 0.22039404294815002,
|
9180 |
+
"grad_norm": 0.0009061134187504649,
|
9181 |
+
"learning_rate": 5.325904326992548e-06,
|
9182 |
+
"loss": 46.0,
|
9183 |
+
"step": 1306
|
9184 |
+
},
|
9185 |
+
{
|
9186 |
+
"epoch": 0.22056279795806438,
|
9187 |
+
"grad_norm": 0.0008035211358219385,
|
9188 |
+
"learning_rate": 5.2562186734137485e-06,
|
9189 |
+
"loss": 46.0,
|
9190 |
+
"step": 1307
|
9191 |
+
},
|
9192 |
+
{
|
9193 |
+
"epoch": 0.22073155296797875,
|
9194 |
+
"grad_norm": 0.0004889973206445575,
|
9195 |
+
"learning_rate": 5.186979614187071e-06,
|
9196 |
+
"loss": 46.0,
|
9197 |
+
"step": 1308
|
9198 |
+
},
|
9199 |
+
{
|
9200 |
+
"epoch": 0.2209003079778931,
|
9201 |
+
"grad_norm": 0.0006611088174395263,
|
9202 |
+
"learning_rate": 5.118187475685121e-06,
|
9203 |
+
"loss": 46.0,
|
9204 |
+
"step": 1309
|
9205 |
+
},
|
9206 |
+
{
|
9207 |
+
"epoch": 0.22106906298780746,
|
9208 |
+
"grad_norm": 0.0021591607946902514,
|
9209 |
+
"learning_rate": 5.049842582173814e-06,
|
9210 |
+
"loss": 46.0,
|
9211 |
+
"step": 1310
|
9212 |
+
},
|
9213 |
+
{
|
9214 |
+
"epoch": 0.2212378179977218,
|
9215 |
+
"grad_norm": 0.0015975041314959526,
|
9216 |
+
"learning_rate": 4.9819452558109045e-06,
|
9217 |
+
"loss": 46.0,
|
9218 |
+
"step": 1311
|
9219 |
+
},
|
9220 |
+
{
|
9221 |
+
"epoch": 0.22140657300763616,
|
9222 |
+
"grad_norm": 0.00249864743091166,
|
9223 |
+
"learning_rate": 4.914495816644449e-06,
|
9224 |
+
"loss": 46.0,
|
9225 |
+
"step": 1312
|
9226 |
+
},
|
9227 |
+
{
|
9228 |
+
"epoch": 0.22157532801755053,
|
9229 |
+
"grad_norm": 0.0006162183126434684,
|
9230 |
+
"learning_rate": 4.847494582611301e-06,
|
9231 |
+
"loss": 46.0,
|
9232 |
+
"step": 1313
|
9233 |
+
},
|
9234 |
+
{
|
9235 |
+
"epoch": 0.22174408302746487,
|
9236 |
+
"grad_norm": 0.0007919514318928123,
|
9237 |
+
"learning_rate": 4.78094186953556e-06,
|
9238 |
+
"loss": 46.0,
|
9239 |
+
"step": 1314
|
9240 |
+
},
|
9241 |
+
{
|
9242 |
+
"epoch": 0.22191283803737924,
|
9243 |
+
"grad_norm": 0.0022155398037284613,
|
9244 |
+
"learning_rate": 4.714837991127186e-06,
|
9245 |
+
"loss": 46.0,
|
9246 |
+
"step": 1315
|
9247 |
+
},
|
9248 |
+
{
|
9249 |
+
"epoch": 0.22208159304729358,
|
9250 |
+
"grad_norm": 0.0005009054439142346,
|
9251 |
+
"learning_rate": 4.6491832589804605e-06,
|
9252 |
+
"loss": 46.0,
|
9253 |
+
"step": 1316
|
9254 |
+
},
|
9255 |
+
{
|
9256 |
+
"epoch": 0.22225034805720795,
|
9257 |
+
"grad_norm": 0.0015528866788372397,
|
9258 |
+
"learning_rate": 4.583977982572463e-06,
|
9259 |
+
"loss": 46.0,
|
9260 |
+
"step": 1317
|
9261 |
+
},
|
9262 |
+
{
|
9263 |
+
"epoch": 0.22241910306712231,
|
9264 |
+
"grad_norm": 0.0009101424948312342,
|
9265 |
+
"learning_rate": 4.519222469261731e-06,
|
9266 |
+
"loss": 46.0,
|
9267 |
+
"step": 1318
|
9268 |
+
},
|
9269 |
+
{
|
9270 |
+
"epoch": 0.22258785807703665,
|
9271 |
+
"grad_norm": 0.000738188624382019,
|
9272 |
+
"learning_rate": 4.454917024286742e-06,
|
9273 |
+
"loss": 46.0,
|
9274 |
+
"step": 1319
|
9275 |
+
},
|
9276 |
+
{
|
9277 |
+
"epoch": 0.22275661308695102,
|
9278 |
+
"grad_norm": 0.0008253224659711123,
|
9279 |
+
"learning_rate": 4.391061950764453e-06,
|
9280 |
+
"loss": 46.0,
|
9281 |
+
"step": 1320
|
9282 |
+
},
|
9283 |
+
{
|
9284 |
+
"epoch": 0.2229253680968654,
|
9285 |
+
"grad_norm": 0.001060123322531581,
|
9286 |
+
"learning_rate": 4.327657549688935e-06,
|
9287 |
+
"loss": 46.0,
|
9288 |
+
"step": 1321
|
9289 |
+
},
|
9290 |
+
{
|
9291 |
+
"epoch": 0.22309412310677973,
|
9292 |
+
"grad_norm": 0.0007293847156688571,
|
9293 |
+
"learning_rate": 4.264704119929897e-06,
|
9294 |
+
"loss": 46.0,
|
9295 |
+
"step": 1322
|
9296 |
+
},
|
9297 |
+
{
|
9298 |
+
"epoch": 0.2232628781166941,
|
9299 |
+
"grad_norm": 0.00065107416594401,
|
9300 |
+
"learning_rate": 4.202201958231322e-06,
|
9301 |
+
"loss": 46.0,
|
9302 |
+
"step": 1323
|
9303 |
+
},
|
9304 |
+
{
|
9305 |
+
"epoch": 0.22343163312660844,
|
9306 |
+
"grad_norm": 0.0006281050737015903,
|
9307 |
+
"learning_rate": 4.140151359210043e-06,
|
9308 |
+
"loss": 46.0,
|
9309 |
+
"step": 1324
|
9310 |
+
},
|
9311 |
+
{
|
9312 |
+
"epoch": 0.2236003881365228,
|
9313 |
+
"grad_norm": 0.000606034358497709,
|
9314 |
+
