Training in progress, step 1095, 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 sha256:9f992333d1ac6c619130e6ab9b8ad83b17635dafa74a4211d798a8b491e46009
|
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 sha256:5ca9fcaacb5832c4c5f23c40e996a41930cd35c3e789c48269dbe7ca2817e3e2
|
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 sha256:091cf724662e7257b815a6442e352013aac78a6c32e5577cdf81716e826ca76f
|
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 sha256:55f30c05338451a8e814015d126dfcb6b31fd73b6ba75f24a484a1ad0c936fec
|
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,
|
@@ -5141,6 +5141,2569 @@
|
|
5141 |
"eval_samples_per_second": 175.97,
|
5142 |
"eval_steps_per_second": 88.021,
|
5143 |
"step": 730
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5144 |
}
|
5145 |
],
|
5146 |
"logging_steps": 1,
|
@@ -5160,7 +7723,7 @@
|
|
5160 |
"attributes": {}
|
5161 |
}
|
5162 |
},
|
5163 |
-
"total_flos":
|
5164 |
"train_batch_size": 2,
|
5165 |
"trial_name": null,
|
5166 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.18478673585622074,
|
5 |
"eval_steps": 365,
|
6 |
+
"global_step": 1095,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
5141 |
"eval_samples_per_second": 175.97,
|
5142 |
"eval_steps_per_second": 88.021,
|
5143 |
"step": 730
|
5144 |
+
},
|
5145 |
+
{
|
5146 |
+
"epoch": 0.12335991224739484,
|
5147 |
+
"grad_norm": 0.0005169582436792552,
|
5148 |
+
"learning_rate": 0.00010054277422737369,
|
5149 |
+
"loss": 46.0,
|
5150 |
+
"step": 731
|
5151 |
+
},
|
5152 |
+
{
|
5153 |
+
"epoch": 0.1235286672573092,
|
5154 |
+
"grad_norm": 0.00028787710471078753,
|
5155 |
+
"learning_rate": 0.0001003256655598189,
|
5156 |
+
"loss": 46.0,
|
5157 |
+
"step": 732
|
5158 |
+
},
|
5159 |
+
{
|
5160 |
+
"epoch": 0.12369742226722356,
|
5161 |
+
"grad_norm": 0.0016796085983514786,
|
5162 |
+
"learning_rate": 0.00010010855535717232,
|
5163 |
+
"loss": 46.0,
|
5164 |
+
"step": 733
|
5165 |
+
},
|
5166 |
+
{
|
5167 |
+
"epoch": 0.12386617727713792,
|
5168 |
+
"grad_norm": 0.0016076072352007031,
|
5169 |
+
"learning_rate": 9.98914446428277e-05,
|
5170 |
+
"loss": 46.0,
|
5171 |
+
"step": 734
|
5172 |
+
},
|
5173 |
+
{
|
5174 |
+
"epoch": 0.12403493228705227,
|
5175 |
+
"grad_norm": 0.0008219809387810528,
|
5176 |
+
"learning_rate": 9.967433444018111e-05,
|
5177 |
+
"loss": 46.0,
|
5178 |
+
"step": 735
|
5179 |
+
},
|
5180 |
+
{
|
5181 |
+
"epoch": 0.12420368729696663,
|
5182 |
+
"grad_norm": 0.002049337374046445,
|
5183 |
+
"learning_rate": 9.945722577262632e-05,
|
5184 |
+
"loss": 46.0,
|
5185 |
+
"step": 736
|
5186 |
+
},
|
5187 |
+
{
|
5188 |
+
"epoch": 0.12437244230688098,
|
5189 |
+
"grad_norm": 0.0005754379089921713,
|
5190 |
+
"learning_rate": 9.92401196635498e-05,
|
5191 |
+
"loss": 46.0,
|
5192 |
+
"step": 737
|
5193 |
+
},
|
5194 |
+
{
|
5195 |
+
"epoch": 0.12454119731679535,
|
5196 |
+
"grad_norm": 0.0013358064461499453,
|
5197 |
+
"learning_rate": 9.902301713632592e-05,
|
5198 |
+
"loss": 46.0,
|
5199 |
+
"step": 738
|
5200 |
+
},
|
5201 |
+
{
|
5202 |
+
"epoch": 0.1247099523267097,
|
5203 |
+
"grad_norm": 0.0006682523526251316,
|
5204 |
+
"learning_rate": 9.880591921431227e-05,
|
5205 |
+
"loss": 46.0,
|
5206 |
+
"step": 739
|
5207 |
+
},
|
5208 |
+
{
|
5209 |
+
"epoch": 0.12487870733662405,
|
5210 |
+
"grad_norm": 0.00043337634997442365,
|
5211 |
+
"learning_rate": 9.858882692084466e-05,
|
5212 |
+
"loss": 46.0,
|
5213 |
+
"step": 740
|
5214 |
+
},
|
5215 |
+
{
|
5216 |
+
"epoch": 0.1250474623465384,
|
5217 |
+
"grad_norm": 0.0005919807590544224,
|
5218 |
+
"learning_rate": 9.837174127923237e-05,
|
5219 |
+
"loss": 46.0,
|
5220 |
+
"step": 741
|
5221 |