"learning_rate": 4.078552615354325e-06,
|
9315 |
+
"loss": 46.0,
|
9316 |
+
"step": 1325
|
9317 |
+
},
|
9318 |
+
{
|
9319 |
+
"epoch": 0.22376914314643717,
|
9320 |
+
"grad_norm": 0.0006844737217761576,
|
9321 |
+
"learning_rate": 4.017406017022607e-06,
|
9322 |
+
"loss": 46.0,
|
9323 |
+
"step": 1326
|
9324 |
+
},
|
9325 |
+
{
|
9326 |
+
"epoch": 0.2239378981563515,
|
9327 |
+
"grad_norm": 0.000884491775650531,
|
9328 |
+
"learning_rate": 3.95671185244193e-06,
|
9329 |
+
"loss": 46.0,
|
9330 |
+
"step": 1327
|
9331 |
+
},
|
9332 |
+
{
|
9333 |
+
"epoch": 0.22410665316626588,
|
9334 |
+
"grad_norm": 0.0005510865012183785,
|
9335 |
+
"learning_rate": 3.896470407706798e-06,
|
9336 |
+
"loss": 46.0,
|
9337 |
+
"step": 1328
|
9338 |
+
},
|
9339 |
+
{
|
9340 |
+
"epoch": 0.22427540817618022,
|
9341 |
+
"grad_norm": 0.0006715485942550004,
|
9342 |
+
"learning_rate": 3.836681966777678e-06,
|
9343 |
+
"loss": 46.0,
|
9344 |
+
"step": 1329
|
9345 |
+
},
|
9346 |
+
{
|
9347 |
+
"epoch": 0.2244441631860946,
|
9348 |
+
"grad_norm": 0.001269484986551106,
|
9349 |
+
"learning_rate": 3.7773468114796764e-06,
|
9350 |
+
"loss": 46.0,
|
9351 |
+
"step": 1330
|
9352 |
+
},
|
9353 |
+
{
|
9354 |
+
"epoch": 0.22461291819600895,
|
9355 |
+
"grad_norm": 0.0022162743844091892,
|
9356 |
+
"learning_rate": 3.718465221501344e-06,
|
9357 |
+
"loss": 46.0,
|
9358 |
+
"step": 1331
|
9359 |
+
},
|
9360 |
+
{
|
9361 |
+
"epoch": 0.2247816732059233,
|
9362 |
+
"grad_norm": 0.00027326264535076916,
|
9363 |
+
"learning_rate": 3.660037474393152e-06,
|
9364 |
+
"loss": 46.0,
|
9365 |
+
"step": 1332
|
9366 |
+
},
|
9367 |
+
{
|
9368 |
+
"epoch": 0.22495042821583766,
|
9369 |
+
"grad_norm": 0.0005874920170754194,
|
9370 |
+
"learning_rate": 3.602063845566361e-06,
|
9371 |
+
"loss": 46.0,
|
9372 |
+
"step": 1333
|
9373 |
+
},
|
9374 |
+
{
|
9375 |
+
"epoch": 0.225119183225752,
|
9376 |
+
"grad_norm": 0.0009573047864250839,
|
9377 |
+
"learning_rate": 3.544544608291622e-06,
|
9378 |
+
"loss": 46.0,
|
9379 |
+
"step": 1334
|
9380 |
+
},
|
9381 |
+
{
|
9382 |
+
"epoch": 0.22528793823566637,
|
9383 |
+
"grad_norm": 0.0007656016387045383,
|
9384 |
+
"learning_rate": 3.487480033697721e-06,
|
9385 |
+
"loss": 46.0,
|
9386 |
+
"step": 1335
|
9387 |
+
},
|
9388 |
+
{
|
9389 |
+
"epoch": 0.22545669324558074,
|
9390 |
+
"grad_norm": 0.00044490606524050236,
|
9391 |
+
"learning_rate": 3.430870390770291e-06,
|
9392 |
+
"loss": 46.0,
|
9393 |
+
"step": 1336
|
9394 |
+
},
|
9395 |
+
{
|
9396 |
+
"epoch": 0.22562544825549508,
|
9397 |
+
"grad_norm": 0.0003960455651395023,
|
9398 |
+
"learning_rate": 3.3747159463505595e-06,
|
9399 |
+
"loss": 46.0,
|
9400 |
+
"step": 1337
|
9401 |
+
},
|
9402 |
+
{
|
9403 |
+
"epoch": 0.22579420326540944,
|
9404 |
+
"grad_norm": 0.000521904497873038,
|
9405 |
+
"learning_rate": 3.3190169651340806e-06,
|
9406 |
+
"loss": 46.0,
|
9407 |
+
"step": 1338
|
9408 |
+
},
|
9409 |
+
{
|
9410 |
+
"epoch": 0.2259629582753238,
|
9411 |
+
"grad_norm": 0.0004184047575108707,
|
9412 |
+
"learning_rate": 3.2637737096695154e-06,
|
9413 |
+
"loss": 46.0,
|
9414 |
+
"step": 1339
|
9415 |
+
},
|
9416 |
+
{
|
9417 |
+
"epoch": 0.22613171328523815,
|
9418 |
+
"grad_norm": 0.001061237882822752,
|
9419 |
+
"learning_rate": 3.2089864403572976e-06,
|
9420 |
+
"loss": 46.0,
|
9421 |
+
"step": 1340
|
9422 |
+
},
|
9423 |
+
{
|
9424 |
+
"epoch": 0.22630046829515252,
|
9425 |
+
"grad_norm": 0.000829793163575232,
|
9426 |
+
"learning_rate": 3.1546554154485486e-06,
|
9427 |
+
"loss": 46.0,
|
9428 |
+
"step": 1341
|
9429 |
+
},
|
9430 |
+
{
|
9431 |
+
"epoch": 0.22646922330506686,
|
9432 |
+
"grad_norm": 0.0010989286238327622,
|
9433 |
+
"learning_rate": 3.100780891043764e-06,
|
9434 |
+
"loss": 46.0,
|
9435 |
+
"step": 1342
|
9436 |
+
},
|
9437 |
+
{
|
9438 |
+
"epoch": 0.22663797831498123,
|
9439 |
+
"grad_norm": 0.000658164091873914,
|
9440 |
+
"learning_rate": 3.0473631210915954e-06,
|
9441 |
+
"loss": 46.0,
|
9442 |
+
"step": 1343
|
9443 |
+
},
|
9444 |
+
{
|
9445 |
+
"epoch": 0.2268067333248956,
|
9446 |
+
"grad_norm": 0.0018878680421039462,
|
9447 |
+
"learning_rate": 2.994402357387738e-06,
|
9448 |
+
"loss": 46.0,
|
9449 |
+
"step": 1344
|
9450 |
+
},
|
9451 |
+
{
|
9452 |
+
"epoch": 0.22697548833480993,
|
9453 |
+
"grad_norm": 0.0011438899673521519,
|
9454 |
+
"learning_rate": 2.9418988495736765e-06,
|
9455 |
+
"loss": 46.0,
|
9456 |
+
"step": 1345
|
9457 |
+
},
|
9458 |
+
{
|
9459 |
+
"epoch": 0.2271442433447243,
|
9460 |
+
"grad_norm": 0.0005721378256566823,
|
9461 |
+
"learning_rate": 2.889852845135521e-06,
|
9462 |
+
"loss": 46.0,
|
9463 |
+
"step": 1346
|
9464 |
+
},
|
9465 |
+
{
|
9466 |
+
"epoch": 0.22731299835463864,
|
9467 |
+
"grad_norm": 0.0004976931377314031,
|
9468 |
+
"learning_rate": 2.8382645894028393e-06,
|
9469 |
+
"loss": 46.0,
|
9470 |
+
"step": 1347
|
9471 |
+
},
|
9472 |
+
{
|
9473 |
+
"epoch": 0.227481753364553,
|
9474 |
+
"grad_norm": 0.00046737625962123275,
|
9475 |
+
"learning_rate": 2.7871343255475135e-06,
|
9476 |
+
"loss": 46.0,
|
9477 |
+
"step": 1348
|
9478 |
+
},
|
9479 |
+
{
|
9480 |
+
"epoch": 0.22765050837446738,
|
9481 |
+
"grad_norm": 0.0006955791031941772,
|
9482 |
+
"learning_rate": 2.736462294582598e-06,
|
9483 |
+
"loss": 46.0,
|
9484 |
+
"step": 1349
|
9485 |
+
},
|
9486 |
+
{
|
9487 |
+
"epoch": 0.22781926338438171,
|
9488 |
+
"grad_norm": 0.0017836546758189797,
|
9489 |
+
"learning_rate": 2.686248735361141e-06,
|
9490 |
+
"loss": 46.0,
|
9491 |
+
"step": 1350
|
9492 |
+
},
|
9493 |
+
{
|
9494 |
+
"epoch": 0.22798801839429608,
|
9495 |
+
"grad_norm": 0.000547305797226727,
|
9496 |
+
"learning_rate": 2.6364938845751196e-06,
|
9497 |
+
"loss": 46.0,
|
9498 |
+
"step": 1351
|
9499 |
+
},
|
9500 |
+
{
|
9501 |
+
"epoch": 0.22815677340421045,
|
9502 |
+
"grad_norm": 0.0018898543203249574,
|
9503 |
+
"learning_rate": 2.5871979767543077e-06,
|
9504 |
+
"loss": 46.0,
|
9505 |
+
"step": 1352
|
9506 |
+
},
|
9507 |
+
{
|
9508 |
+
"epoch": 0.2283255284141248,
|
9509 |
+
"grad_norm": 0.0005670526879839599,
|
9510 |
+
"learning_rate": 2.5383612442650973e-06,
|
9511 |
+
"loss": 46.0,
|
9512 |
+
"step": 1353
|
9513 |
+
},
|
9514 |
+
{
|
9515 |
+
"epoch": 0.22849428342403916,
|
9516 |
+
"grad_norm": 0.0006443196325562894,
|
9517 |
+
"learning_rate": 2.489983917309502e-06,
|
9518 |
+
"loss": 46.0,
|
9519 |
+
"step": 1354
|
9520 |
+
},
|
9521 |
+
{
|
9522 |
+
"epoch": 0.2286630384339535,
|
9523 |
+
"grad_norm": 0.000439947412814945,
|
9524 |
+
"learning_rate": 2.4420662239240666e-06,
|
9525 |
+
"loss": 46.0,
|
9526 |
+
"step": 1355
|
9527 |
+
},
|
9528 |
+
{
|
9529 |
+
"epoch": 0.22883179344386786,
|
9530 |
+
"grad_norm": 0.0012257567141205072,
|
9531 |
+
"learning_rate": 2.3946083899786697e-06,
|
9532 |
+
"loss": 46.0,
|
9533 |
+
"step": 1356
|
9534 |
+
},
|
9535 |
+
{
|
9536 |
+
"epoch": 0.22900054845378223,
|
9537 |
+
"grad_norm": 0.0011901329271495342,
|
9538 |
+
"learning_rate": 2.3476106391756347e-06,
|
9539 |
+
"loss": 46.0,
|
9540 |
+
"step": 1357
|
9541 |
+
},
|
9542 |
+
{
|
9543 |
+
"epoch": 0.22916930346369657,
|
9544 |
+
"grad_norm": 0.0009685796103440225,
|
9545 |
+
"learning_rate": 2.3010731930485417e-06,
|
9546 |
+
"loss": 46.0,
|
9547 |
+
"step": 1358
|
9548 |
+
},
|
9549 |
+
{
|
9550 |
+
"epoch": 0.22933805847361094,
|
9551 |
+
"grad_norm": 0.0011787050170823932,
|
9552 |
+
"learning_rate": 2.2549962709612293e-06,
|
9553 |
+
"loss": 46.0,
|
9554 |
+
"step": 1359
|
9555 |
+
},
|
9556 |
+
{
|
9557 |
+
"epoch": 0.22950681348352528,
|
9558 |
+
"grad_norm": 0.0006927854265086353,
|
9559 |
+
"learning_rate": 2.209380090106794e-06,
|
9560 |
+
"loss": 46.0,
|
9561 |
+
"step": 1360
|
9562 |
+
},
|
9563 |
+
{
|
9564 |
+
"epoch": 0.22967556849343965,
|
9565 |
+
"grad_norm": 0.0021828326862305403,
|
9566 |
+
"learning_rate": 2.164224865506492e-06,
|
9567 |
+
"loss": 46.0,
|
9568 |
+
"step": 1361
|
9569 |
+
},
|
9570 |
+
{
|
9571 |
+
"epoch": 0.22984432350335401,
|
9572 |
+
"grad_norm": 0.0011952252825722098,
|
9573 |
+
"learning_rate": 2.1195308100087964e-06,
|
9574 |
+
"loss": 46.0,
|
9575 |
+
"step": 1362
|
9576 |
+
},
|
9577 |
+
{
|
9578 |
+
"epoch": 0.23001307851326835,
|
9579 |
+
"grad_norm": 0.0008620006265118718,
|
9580 |
+
"learning_rate": 2.0752981342883504e-06,
|
9581 |
+
"loss": 46.0,
|
9582 |
+
"step": 1363
|
9583 |
+
},
|
9584 |
+
{
|
9585 |
+
"epoch": 0.23018183352318272,