+
},
|
5222 |
+
{
|
5223 |
+
"epoch": 0.12521621735645277,
|
5224 |
+
"grad_norm": 0.0013382198521867394,
|
5225 |
+
"learning_rate": 9.815466331275335e-05,
|
5226 |
+
"loss": 46.0,
|
5227 |
+
"step": 742
|
5228 |
+
},
|
5229 |
+
{
|
5230 |
+
"epoch": 0.12538497236636711,
|
5231 |
+
"grad_norm": 0.000782837625592947,
|
5232 |
+
"learning_rate": 9.793759404464936e-05,
|
5233 |
+
"loss": 46.0,
|
5234 |
+
"step": 743
|
5235 |
+
},
|
5236 |
+
{
|
5237 |
+
"epoch": 0.12555372737628148,
|
5238 |
+
"grad_norm": 0.0009372693602927029,
|
5239 |
+
"learning_rate": 9.772053449812118e-05,
|
5240 |
+
"loss": 46.0,
|
5241 |
+
"step": 744
|
5242 |
+
},
|
5243 |
+
{
|
5244 |
+
"epoch": 0.12572248238619585,
|
5245 |
+
"grad_norm": 0.000794100807979703,
|
5246 |
+
"learning_rate": 9.750348569632375e-05,
|
5247 |
+
"loss": 46.0,
|
5248 |
+
"step": 745
|
5249 |
+
},
|
5250 |
+
{
|
5251 |
+
"epoch": 0.1258912373961102,
|
5252 |
+
"grad_norm": 0.0010177150834351778,
|
5253 |
+
"learning_rate": 9.728644866236132e-05,
|
5254 |
+
"loss": 46.0,
|
5255 |
+
"step": 746
|
5256 |
+
},
|
5257 |
+
{
|
5258 |
+
"epoch": 0.12605999240602456,
|
5259 |
+
"grad_norm": 0.0010124148102477193,
|
5260 |
+
"learning_rate": 9.706942441928273e-05,
|
5261 |
+
"loss": 46.0,
|
5262 |
+
"step": 747
|
5263 |
+
},
|
5264 |
+
{
|
5265 |
+
"epoch": 0.12622874741593892,
|
5266 |
+
"grad_norm": 0.0010855060536414385,
|
5267 |
+
"learning_rate": 9.68524139900765e-05,
|
5268 |
+
"loss": 46.0,
|
5269 |
+
"step": 748
|
5270 |
+
},
|
5271 |
+
{
|
5272 |
+
"epoch": 0.12639750242585326,
|
5273 |
+
"grad_norm": 0.0008628432988189161,
|
5274 |
+
"learning_rate": 9.66354183976661e-05,
|
5275 |
+
"loss": 46.0,
|
5276 |
+
"step": 749
|
5277 |
+
},
|
5278 |
+
{
|
5279 |
+
"epoch": 0.12656625743576763,
|
5280 |
+
"grad_norm": 0.0007373031694442034,
|
5281 |
+
"learning_rate": 9.641843866490492e-05,
|
5282 |
+
"loss": 46.0,
|
5283 |
+
"step": 750
|
5284 |
+
},
|
5285 |
+
{
|
5286 |
+
"epoch": 0.12673501244568197,
|
5287 |
+
"grad_norm": 0.0007096838089637458,
|
5288 |
+
"learning_rate": 9.62014758145717e-05,
|
5289 |
+
"loss": 46.0,
|
5290 |
+
"step": 751
|
5291 |
+
},
|
5292 |
+
{
|
5293 |
+
"epoch": 0.12690376745559634,
|
5294 |
+
"grad_norm": 0.0009518765727989376,
|
5295 |
+
"learning_rate": 9.598453086936559e-05,
|
5296 |
+
"loss": 46.0,
|
5297 |
+
"step": 752
|
5298 |
+
},
|
5299 |
+
{
|
5300 |
+
"epoch": 0.1270725224655107,
|
5301 |
+
"grad_norm": 0.0013600359670817852,
|
5302 |
+
"learning_rate": 9.576760485190137e-05,
|
5303 |
+
"loss": 46.0,
|
5304 |
+
"step": 753
|
5305 |
+
},
|
5306 |
+
{
|
5307 |
+
"epoch": 0.12724127747542505,
|
5308 |
+
"grad_norm": 0.0011481187539175153,
|
5309 |
+
"learning_rate": 9.555069878470449e-05,
|
5310 |
+
"loss": 46.0,
|
5311 |
+
"step": 754
|
5312 |
+
},
|
5313 |
+
{
|
5314 |
+
"epoch": 0.12741003248533941,
|
5315 |
+
"grad_norm": 0.000521646230481565,
|
5316 |
+
"learning_rate": 9.533381369020646e-05,
|
5317 |
+
"loss": 46.0,
|
5318 |
+
"step": 755
|
5319 |
+
},
|
5320 |
+
{
|
5321 |
+
"epoch": 0.12757878749525375,
|
5322 |
+
"grad_norm": 0.0005510274204425514,
|
5323 |
+
"learning_rate": 9.511695059073989e-05,
|
5324 |
+
"loss": 46.0,
|
5325 |
+
"step": 756
|
5326 |
+
},
|
5327 |
+
{
|
5328 |
+
"epoch": 0.12774754250516812,
|
5329 |
+
"grad_norm": 0.003858919721096754,
|
5330 |
+
"learning_rate": 9.490011050853375e-05,
|
5331 |
+
"loss": 46.0,
|
5332 |
+
"step": 757
|
5333 |
+
},
|
5334 |
+
{
|
5335 |
+
"epoch": 0.1279162975150825,
|
5336 |
+
"grad_norm": 0.0031276738736778498,
|
5337 |
+
"learning_rate": 9.468329446570848e-05,
|
5338 |
+
"loss": 46.0,
|
5339 |
+
"step": 758
|
5340 |
+
},
|
5341 |
+
{
|
5342 |