|
9586 |
+
"grad_norm": 0.0007816816796548665,
|
9587 |
+
"learning_rate": 2.031527046844994e-06,
|
9588 |
+
"loss": 46.0,
|
9589 |
+
"step": 1364
|
9590 |
+
},
|
9591 |
+
{
|
9592 |
+
"epoch": 0.2303505885330971,
|
9593 |
+
"grad_norm": 0.00032747103250585496,
|
9594 |
+
"learning_rate": 1.9882177540027856e-06,
|
9595 |
+
"loss": 46.0,
|
9596 |
+
"step": 1365
|
9597 |
+
},
|
9598 |
+
{
|
9599 |
+
"epoch": 0.23051934354301143,
|
9600 |
+
"grad_norm": 0.0009276359342038631,
|
9601 |
+
"learning_rate": 1.9453704599089906e-06,
|
9602 |
+
"loss": 46.0,
|
9603 |
+
"step": 1366
|
9604 |
+
},
|
9605 |
+
{
|
9606 |
+
"epoch": 0.2306880985529258,
|
9607 |
+
"grad_norm": 0.0007578957010991871,
|
9608 |
+
"learning_rate": 1.902985366533172e-06,
|
9609 |
+
"loss": 46.0,
|
9610 |
+
"step": 1367
|
9611 |
+
},
|
9612 |
+
{
|
9613 |
+
"epoch": 0.23085685356284014,
|
9614 |
+
"grad_norm": 0.0009802145650610328,
|
9615 |
+
"learning_rate": 1.8610626736662252e-06,
|
9616 |
+
"loss": 46.0,
|
9617 |
+
"step": 1368
|
9618 |
+
},
|
9619 |
+
{
|
9620 |
+
"epoch": 0.2310256085727545,
|
9621 |
+
"grad_norm": 0.0017398048657923937,
|
9622 |
+
"learning_rate": 1.8196025789193771e-06,
|
9623 |
+
"loss": 46.0,
|
9624 |
+
"step": 1369
|
9625 |
+
},
|
9626 |
+
{
|
9627 |
+
"epoch": 0.23119436358266887,
|
9628 |
+
"grad_norm": 0.001226657535880804,
|
9629 |
+
"learning_rate": 1.7786052777233663e-06,
|
9630 |
+
"loss": 46.0,
|
9631 |
+
"step": 1370
|
9632 |
+
},
|
9633 |
+
{
|
9634 |
+
"epoch": 0.2313631185925832,
|
9635 |
+
"grad_norm": 0.0007180358516052365,
|
9636 |
+
"learning_rate": 1.7380709633274095e-06,
|
9637 |
+
"loss": 46.0,
|
9638 |
+
"step": 1371
|
9639 |
+
},
|
9640 |
+
{
|
9641 |
+
"epoch": 0.23153187360249758,
|
9642 |
+
"grad_norm": 0.002623442793264985,
|
9643 |
+
"learning_rate": 1.6979998267983577e-06,
|
9644 |
+
"loss": 46.0,
|
9645 |
+
"step": 1372
|
9646 |
+
},
|
9647 |
+
{
|
9648 |
+
"epoch": 0.23170062861241192,
|
9649 |
+
"grad_norm": 0.0011678035371005535,
|
9650 |
+
"learning_rate": 1.6583920570197752e-06,
|
9651 |
+
"loss": 46.0,
|
9652 |
+
"step": 1373
|
9653 |
+
},
|
9654 |
+
{
|
9655 |
+
"epoch": 0.23186938362232629,
|
9656 |
+
"grad_norm": 0.0005888472078368068,
|
9657 |
+
"learning_rate": 1.6192478406910626e-06,
|
9658 |
+
"loss": 46.0,
|
9659 |
+
"step": 1374
|
9660 |
+
},
|
9661 |
+
{
|
9662 |
+
"epoch": 0.23203813863224065,
|
9663 |
+
"grad_norm": 0.001251881243661046,
|
9664 |
+
"learning_rate": 1.5805673623265572e-06,
|
9665 |
+
"loss": 46.0,
|
9666 |
+
"step": 1375
|
9667 |
+
},
|
9668 |
+
{
|
9669 |
+
"epoch": 0.232206893642155,
|
9670 |
+
"grad_norm": 0.0005478327511809766,
|
9671 |
+
"learning_rate": 1.5423508042546553e-06,
|
9672 |
+
"loss": 46.0,
|
9673 |
+
"step": 1376
|
9674 |
+
},
|
9675 |
+
{
|
9676 |
+
"epoch": 0.23237564865206936,
|
9677 |
+
"grad_norm": 0.0005645382916554809,
|
9678 |
+
"learning_rate": 1.504598346616981e-06,
|
9679 |
+
"loss": 46.0,
|
9680 |
+
"step": 1377
|
9681 |
+
},
|
9682 |
+
{
|
9683 |
+
"epoch": 0.23254440366198373,
|
9684 |
+
"grad_norm": 0.0011800862848758698,
|
9685 |
+
"learning_rate": 1.4673101673675639e-06,
|
9686 |
+
"loss": 46.0,
|
9687 |
+
"step": 1378
|
9688 |
+
},
|
9689 |
+
{
|
9690 |
+
"epoch": 0.23271315867189807,
|
9691 |
+
"grad_norm": 0.0021919957362115383,
|
9692 |
+
"learning_rate": 1.4304864422718945e-06,
|
9693 |
+
"loss": 46.0,
|
9694 |
+
"step": 1379
|
9695 |
+
},
|
9696 |
+
{
|
9697 |
+
"epoch": 0.23288191368181244,
|
9698 |
+
"grad_norm": 0.000592592463362962,
|
9699 |
+
"learning_rate": 1.3941273449062041e-06,
|
9700 |
+
"loss": 46.0,
|
9701 |
+
"step": 1380
|
9702 |
+
},
|
9703 |
+
{
|
9704 |
+
"epoch": 0.23305066869172678,
|
9705 |
+
"grad_norm": 0.002734198933467269,
|
9706 |
+
"learning_rate": 1.3582330466565985e-06,
|
9707 |
+
"loss": 46.0,
|
9708 |
+
"step": 1381
|
9709 |
+
},
|
9710 |
+
{
|
9711 |
+
"epoch": 0.23321942370164114,
|
9712 |
+
"grad_norm": 0.0006839183042757213,
|
9713 |
+
"learning_rate": 1.3228037167182573e-06,
|
9714 |
+
"loss": 46.0,
|
9715 |
+
"step": 1382
|
9716 |
+
},
|
9717 |
+
{
|
9718 |
+
"epoch": 0.2333881787115555,
|
9719 |
+
"grad_norm": 0.0003775713557843119,
|
9720 |
+
"learning_rate": 1.2878395220946248e-06,
|
9721 |
+
"loss": 46.0,
|
9722 |
+
"step": 1383
|
9723 |
+
},
|
9724 |
+
{
|
9725 |
+
"epoch": 0.23355693372146985,
|
9726 |
+
"grad_norm": 0.00045846737339161336,
|
9727 |
+
"learning_rate": 1.253340627596644e-06,
|
9728 |
+
"loss": 46.0,
|
9729 |
+
"step": 1384
|
9730 |
+
},
|
9731 |
+
{
|
9732 |
+
"epoch": 0.23372568873138422,
|
9733 |
+
"grad_norm": 0.0007842537015676498,
|
9734 |
+
"learning_rate": 1.2193071958419788e-06,
|
9735 |
+
"loss": 46.0,
|
9736 |
+
"step": 1385
|
9737 |
+
},
|
9738 |
+
{
|
9739 |
+
"epoch": 0.23389444374129856,
|
9740 |
+
"grad_norm": 0.0012929689837619662,
|
9741 |
+
"learning_rate": 1.1857393872542255e-06,
|
9742 |
+
"loss": 46.0,
|
9743 |
+
"step": 1386
|
9744 |
+
},
|
9745 |
+
{
|
9746 |
+
"epoch": 0.23406319875121293,
|
9747 |
+
"grad_norm": 0.0015821013366803527,
|
9748 |
+
"learning_rate": 1.1526373600621699e-06,
|
9749 |
+
"loss": 46.0,
|
9750 |
+
"step": 1387
|
9751 |
+
},
|
9752 |
+
{
|
9753 |
+
"epoch": 0.2342319537611273,
|
9754 |
+
"grad_norm": 0.0006062084576115012,
|
9755 |
+
"learning_rate": 1.1200012702990425e-06,
|
9756 |
+
"loss": 46.0,
|
9757 |
+
"step": 1388
|
9758 |
+
},
|
9759 |
+
{
|
9760 |
+
"epoch": 0.23440070877104163,
|
9761 |
+
"grad_norm": 0.0006992828566581011,
|
9762 |
+
"learning_rate": 1.0878312718017868e-06,
|
9763 |
+
"loss": 46.0,
|
9764 |
+
"step": 1389
|
9765 |
+
},
|
9766 |
+
{
|
9767 |
+
"epoch": 0.234569463780956,
|
9768 |
+
"grad_norm": 0.001030724961310625,
|
9769 |
+
"learning_rate": 1.0561275162103035e-06,
|
9770 |
+
"loss": 46.0,
|
9771 |
+
"step": 1390
|
9772 |
+
},
|
9773 |
+
{
|
9774 |
+
"epoch": 0.23473821879087037,
|
9775 |
+
"grad_norm": 0.001839980366639793,
|
9776 |
+
"learning_rate": 1.0248901529668286e-06,
|
9777 |
+
"loss": 46.0,
|
9778 |
+
"step": 1391
|
9779 |
+
},
|
9780 |
+
{
|
9781 |
+
"epoch": 0.2349069738007847,
|
9782 |
+
"grad_norm": 0.0017480964306741953,
|
9783 |
+
"learning_rate": 9.941193293150796e-07,
|
9784 |
+
"loss": 46.0,
|
9785 |
+
"step": 1392
|
9786 |
+
},
|
9787 |
+
{
|
9788 |
+
"epoch": 0.23507572881069907,
|
9789 |
+
"grad_norm": 0.0008748589316383004,
|
9790 |
+
"learning_rate": 9.638151902996773e-07,
|
9791 |
+
"loss": 46.0,
|
9792 |
+
"step": 1393
|
9793 |
+
},
|
9794 |
+
{
|
9795 |
+
"epoch": 0.23524448382061341,
|
9796 |
+
"grad_norm": 0.0008244823548011482,
|
9797 |
+
"learning_rate": 9.339778787654574e-07,
|
9798 |
+
"loss": 46.0,
|
9799 |
+
"step": 1394
|
9800 |
+
},
|
9801 |
+
{
|
9802 |
+
"epoch": 0.23541323883052778,
|
9803 |
+
"grad_norm": 0.001938622328452766,
|
9804 |
+
"learning_rate": 9.046075353567163e-07,
|
9805 |
+
"loss": 46.0,
|
9806 |
+
"step": 1395
|
9807 |
+
},
|
9808 |
+
{
|
9809 |
+
"epoch": 0.23558199384044215,
|
9810 |
+
"grad_norm": 0.0008106924360617995,
|
9811 |
+
"learning_rate": 8.757042985166331e-07,
|
9812 |
+
"loss": 46.0,
|
9813 |
+
"step": 1396
|
9814 |
+
},
|
9815 |
+
{
|
9816 |
+
"epoch": 0.2357507488503565,
|
9817 |
+
"grad_norm": 0.0010104329558089375,
|
9818 |
+
"learning_rate": 8.472683044865815e-07,
|
9819 |
+
"loss": 46.0,
|
9820 |
+
"step": 1397
|
9821 |
+
},
|
9822 |
+
{
|
9823 |
+
"epoch": 0.23591950386027086,
|
9824 |
+
"grad_norm": 0.0008030526805669069,
|
9825 |
+
"learning_rate": 8.192996873054748e-07,
|
9826 |
+
"loss": 46.0,
|
9827 |
+
"step": 1398
|
9828 |
+
},
|
9829 |
+
{
|
9830 |
+
"epoch": 0.2360882588701852,
|
9831 |
+
"grad_norm": 0.003335982095450163,
|
9832 |
+
"learning_rate": 7.917985788091553e-07,
|
9833 |
+
"loss": 46.0,
|
9834 |
+
"step": 1399
|
9835 |
+
},
|
9836 |
+
{
|
9837 |
+
"epoch": 0.23625701388009956,
|
9838 |
+
"grad_norm": 0.0006814859807491302,
|
9839 |
+
"learning_rate": 7.647651086297835e-07,
|
9840 |
+
"loss": 46.0,
|
9841 |
+
"step": 1400
|
9842 |
+
},
|
9843 |
+
{
|
9844 |
+
"epoch": 0.23642576889001393,
|
9845 |
+
"grad_norm": 0.001208942150697112,
|
9846 |
+
"learning_rate": 7.381994041951945e-07,
|
9847 |
+
"loss": 46.0,
|
9848 |
+
"step": 1401
|
9849 |
+
},
|
9850 |
+
{
|
9851 |
+
"epoch": 0.23659452389992827,
|
9852 |
+
"grad_norm": 0.0026390880811959505,
|
9853 |
+
"learning_rate": 7.12101590728298e-07,
|
9854 |
+
"loss": 46.0,
|
9855 |