+
"epoch": 0.12808505252499683,
|
5343 |
+
"grad_norm": 0.0007159564993344247,
|
5344 |
+
"learning_rate": 9.446650348427117e-05,
|
5345 |
+
"loss": 46.0,
|
5346 |
+
"step": 759
|
5347 |
+
},
|
5348 |
+
{
|
5349 |
+
"epoch": 0.1282538075349112,
|
5350 |
+
"grad_norm": 0.0005575700779445469,
|
5351 |
+
"learning_rate": 9.424973858611088e-05,
|
5352 |
+
"loss": 46.0,
|
5353 |
+
"step": 760
|
5354 |
+
},
|
5355 |
+
{
|
5356 |
+
"epoch": 0.12842256254482554,
|
5357 |
+
"grad_norm": 0.00020999553089495748,
|
5358 |
+
"learning_rate": 9.403300079299364e-05,
|
5359 |
+
"loss": 46.0,
|
5360 |
+
"step": 761
|
5361 |
+
},
|
5362 |
+
{
|
5363 |
+
"epoch": 0.1285913175547399,
|
5364 |
+
"grad_norm": 0.0008280250476673245,
|
5365 |
+
"learning_rate": 9.38162911265577e-05,
|
5366 |
+
"loss": 46.0,
|
5367 |
+
"step": 762
|
5368 |
+
},
|
5369 |
+
{
|
5370 |
+
"epoch": 0.12876007256465427,
|
5371 |
+
"grad_norm": 0.0007614243077114224,
|
5372 |
+
"learning_rate": 9.35996106083088e-05,
|
5373 |
+
"loss": 46.0,
|
5374 |
+
"step": 763
|
5375 |
+
},
|
5376 |
+
{
|
5377 |
+
"epoch": 0.1289288275745686,
|
5378 |
+
"grad_norm": 0.0007390084792859852,
|
5379 |
+
"learning_rate": 9.338296025961521e-05,
|
5380 |
+
"loss": 46.0,
|
5381 |
+
"step": 764
|
5382 |
+
},
|
5383 |
+
{
|
5384 |
+
"epoch": 0.12909758258448298,
|
5385 |
+
"grad_norm": 0.0005527955945581198,
|
5386 |
+
"learning_rate": 9.31663411017031e-05,
|
5387 |
+
"loss": 46.0,
|
5388 |
+
"step": 765
|
5389 |
+
},
|
5390 |
+
{
|
5391 |
+
"epoch": 0.12926633759439735,
|
5392 |
+
"grad_norm": 0.0007797405123710632,
|
5393 |
+
"learning_rate": 9.294975415565151e-05,
|
5394 |
+
"loss": 46.0,
|
5395 |
+
"step": 766
|
5396 |
+
},
|
5397 |
+
{
|
5398 |
+
"epoch": 0.12943509260431169,
|
5399 |
+
"grad_norm": 0.0008938325918279588,
|
5400 |
+
"learning_rate": 9.273320044238763e-05,
|
5401 |
+
"loss": 46.0,
|
5402 |
+
"step": 767
|
5403 |
+
},
|
5404 |
+
{
|
5405 |
+
"epoch": 0.12960384761422605,
|
5406 |
+
"grad_norm": 0.0011889681918546557,
|
5407 |
+
"learning_rate": 9.251668098268209e-05,
|
5408 |
+
"loss": 46.0,
|
5409 |
+
"step": 768
|
5410 |
+
},
|
5411 |
+
{
|
5412 |
+
"epoch": 0.1297726026241404,
|
5413 |
+
"grad_norm": 0.0007662948337383568,
|
5414 |
+
"learning_rate": 9.230019679714405e-05,
|
5415 |
+
"loss": 46.0,
|
5416 |
+
"step": 769
|
5417 |
+
},
|
5418 |
+
{
|
5419 |
+
"epoch": 0.12994135763405476,
|
5420 |
+
"grad_norm": 0.0007776907877996564,
|
5421 |
+
"learning_rate": 9.208374890621628e-05,
|
5422 |
+
"loss": 46.0,
|
5423 |
+
"step": 770
|
5424 |
+
},
|
5425 |
+
{
|
5426 |
+
"epoch": 0.13011011264396913,
|
5427 |
+
"grad_norm": 0.0019411947578191757,
|
5428 |
+
"learning_rate": 9.186733833017061e-05,
|
5429 |
+
"loss": 46.0,
|
5430 |
+
"step": 771
|
5431 |
+
},
|
5432 |
+
{
|
5433 |
+
"epoch": 0.13027886765388347,
|
5434 |
+
"grad_norm": 0.0007572571048513055,
|
5435 |
+
"learning_rate": 9.165096608910287e-05,
|
5436 |
+
"loss": 46.0,
|
5437 |
+
"step": 772
|
5438 |
+
},
|
5439 |
+
{
|
5440 |
+
"epoch": 0.13044762266379784,
|
5441 |
+
"grad_norm": 0.0006218423368409276,
|
5442 |
+
"learning_rate": 9.143463320292832e-05,
|
5443 |
+
"loss": 46.0,
|
5444 |
+
"step": 773
|
5445 |
+
},
|
5446 |
+
{
|
5447 |
+
"epoch": 0.13061637767371218,
|
5448 |
+
"grad_norm": 0.0005072550848126411,
|
5449 |
+
"learning_rate": 9.12183406913766e-05,
|
5450 |
+
"loss": 46.0,
|
5451 |
+
"step": 774
|
5452 |
+
},
|
5453 |
+
{
|
5454 |
+
"epoch": 0.13078513268362654,
|
5455 |
+
"grad_norm": 0.0011237628059461713,
|
5456 |
+
"learning_rate": 9.100208957398703e-05,
|
5457 |
+
"loss": 46.0,
|
5458 |
+
"step": 775
|
5459 |
+
},
|
5460 |
+
{
|
5461 |
+
"epoch": 0.1309538876935409,