+
"step": 1402
|
9856 |
+
},
|
9857 |
+
{
|
9858 |
+
"epoch": 0.23676327890984264,
|
9859 |
+
"grad_norm": 0.0010781860910356045,
|
9860 |
+
"learning_rate": 6.864717912465346e-07,
|
9861 |
+
"loss": 46.0,
|
9862 |
+
"step": 1403
|
9863 |
+
},
|
9864 |
+
{
|
9865 |
+
"epoch": 0.23693203391975698,
|
9866 |
+
"grad_norm": 0.001180395483970642,
|
9867 |
+
"learning_rate": 6.613101265612431e-07,
|
9868 |
+
"loss": 46.0,
|
9869 |
+
"step": 1404
|
9870 |
+
},
|
9871 |
+
{
|
9872 |
+
"epoch": 0.23710078892967135,
|
9873 |
+
"grad_norm": 0.0006595318554900587,
|
9874 |
+
"learning_rate": 6.366167152771274e-07,
|
9875 |
+
"loss": 46.0,
|
9876 |
+
"step": 1405
|
9877 |
+
},
|
9878 |
+
{
|
9879 |
+
"epoch": 0.23726954393958571,
|
9880 |
+
"grad_norm": 0.0005041745025664568,
|
9881 |
+
"learning_rate": 6.123916737916568e-07,
|
9882 |
+
"loss": 46.0,
|
9883 |
+
"step": 1406
|
9884 |
+
},
|
9885 |
+
{
|
9886 |
+
"epoch": 0.23743829894950005,
|
9887 |
+
"grad_norm": 0.0013777145650237799,
|
9888 |
+
"learning_rate": 5.886351162945891e-07,
|
9889 |
+
"loss": 46.0,
|
9890 |
+
"step": 1407
|
9891 |
+
},
|
9892 |
+
{
|
9893 |
+
"epoch": 0.23760705395941442,
|
9894 |
+
"grad_norm": 0.0009335106587968767,
|
9895 |
+
"learning_rate": 5.653471547673262e-07,
|
9896 |
+
"loss": 46.0,
|
9897 |
+
"step": 1408
|
9898 |
+
},
|
9899 |
+
{
|
9900 |
+
"epoch": 0.2377758089693288,
|
9901 |
+
"grad_norm": 0.0007560536614619195,
|
9902 |
+
"learning_rate": 5.425278989824922e-07,
|
9903 |
+
"loss": 46.0,
|
9904 |
+
"step": 1409
|
9905 |
+
},
|
9906 |
+
{
|
9907 |
+
"epoch": 0.23794456397924313,
|
9908 |
+
"grad_norm": 0.0007272138609550893,
|
9909 |
+
"learning_rate": 5.201774565033679e-07,
|
9910 |
+
"loss": 46.0,
|
9911 |
+
"step": 1410
|
9912 |
+
},
|
9913 |
+
{
|
9914 |
+
"epoch": 0.2381133189891575,
|
9915 |
+
"grad_norm": 0.0035329603124409914,
|
9916 |
+
"learning_rate": 4.982959326833347e-07,
|
9917 |
+
"loss": 46.0,
|
9918 |
+
"step": 1411
|
9919 |
+
},
|
9920 |
+
{
|
9921 |
+
"epoch": 0.23828207399907184,
|
9922 |
+
"grad_norm": 0.0006473190151154995,
|
9923 |
+
"learning_rate": 4.76883430665509e-07,
|
9924 |
+
"loss": 46.0,
|
9925 |
+
"step": 1412
|
9926 |
+
},
|
9927 |
+
{
|
9928 |
+
"epoch": 0.2384508290089862,
|
9929 |
+
"grad_norm": 0.0016907091485336423,
|
9930 |
+
"learning_rate": 4.5594005138211993e-07,
|
9931 |
+
"loss": 46.0,
|
9932 |
+
"step": 1413
|
9933 |
+
},
|
9934 |
+
{
|
9935 |
+
"epoch": 0.23861958401890057,
|
9936 |
+
"grad_norm": 0.0016014818102121353,
|
9937 |
+
"learning_rate": 4.3546589355409894e-07,
|
9938 |
+
"loss": 46.0,
|
9939 |
+
"step": 1414
|
9940 |
+
},
|
9941 |
+
{
|
9942 |
+
"epoch": 0.2387883390288149,
|
9943 |
+
"grad_norm": 0.0006536655710078776,
|
9944 |
+
"learning_rate": 4.154610536906134e-07,
|
9945 |
+
"loss": 46.0,
|
9946 |
+
"step": 1415
|
9947 |
+
},
|
9948 |
+
{
|
9949 |
+
"epoch": 0.23895709403872928,
|
9950 |
+
"grad_norm": 0.0007482526707462966,
|
9951 |
+
"learning_rate": 3.959256260886113e-07,
|
9952 |
+
"loss": 46.0,
|
9953 |
+
"step": 1416
|
9954 |
+
},
|
9955 |
+
{
|
9956 |
+
"epoch": 0.23912584904864362,
|
9957 |
+
"grad_norm": 0.0012404083972796798,
|
9958 |
+
"learning_rate": 3.7685970283234395e-07,
|
9959 |
+
"loss": 46.0,
|
9960 |
+
"step": 1417
|
9961 |
+
},
|
9962 |
+
{
|
9963 |
+
"epoch": 0.23929460405855799,
|
9964 |
+
"grad_norm": 0.0006575691513717175,
|
9965 |
+
"learning_rate": 3.5826337379297746e-07,
|
9966 |
+
"loss": 46.0,
|
9967 |
+
"step": 1418
|
9968 |
+
},
|
9969 |
+
{
|
9970 |
+
"epoch": 0.23946335906847235,
|
9971 |
+
"grad_norm": 0.0005846316926181316,
|
9972 |
+
"learning_rate": 3.4013672662815964e-07,
|
9973 |
+
"loss": 46.0,
|
9974 |
+
"step": 1419
|
9975 |
+
},
|
9976 |
+
{
|
9977 |
+
"epoch": 0.2396321140783867,
|
9978 |
+
"grad_norm": 0.0015353142516687512,
|
9979 |
+
"learning_rate": 3.2247984678155374e-07,
|
9980 |
+
"loss": 46.0,
|
9981 |
+
"step": 1420
|
9982 |
+
},
|
9983 |
+
{
|
9984 |
+
"epoch": 0.23980086908830106,
|
9985 |
+
"grad_norm": 0.0007021796191111207,
|
9986 |
+
"learning_rate": 3.0529281748252757e-07,
|
9987 |
+
"loss": 46.0,
|
9988 |
+
"step": 1421
|
9989 |
+
},
|
9990 |
+
{
|
9991 |
+
"epoch": 0.23996962409821543,
|
9992 |
+
"grad_norm": 0.00043599194032140076,
|
9993 |
+
"learning_rate": 2.8857571974567623e-07,
|
9994 |
+
"loss": 46.0,
|
9995 |
+
"step": 1422
|
9996 |
+
},
|
9997 |
+
{
|
9998 |
+
"epoch": 0.24013837910812977,
|
9999 |
+
"grad_norm": 0.0008538305992260575,
|
10000 |
+
"learning_rate": 2.723286323704888e-07,
|
10001 |
+
"loss": 46.0,
|
10002 |
+
"step": 1423
|
10003 |
+
},
|
10004 |
+
{
|
10005 |
+
"epoch": 0.24030713411804414,
|
10006 |
+
"grad_norm": 0.0007377176661975682,
|
10007 |
+
"learning_rate": 2.565516319409711e-07,
|
10008 |
+
"loss": 46.0,
|
10009 |
+
"step": 1424
|
10010 |
+
},
|
10011 |
+
{
|
10012 |
+
"epoch": 0.24047588912795848,
|
10013 |
+
"grad_norm": 0.001053746440447867,
|
10014 |
+
"learning_rate": 2.412447928252459e-07,
|
10015 |
+
"loss": 46.0,
|
10016 |
+
"step": 1425
|
10017 |
+
},
|
10018 |
+
{
|
10019 |
+
"epoch": 0.24064464413787284,
|
10020 |
+
"grad_norm": 0.00034214206971228123,
|
10021 |
+
"learning_rate": 2.2640818717527546e-07,
|
10022 |
+
"loss": 46.0,
|
10023 |
+
"step": 1426
|
10024 |
+
},
|
10025 |
+
{
|
10026 |
+
"epoch": 0.2408133991477872,
|
10027 |
+
"grad_norm": 0.0006055811536498368,
|
10028 |
+
"learning_rate": 2.120418849264616e-07,
|
10029 |
+
"loss": 46.0,
|
10030 |
+
"step": 1427
|
10031 |
+
},
|
10032 |
+
{
|
10033 |
+
"epoch": 0.24098215415770155,
|
10034 |
+
"grad_norm": 0.0032110221218317747,
|
10035 |
+
"learning_rate": 1.9814595379732405e-07,
|
10036 |
+
"loss": 46.0,
|
10037 |
+
"step": 1428
|
10038 |
+
},
|
10039 |
+
{
|
10040 |
+
"epoch": 0.24115090916761592,
|
10041 |
+
"grad_norm": 0.0006545573123730719,
|
10042 |
+
"learning_rate": 1.8472045928920045e-07,
|
10043 |
+
"loss": 46.0,
|
10044 |
+
"step": 1429
|
10045 |
+
},
|
10046 |
+
{
|
10047 |
+
"epoch": 0.24131966417753026,
|
10048 |
+
"grad_norm": 0.0010023179929703474,
|
10049 |
+
"learning_rate": 1.7176546468592457e-07,
|
10050 |
+
"loss": 46.0,
|
10051 |
+
"step": 1430
|
10052 |
+
},
|
10053 |
+
{
|
10054 |
+
"epoch": 0.24148841918744463,
|
10055 |
+
"grad_norm": 0.0009001618018373847,
|
10056 |
+
"learning_rate": 1.5928103105354864e-07,
|
10057 |
+
"loss": 46.0,
|
10058 |
+
"step": 1431
|
10059 |
+
},
|
10060 |
+
{
|
10061 |
+
"epoch": 0.241657174197359,
|
10062 |
+
"grad_norm": 0.000825360300950706,
|
10063 |
+
"learning_rate": 1.4726721724001024e-07,
|
10064 |
+
"loss": 46.0,
|
10065 |
+
"step": 1432
|
10066 |
+
},
|
10067 |
+
{
|
10068 |
+
"epoch": 0.24182592920727333,
|
10069 |
+
"grad_norm": 0.0018932815873995423,
|
10070 |
+
"learning_rate": 1.3572407987491042e-07,
|
10071 |
+
"loss": 46.0,
|
10072 |
+
"step": 1433
|
10073 |
+
},
|
10074 |
+
{
|
10075 |
+
"epoch": 0.2419946842171877,
|
10076 |
+
"grad_norm": 0.0010684579610824585,
|
10077 |
+
"learning_rate": 1.2465167336920268e-07,
|
10078 |
+
"loss": 46.0,
|
10079 |
+
"step": 1434
|
10080 |
+
},
|
10081 |
+
{
|
10082 |
+
"epoch": 0.24216343922710207,
|
10083 |
+
"grad_norm": 0.0006767999730072916,
|
10084 |
+
"learning_rate": 1.1405004991495993e-07,
|
10085 |
+
"loss": 46.0,
|
10086 |
+
"step": 1435
|
10087 |
+
},
|
10088 |
+
{
|
10089 |
+
"epoch": 0.2423321942370164,
|
10090 |
+
"grad_norm": 0.0002995872055180371,
|
10091 |
+
"learning_rate": 1.0391925948511904e-07,
|
10092 |
+
"loss": 46.0,
|
10093 |
+
"step": 1436
|
10094 |
+
},
|
10095 |
+
{
|
10096 |
+
"epoch": 0.24250094924693077,
|
10097 |
+
"grad_norm": 0.0010398075683042407,
|
10098 |
+
"learning_rate": 9.425934983323669e-08,
|
10099 |
+
"loss": 46.0,
|
10100 |
+
"step": 1437
|
10101 |
+
},
|
10102 |
+
{
|
10103 |
+
"epoch": 0.24266970425684511,
|
10104 |
+
"grad_norm": 0.0006404778105206788,
|
10105 |
+
"learning_rate": 8.507036649331168e-08,
|
10106 |
+
"loss": 46.0,
|
10107 |
+
"step": 1438
|
10108 |
+
},
|
10109 |
+
{
|
10110 |
+
"epoch": 0.24283845926675948,
|
10111 |
+
"grad_norm": 0.0009962361073121428,
|
10112 |
+
"learning_rate": 7.635235277950736e-08,
|
10113 |
+
"loss": 46.0,
|
10114 |
+
"step": 1439
|
10115 |
+
},
|
10116 |
+
{
|
10117 |
+
"epoch": 0.24300721427667385,
|
10118 |
+
"grad_norm": 0.0007267189212143421,
|
10119 |
+
"learning_rate": 6.810534978597404e-08,
|
10120 |
+
"loss": 46.0,
|
10121 |
+
"step": 1440