|
5462 |
+
"grad_norm": 0.0009162898059003055,
|
5463 |
+
"learning_rate": 9.078588087010389e-05,
|
5464 |
+
"loss": 46.0,
|
5465 |
+
"step": 776
|
5466 |
+
},
|
5467 |
+
{
|
5468 |
+
"epoch": 0.13112264270345525,
|
5469 |
+
"grad_norm": 0.003182788612321019,
|
5470 |
+
"learning_rate": 9.056971559887145e-05,
|
5471 |
+
"loss": 46.0,
|
5472 |
+
"step": 777
|
5473 |
+
},
|
5474 |
+
{
|
5475 |
+
"epoch": 0.13129139771336962,
|
5476 |
+
"grad_norm": 0.0018501750892028213,
|
5477 |
+
"learning_rate": 9.035359477922936e-05,
|
5478 |
+
"loss": 46.0,
|
5479 |
+
"step": 778
|
5480 |
+
},
|
5481 |
+
{
|
5482 |
+
"epoch": 0.13146015272328399,
|
5483 |
+
"grad_norm": 0.0009051130618900061,
|
5484 |
+
"learning_rate": 9.013751942990766e-05,
|
5485 |
+
"loss": 46.0,
|
5486 |
+
"step": 779
|
5487 |
+
},
|
5488 |
+
{
|
5489 |
+
"epoch": 0.13162890773319832,
|
5490 |
+
"grad_norm": 0.0003044608747586608,
|
5491 |
+
"learning_rate": 8.9921490569422e-05,
|
5492 |
+
"loss": 46.0,
|
5493 |
+
"step": 780
|
5494 |
+
},
|
5495 |
+
{
|
5496 |
+
"epoch": 0.1317976627431127,
|
5497 |
+
"grad_norm": 0.0008892896585166454,
|
5498 |
+
"learning_rate": 8.970550921606903e-05,
|
5499 |
+
"loss": 46.0,
|
5500 |
+
"step": 781
|
5501 |
+
},
|
5502 |
+
{
|
5503 |
+
"epoch": 0.13196641775302703,
|
5504 |
+
"grad_norm": 0.002685688668861985,
|
5505 |
+
"learning_rate": 8.948957638792144e-05,
|
5506 |
+
"loss": 46.0,
|
5507 |
+
"step": 782
|
5508 |
+
},
|
5509 |
+
{
|
5510 |
+
"epoch": 0.1321351727629414,
|
5511 |
+
"grad_norm": 0.0016494488809257746,
|
5512 |
+
"learning_rate": 8.927369310282302e-05,
|
5513 |
+
"loss": 46.0,
|
5514 |
+
"step": 783
|
5515 |
+
},
|
5516 |
+
{
|
5517 |
+
"epoch": 0.13230392777285577,
|
5518 |
+
"grad_norm": 0.0006050110096111894,
|
5519 |
+
"learning_rate": 8.905786037838426e-05,
|
5520 |
+
"loss": 46.0,
|
5521 |
+
"step": 784
|
5522 |
+
},
|
5523 |
+
{
|
5524 |
+
"epoch": 0.1324726827827701,
|
5525 |
+
"grad_norm": 0.0009582451893948019,
|
5526 |
+
"learning_rate": 8.884207923197715e-05,
|
5527 |
+
"loss": 46.0,
|
5528 |
+
"step": 785
|
5529 |
+
},
|
5530 |
+
{
|
5531 |
+
"epoch": 0.13264143779268447,
|
5532 |
+
"grad_norm": 0.004504833836108446,
|
5533 |
+
"learning_rate": 8.86263506807307e-05,
|
5534 |
+
"loss": 46.0,
|
5535 |
+
"step": 786
|
5536 |
+
},
|
5537 |
+
{
|
5538 |
+
"epoch": 0.13281019280259881,
|
5539 |
+
"grad_norm": 0.0005395681946538389,
|
5540 |
+
"learning_rate": 8.841067574152589e-05,
|
5541 |
+
"loss": 46.0,
|
5542 |
+
"step": 787
|
5543 |
+
},
|
5544 |
+
{
|
5545 |
+
"epoch": 0.13297894781251318,
|
5546 |
+
"grad_norm": 0.001102139474824071,
|
5547 |
+
"learning_rate": 8.819505543099099e-05,
|
5548 |
+
"loss": 46.0,
|
5549 |
+
"step": 788
|
5550 |
+
},
|
5551 |
+
{
|
5552 |
+
"epoch": 0.13314770282242755,
|
5553 |
+
"grad_norm": 0.000723580364137888,
|
5554 |
+
"learning_rate": 8.797949076549685e-05,
|
5555 |
+
"loss": 46.0,
|
5556 |
+
"step": 789
|
5557 |
+
},
|
5558 |
+
{
|
5559 |
+
"epoch": 0.1333164578323419,
|
5560 |
+
"grad_norm": 0.002134741051122546,
|
5561 |
+
"learning_rate": 8.776398276115198e-05,
|
5562 |
+
"loss": 46.0,
|
5563 |
+
"step": 790
|
5564 |
+
},
|
5565 |
+
{
|
5566 |
+
"epoch": 0.13348521284225626,
|
5567 |
+
"grad_norm": 0.0018205747473984957,
|
5568 |
+
"learning_rate": 8.75485324337977e-05,
|
5569 |
+
"loss": 46.0,
|
5570 |
+
"step": 791
|
5571 |
+
},
|
5572 |
+
{
|
5573 |
+
"epoch": 0.13365396785217062,
|
5574 |
+
"grad_norm": 0.0017829221906140447,
|
5575 |
+
"learning_rate": 8.733314079900372e-05,
|
5576 |
+
"loss": 46.0,
|
5577 |
+
"step": 792
|
5578 |
+
},
|
5579 |
+
{
|
5580 |
+
"epoch": 0.13382272286208496,
|
5581 |
+
"grad_norm": 0.000708136591129005,
|
5582 |
+
"learning_rate": 8.71178088720628e-05,
|
5583 |
+
"loss": 46.0,
|
5584 |
+
"step": 793
|
5585 |
+
},
|
5586 |
+
{
|
5587 |
+
"epoch": 0.13399147787199933,
|
5588 |
+
"grad_norm": 0.0009631026186980307,
|
5589 |
+
"learning_rate": 8.690253766798644e-05,
|
5590 |
+
"loss": 46.0,
|
5591 |
+
"step": 794
|
5592 |
+
},
|
5593 |
+
{
|
5594 |
+
"epoch": 0.13416023288191367,
|
5595 |
+
"grad_norm": 0.0010598397348076105,
|
5596 |
+
"learning_rate": 8.668732820149989e-05,
|
5597 |
+
"loss": 46.0,
|
5598 |
+
"step": 795
|
5599 |
+
},
|
5600 |
+
{
|
5601 |
+
"epoch": 0.13432898789182804,
|
5602 |
+
"grad_norm": 0.0007573203183710575,
|
5603 |
+
"learning_rate": 8.647218148703727e-05,
|
5604 |
+
"loss": 46.0,
|
5605 |
+
"step": 796
|
5606 |
+
},
|
5607 |
+
{
|
5608 |
+
"epoch": 0.1344977429017424,
|
5609 |
+
"grad_norm": 0.0017247875221073627,
|
5610 |
+
"learning_rate": 8.625709853873705e-05,
|
5611 |
+
"loss": 46.0,
|
5612 |
+
"step": 797
|
5613 |
+
},
|
5614 |
+
{
|
5615 |
+
"epoch": 0.13466649791165675,
|
5616 |
+
"grad_norm": 0.000708206498529762,
|
5617 |
+
"learning_rate": 8.6042080370437e-05,
|
5618 |
+
"loss": 46.0,
|
5619 |
+
"step": 798
|
5620 |
+
},
|
5621 |
+
{
|
5622 |
+
"epoch": 0.1348352529215711,
|
5623 |
+
"grad_norm": 0.0007804720080457628,
|
5624 |
+
"learning_rate": 8.582712799566969e-05,
|
5625 |
+
"loss": 46.0,
|
5626 |
+
"step": 799
|
5627 |
+
},
|
5628 |
+
{
|
5629 |
+
"epoch": 0.13500400793148545,
|
5630 |
+
"grad_norm": 0.002211306942626834,
|
5631 |
+
"learning_rate": 8.561224242765744e-05,
|
5632 |
+
"loss": 46.0,
|
5633 |
+
"step": 800
|
5634 |
+
},
|
5635 |
+
{
|
5636 |
+
"epoch": 0.13517276294139982,
|
5637 |
+
"grad_norm": 0.0037598570343106985,
|
5638 |
+
"learning_rate": 8.539742467930761e-05,
|
5639 |
+
"loss": 46.0,
|
5640 |
+
"step": 801
|
5641 |
+
},
|
5642 |
+
{
|
5643 |
+
"epoch": 0.1353415179513142,
|
5644 |
+
"grad_norm": 0.0007354649715125561,
|
5645 |
+
"learning_rate": 8.518267576320806e-05,
|
5646 |
+
"loss": 46.0,
|
5647 |
+
"step": 802
|
5648 |
+
},
|
5649 |
+
{
|
5650 |
+
"epoch": 0.13551027296122853,
|
5651 |
+
"grad_norm": 0.00044775562128052115,
|
5652 |
+
"learning_rate": 8.496799669162206e-05,
|
5653 |
+
"loss": 46.0,
|
5654 |
+
"step": 803
|
5655 |
+
},
|
5656 |
+
{
|
5657 |
+
"epoch": 0.1356790279711429,
|
5658 |
+
"grad_norm": 0.0005587537889368832,
|
5659 |
+
"learning_rate": 8.475338847648361e-05,
|
5660 |
+
"loss": 46.0,
|
5661 |
+
"step": 804
|
5662 |
+
},
|
5663 |
+
{
|
5664 |
+
"epoch": 0.13584778298105726,
|
5665 |
+
"grad_norm": 0.0005210166564211249,
|
5666 |
+
"learning_rate": 8.453885212939294e-05,
|
5667 |
+
"loss": 46.0,
|
5668 |
+
"step": 805
|
5669 |
+
},
|
5670 |
+
{
|
5671 |
+
"epoch": 0.1360165379909716,
|
5672 |
+
"grad_norm": 0.0007426153169944882,
|
5673 |
+
"learning_rate": 8.432438866161124e-05,
|
5674 |
+
"loss": 46.0,
|
5675 |
+
"step": 806
|
5676 |
+
},
|
5677 |
+
{
|
5678 |
+
"epoch": 0.13618529300088597,
|
5679 |
+
"grad_norm": 0.0013306104810908437,
|
5680 |
+
"learning_rate": 8.410999908405635e-05,
|
5681 |
+
"loss": 46.0,
|
5682 |
+
"step": 807
|
5683 |
+
},
|
5684 |
+
{
|
5685 |
+
"epoch": 0.1363540480108003,
|
5686 |
+
"grad_norm": 0.0013473851140588522,
|
5687 |
+
"learning_rate": 8.389568440729776e-05,
|
5688 |
+
"loss": 46.0,
|
5689 |
+
"step": 808
|
5690 |
+
},
|
5691 |
+
{
|
5692 |
+
"epoch": 0.13652280302071468,
|
5693 |
+
"grad_norm": 0.00044588756281882524,
|
5694 |
+
"learning_rate": 8.368144564155185e-05,
|
5695 |
+
"loss": 46.0,
|
5696 |
+
"step": 809
|
5697 |
+
},
|
5698 |
+
{
|
5699 |
+
"epoch": 0.13669155803062905,
|
5700 |
+
"grad_norm": 0.0006963923806324601,