|
10122 |
+
},
|
10123 |
+
{
|
10124 |
+
"epoch": 0.2431759692865882,
|
10125 |
+
"grad_norm": 0.000724569137673825,
|
10126 |
+
"learning_rate": 6.032939638664914e-08,
|
10127 |
+
"loss": 46.0,
|
10128 |
+
"step": 1441
|
10129 |
+
},
|
10130 |
+
{
|
10131 |
+
"epoch": 0.24334472429650256,
|
10132 |
+
"grad_norm": 0.0003782059939112514,
|
10133 |
+
"learning_rate": 5.3024529235112807e-08,
|
10134 |
+
"loss": 46.0,
|
10135 |
+
"step": 1442
|
10136 |
+
},
|
10137 |
+
{
|
10138 |
+
"epoch": 0.2435134793064169,
|
10139 |
+
"grad_norm": 0.001214990857988596,
|
10140 |
+
"learning_rate": 4.619078276436595e-08,
|
10141 |
+
"loss": 46.0,
|
10142 |
+
"step": 1443
|
10143 |
+
},
|
10144 |
+
{
|
10145 |
+
"epoch": 0.24368223431633126,
|
10146 |
+
"grad_norm": 0.0013174681225791574,
|
10147 |
+
"learning_rate": 3.982818918665254e-08,
|
10148 |
+
"loss": 46.0,
|
10149 |
+
"step": 1444
|
10150 |
+
},
|
10151 |
+
{
|
10152 |
+
"epoch": 0.24385098932624563,
|
10153 |
+
"grad_norm": 0.0019085375824943185,
|
10154 |
+
"learning_rate": 3.393677849340415e-08,
|
10155 |
+
"loss": 46.0,
|
10156 |
+
"step": 1445
|
10157 |
+
},
|
10158 |
+
{
|
10159 |
+
"epoch": 0.24401974433615997,
|
10160 |
+
"grad_norm": 0.0005340630887076259,
|
10161 |
+
"learning_rate": 2.8516578454973465e-08,
|
10162 |
+
"loss": 46.0,
|
10163 |
+
"step": 1446
|
10164 |
+
},
|
10165 |
+
{
|
10166 |
+
"epoch": 0.24418849934607434,
|
10167 |
+
"grad_norm": 0.0010641046101227403,
|
10168 |
+
"learning_rate": 2.3567614620600975e-08,
|
10169 |
+
"loss": 46.0,
|
10170 |
+
"step": 1447
|
10171 |
+
},
|
10172 |
+
{
|
10173 |
+
"epoch": 0.2443572543559887,
|
10174 |
+
"grad_norm": 0.0006958871381357312,
|
10175 |
+
"learning_rate": 1.9089910318259572e-08,
|
10176 |
+
"loss": 46.0,
|
10177 |
+
"step": 1448
|
10178 |
+
},
|
10179 |
+
{
|
10180 |
+
"epoch": 0.24452600936590305,
|
10181 |
+
"grad_norm": 0.0013515378814190626,
|
10182 |
+
"learning_rate": 1.5083486654510203e-08,
|
10183 |
+
"loss": 46.0,
|
10184 |
+
"step": 1449
|
10185 |
+
},
|
10186 |
+
{
|
10187 |
+
"epoch": 0.24469476437581741,
|
10188 |
+
"grad_norm": 0.0014520830009132624,
|
10189 |
+
"learning_rate": 1.1548362514479661e-08,
|
10190 |
+
"loss": 46.0,
|
10191 |
+
"step": 1450
|
10192 |
+
},
|
10193 |
+
{
|
10194 |
+
"epoch": 0.24486351938573175,
|
10195 |
+
"grad_norm": 0.0015810457989573479,
|
10196 |
+
"learning_rate": 8.484554561682956e-09,
|
10197 |
+
"loss": 46.0,
|
10198 |
+
"step": 1451
|
10199 |
+
},
|
10200 |
+
{
|
10201 |
+
"epoch": 0.24503227439564612,
|
10202 |
+
"grad_norm": 0.001024417346343398,
|
10203 |
+
"learning_rate": 5.8920772380233195e-09,
|
10204 |
+
"loss": 46.0,
|
10205 |
+
"step": 1452
|
10206 |
+
},
|
10207 |
+
{
|
10208 |
+
"epoch": 0.2452010294055605,
|
10209 |
+
"grad_norm": 0.000991427805274725,
|
10210 |
+
"learning_rate": 3.7709427636700714e-09,
|
10211 |
+
"loss": 46.0,
|
10212 |
+
"step": 1453
|
10213 |
+
},
|
10214 |
+
{
|
10215 |
+
"epoch": 0.24536978441547483,
|
10216 |
+
"grad_norm": 0.001307089813053608,
|
10217 |
+
"learning_rate": 2.1211611370475226e-09,
|
10218 |
+
"loss": 46.0,
|
10219 |
+
"step": 1454
|
10220 |
+
},
|
10221 |
+
{
|
10222 |
+
"epoch": 0.2455385394253892,
|
10223 |
+
"grad_norm": 0.0008045461145229638,
|
10224 |
+
"learning_rate": 9.427401347128495e-10,
|
10225 |
+
"loss": 46.0,
|
10226 |
+
"step": 1455
|
10227 |
+
},
|
10228 |
+
{
|
10229 |
+
"epoch": 0.24570729443530354,
|
10230 |
+
"grad_norm": 0.0017816101899370551,
|
10231 |
+
"learning_rate": 2.356853114116042e-10,
|
10232 |
+
"loss": 46.0,
|
10233 |
+
"step": 1456
|
10234 |
+
},
|
10235 |
+
{
|
10236 |
+
"epoch": 0.2458760494452179,
|
10237 |
+
"grad_norm": 0.000787016935646534,
|
10238 |
+
"learning_rate": 0.0,
|
10239 |
+
"loss": 46.0,
|
10240 |
+
"step": 1457
|
10241 |
}
|
10242 |
],
|
10243 |
"logging_steps": 1,
|
|
|
10252 |
"should_evaluate": false,
|
10253 |
"should_log": false,
|
10254 |
"should_save": true,
|
10255 |
+
"should_training_stop": true
|
10256 |
},
|
10257 |
"attributes": {}
|
10258 |
}
|
10259 |
},
|
10260 |
+
"total_flos": 29027633922048.0,
|
10261 |
"train_batch_size": 2,
|
10262 |
"trial_name": null,
|
10263 |
"trial_params": null
|