|
5701 |
+
"learning_rate": 8.346728379667727e-05,
|
5702 |
+
"loss": 46.0,
|
5703 |
+
"step": 810
|
5704 |
+
},
|
5705 |
+
{
|
5706 |
+
"epoch": 0.13686031304054339,
|
5707 |
+
"grad_norm": 0.001505714375525713,
|
5708 |
+
"learning_rate": 8.325319988217004e-05,
|
5709 |
+
"loss": 46.0,
|
5710 |
+
"step": 811
|
5711 |
+
},
|
5712 |
+
{
|
5713 |
+
"epoch": 0.13702906805045775,
|
5714 |
+
"grad_norm": 0.0009234807803295553,
|
5715 |
+
"learning_rate": 8.303919490715881e-05,
|
5716 |
+
"loss": 46.0,
|
5717 |
+
"step": 812
|
5718 |
+
},
|
5719 |
+
{
|
5720 |
+
"epoch": 0.1371978230603721,
|
5721 |
+
"grad_norm": 0.001101883128285408,
|
5722 |
+
"learning_rate": 8.282526988040022e-05,
|
5723 |
+
"loss": 46.0,
|
5724 |
+
"step": 813
|
5725 |
+
},
|
5726 |
+
{
|
5727 |
+
"epoch": 0.13736657807028646,
|
5728 |
+
"grad_norm": 0.00047088676365092397,
|
5729 |
+
"learning_rate": 8.261142581027392e-05,
|
5730 |
+
"loss": 46.0,
|
5731 |
+
"step": 814
|
5732 |
+
},
|
5733 |
+
{
|
5734 |
+
"epoch": 0.13753533308020083,
|
5735 |
+
"grad_norm": 0.0010181186953559518,
|
5736 |
+
"learning_rate": 8.239766370477811e-05,
|
5737 |
+
"loss": 46.0,
|
5738 |
+
"step": 815
|
5739 |
+
},
|
5740 |
+
{
|
5741 |
+
"epoch": 0.13770408809011517,
|
5742 |
+
"grad_norm": 0.0008076895028352737,
|
5743 |
+
"learning_rate": 8.218398457152454e-05,
|
5744 |
+
"loss": 46.0,
|
5745 |
+
"step": 816
|
5746 |
+
},
|
5747 |
+
{
|
5748 |
+
"epoch": 0.13787284310002954,
|
5749 |
+
"grad_norm": 0.000841393368318677,
|
5750 |
+
"learning_rate": 8.197038941773381e-05,
|
5751 |
+
"loss": 46.0,
|
5752 |
+
"step": 817
|
5753 |
+
},
|
5754 |
+
{
|
5755 |
+
"epoch": 0.1380415981099439,
|
5756 |
+
"grad_norm": 0.0008007394499145448,
|
5757 |
+
"learning_rate": 8.175687925023082e-05,
|
5758 |
+
"loss": 46.0,
|
5759 |
+
"step": 818
|
5760 |
+
},
|
5761 |
+
{
|
5762 |
+
"epoch": 0.13821035311985824,
|
5763 |
+
"grad_norm": 0.000764928525313735,
|
5764 |
+
"learning_rate": 8.154345507543972e-05,
|
5765 |
+
"loss": 46.0,
|
5766 |
+
"step": 819
|
5767 |
+
},
|
5768 |
+
{
|
5769 |
+
"epoch": 0.1383791081297726,
|
5770 |
+
"grad_norm": 0.0015394032234326005,
|
5771 |
+
"learning_rate": 8.133011789937937e-05,
|
5772 |
+
"loss": 46.0,
|
5773 |
+
"step": 820
|
5774 |
+
},
|
5775 |
+
{
|
5776 |
+
"epoch": 0.13854786313968695,
|
5777 |
+
"grad_norm": 0.0008911428158171475,
|
5778 |
+
"learning_rate": 8.111686872765859e-05,
|
5779 |
+
"loss": 46.0,
|
5780 |
+
"step": 821
|
5781 |
+
},
|
5782 |
+
{
|
5783 |
+
"epoch": 0.13871661814960132,
|
5784 |
+
"grad_norm": 0.0006434863316826522,
|
5785 |
+
"learning_rate": 8.090370856547127e-05,
|
5786 |
+
"loss": 46.0,
|
5787 |
+
"step": 822
|
5788 |
+
},
|
5789 |
+
{
|
5790 |
+
"epoch": 0.13888537315951568,
|
5791 |
+
"grad_norm": 0.0009041949524544179,
|
5792 |
+
"learning_rate": 8.069063841759182e-05,
|
5793 |
+
"loss": 46.0,
|
5794 |
+
"step": 823
|
5795 |
+
},
|
5796 |
+
{
|
5797 |
+
"epoch": 0.13905412816943002,
|
5798 |
+
"grad_norm": 0.0003908716607838869,
|
5799 |
+
"learning_rate": 8.047765928837034e-05,
|
5800 |
+
"loss": 46.0,
|
5801 |
+
"step": 824
|
5802 |
+
},
|
5803 |
+
{
|
5804 |
+
"epoch": 0.1392228831793444,
|
5805 |
+
"grad_norm": 0.0004880438500549644,
|
5806 |
+
"learning_rate": 8.026477218172785e-05,
|
5807 |
+
"loss": 46.0,
|
5808 |
+
"step": 825
|
5809 |
+
},
|
5810 |
+
{
|
5811 |
+
"epoch": 0.13939163818925873,
|
5812 |
+
"grad_norm": 0.0008036144427023828,
|
5813 |
+
"learning_rate": 8.005197810115168e-05,
|
5814 |
+
"loss": 46.0,
|
5815 |
+
"step": 826
|
5816 |
+
},
|
5817 |
+
{
|
5818 |
+
"epoch": 0.1395603931991731,
|
5819 |
+
"grad_norm": 0.00038311423850245774,
|
5820 |
+
"learning_rate": 7.983927804969054e-05,
|
5821 |
+
"loss": 46.0,
|
5822 |
+
"step": 827
|
5823 |
+
},
|
5824 |
+
{
|
5825 |
+
"epoch": 0.13972914820908747,
|
5826 |
+
"grad_norm": 0.0007264793966896832,
|
5827 |
+
"learning_rate": 7.962667302995004e-05,
|
5828 |
+
"loss": 46.0,
|
5829 |
+
"step": 828
|
5830 |
+
},
|
5831 |
+
{
|
5832 |
+
"epoch": 0.1398979032190018,
|
5833 |
+
"grad_norm": 0.006564402021467686,
|
5834 |
+
"learning_rate": 7.941416404408778e-05,
|
5835 |
+
"loss": 46.0,
|
5836 |
+
"step": 829
|
5837 |
+
},
|
5838 |
+
{
|
5839 |
+
"epoch": 0.14006665822891617,
|
5840 |
+
"grad_norm": 0.0009483057074248791,
|
5841 |
+
"learning_rate": 7.920175209380865e-05,
|
5842 |
+
"loss": 46.0,
|
5843 |
+
"step": 830
|
5844 |
+
},
|
5845 |
+
{
|
5846 |
+
"epoch": 0.14023541323883051,
|
5847 |
+
"grad_norm": 0.001220498699694872,
|
5848 |
+
"learning_rate": 7.898943818036024e-05,
|
5849 |
+
"loss": 46.0,
|
5850 |
+
"step": 831
|
5851 |
+
},
|
5852 |
+
{
|
5853 |
+
"epoch": 0.14040416824874488,
|
5854 |
+
"grad_norm": 0.0012295391643419862,
|
5855 |
+
"learning_rate": 7.877722330452795e-05,
|
5856 |
+
"loss": 46.0,
|
5857 |
+
"step": 832
|
5858 |
+
},
|
5859 |
+
{
|
5860 |
+
"epoch": 0.14057292325865925,
|
5861 |
+
"grad_norm": 0.0008359167259186506,
|
5862 |
+
"learning_rate": 7.856510846663031e-05,
|
5863 |
+
"loss": 46.0,
|
5864 |
+
"step": 833
|
5865 |
+
},
|
5866 |
+
{
|
5867 |
+
"epoch": 0.1407416782685736,
|
5868 |
+
"grad_norm": 0.001273457077331841,
|
5869 |
+
"learning_rate": 7.835309466651444e-05,
|
5870 |
+
"loss": 46.0,
|
5871 |
+
"step": 834
|
5872 |
+
},
|
5873 |
+
{
|
5874 |
+
"epoch": 0.14091043327848796,
|
5875 |
+
"grad_norm": 0.0027280368376523256,
|
5876 |
+
"learning_rate": 7.814118290355102e-05,
|
5877 |
+
"loss": 46.0,
|
5878 |
+
"step": 835
|
5879 |
+
},
|
5880 |
+
{
|
5881 |
+
"epoch": 0.14107918828840232,
|
5882 |
+
"grad_norm": 0.0015108464285731316,
|
5883 |
+
"learning_rate": 7.792937417662993e-05,
|
5884 |
+
"loss": 46.0,
|
5885 |
+
"step": 836
|
5886 |
+
},
|
5887 |
+
{
|
5888 |
+
"epoch": 0.14124794329831666,
|
5889 |
+
"grad_norm": 0.002349273534491658,
|
5890 |
+
"learning_rate": 7.771766948415525e-05,
|
5891 |
+
"loss": 46.0,
|
5892 |
+
"step": 837
|
5893 |
+
},
|
5894 |
+
{
|
5895 |
+
"epoch": 0.14141669830823103,
|
5896 |
+
"grad_norm": 0.0012695115292444825,
|
5897 |
+
"learning_rate": 7.750606982404065e-05,
|
5898 |
+
"loss": 46.0,
|
5899 |
+
"step": 838
|
5900 |
+
},
|
5901 |
+
{
|
5902 |
+
"epoch": 0.14158545331814537,
|
5903 |
+
"grad_norm": 0.0009470002260059118,
|
5904 |
+
"learning_rate": 7.729457619370489e-05,
|
5905 |
+
"loss": 46.0,
|
5906 |
+
"step": 839
|
5907 |
+
},
|
5908 |
+
{
|
5909 |
+
"epoch": 0.14175420832805974,
|
5910 |
+
"grad_norm": 0.0006858884589746594,
|
5911 |
+
"learning_rate": 7.708318959006669e-05,
|
5912 |
+
"loss": 46.0,
|
5913 |
+
"step": 840
|
5914 |
+
},
|
5915 |
+
{
|
5916 |
+
"epoch": 0.1419229633379741,
|
5917 |
+
"grad_norm": 0.0016968476120382547,
|
5918 |
+
"learning_rate": 7.687191100954046e-05,
|
5919 |
+
"loss": 46.0,
|
5920 |
+
"step": 841
|
5921 |
+
},
|
5922 |
+
{
|
5923 |
+
"epoch": 0.14209171834788845,
|
5924 |
+
"grad_norm": 0.002445329912006855,
|
5925 |
+
"learning_rate": 7.666074144803139e-05,
|
5926 |
+
"loss": 46.0,
|
5927 |
+
"step": 842
|
5928 |
+
},
|
5929 |
+
{
|
5930 |
+
"epoch": 0.1422604733578028,
|
5931 |
+
"grad_norm": 0.0007218089303933084,
|
5932 |
+
"learning_rate": 7.644968190093067e-05,
|
5933 |
+
"loss": 46.0,
|
5934 |
+
"step": 843
|
5935 |
+
},
|
5936 |
+
{
|
5937 |
+
"epoch": 0.14242922836771715,
|
5938 |
+
"grad_norm": 0.0014190871734172106,
|
5939 |
+
"learning_rate": 7.623873336311108e-05,
|
5940 |
+
"loss": 46.0,
|
5941 |
+
"step": 844
|
5942 |
+
},
|
5943 |
+
{
|
5944 |
+
"epoch": 0.14259798337763152,
|
5945 |
+
"grad_norm": 0.0006293446640484035,
|
5946 |
+
"learning_rate": 7.602789682892203e-05,
|
5947 |
+
"loss": 46.0,
|
5948 |
+
"step": 845
|
5949 |
+
},
|
5950 |
+
{
|
5951 |
+
"epoch": 0.1427667383875459,
|
5952 |
+
"grad_norm": 0.0012009006459265947,
|
5953 |
+
"learning_rate": 7.581717329218499e-05,
|
5954 |
+
"loss": 46.0,
|
5955 |
+
"step": 846
|
5956 |
+
},
|
5957 |
+
{
|
5958 |
+
"epoch": 0.14293549339746023,
|
5959 |
+
"grad_norm": 0.0011125734308734536,
|
5960 |
+
"learning_rate": 7.560656374618886e-05,
|
5961 |
+
"loss": 46.0,
|
5962 |
+
"step": 847
|
5963 |
+
},
|
5964 |
+
{
|
5965 |
+
"epoch": 0.1431042484073746,
|
5966 |
+
"grad_norm": 0.002301817527040839,
|
5967 |
+
"learning_rate": 7.539606918368507e-05,
|
5968 |
+
"loss": 46.0,
|
5969 |
+
"step": 848
|
5970 |
+
},
|
5971 |
+
{
|
5972 |
+
"epoch": 0.14327300341728896,
|
5973 |
+
"grad_norm": 0.0006231066072359681,
|
5974 |
+
"learning_rate": 7.518569059688325e-05,
|
5975 |
+
"loss": 46.0,
|
5976 |
+
"step": 849
|
5977 |
+
},
|
5978 |
+
{
|
5979 |
+
"epoch": 0.1434417584272033,
|
5980 |
+
"grad_norm": 0.002790312049910426,
|
5981 |
+
"learning_rate": 7.497542897744622e-05,
|
5982 |
+
"loss": 46.0,
|
5983 |
+
"step": 850
|
5984 |
+
},
|
5985 |
+
{
|
5986 |
+
"epoch": 0.14361051343711767,
|
5987 |
+
"grad_norm": 0.0007836490985937417,
|
5988 |
+
"learning_rate": 7.476528531648544e-05,
|
5989 |
+
"loss": 46.0,
|
5990 |
+
"step": 851
|
5991 |
+
},
|
5992 |
+
{
|
5993 |
+
"epoch": 0.143779268447032,
|
5994 |
+
"grad_norm": 0.0008998861303552985,
|
5995 |
+
"learning_rate": 7.455526060455648e-05,
|
5996 |
+
"loss": 46.0,
|
5997 |
+
"step": 852
|
5998 |
+
},
|
5999 |
+
{
|
6000 |
+
"epoch": 0.14394802345694638,
|
6001 |
+
"grad_norm": 0.0022323448210954666,
|
6002 |
+
"learning_rate": 7.434535583165408e-05,
|
6003 |
+
"loss": 46.0,
|
6004 |
+
"step": 853
|
6005 |
+
},
|
6006 |
+
{
|
6007 |
+
"epoch": 0.14411677846686075,
|
6008 |
+
"grad_norm": 0.0007675597444176674,
|
6009 |
+
"learning_rate": 7.413557198720765e-05,
|
6010 |
+
"loss": 46.0,
|
6011 |
+
"step": 854
|
6012 |
+
},
|
6013 |
+
{
|
6014 |
+
"epoch": 0.14428553347677509,
|
6015 |
+
"grad_norm": 0.0015968787483870983,
|
6016 |
+
"learning_rate": 7.392591006007666e-05,
|
6017 |
+
"loss": 46.0,
|
6018 |
+
"step": 855
|
6019 |
+
},
|
6020 |
+
{
|
6021 |
+
"epoch": 0.14445428848668945,
|
6022 |
+
"grad_norm": 0.0005020895623601973,
|
6023 |
+
"learning_rate": 7.37163710385458e-05,
|
6024 |
+
"loss": 46.0,
|
6025 |
+
"step": 856
|
6026 |
+
},
|
6027 |
+
{
|
6028 |
+
"epoch": 0.1446230434966038,
|
6029 |
+
"grad_norm": 0.0010543781099840999,
|
6030 |
+
"learning_rate": 7.350695591032049e-05,
|
6031 |
+
"loss": 46.0,
|
6032 |
+
"step": 857
|
6033 |
+
},
|
6034 |
+
{
|
6035 |
+
"epoch": 0.14479179850651816,
|
6036 |
+
"grad_norm": 0.000769981590565294,
|
6037 |
+
"learning_rate": 7.329766566252212e-05,
|
6038 |
+
"loss": 46.0,
|
6039 |
+
"step": 858
|
6040 |
+
},
|
6041 |
+
{
|
6042 |
+
"epoch": 0.14496055351643253,
|
6043 |
+
"grad_norm": 0.0009824162116274238,
|
6044 |
+
"learning_rate": 7.30885012816834e-05,
|
6045 |
+
"loss": 46.0,
|
6046 |
+
"step": 859
|
6047 |
+
},
|
6048 |
+
{
|
6049 |
+
"epoch": 0.14512930852634687,
|
6050 |
+
"grad_norm": 0.0020662923343479633,
|
6051 |
+
"learning_rate": 7.287946375374385e-05,
|
6052 |
+
"loss": 46.0,
|
6053 |
+
"step": 860
|
6054 |
+
},
|
6055 |
+
{
|
6056 |
+
"epoch": 0.14529806353626123,
|
6057 |
+
"grad_norm": 0.002364357467740774,
|
6058 |
+
"learning_rate": 7.26705540640449e-05,
|
6059 |
+
"loss": 46.0,
|
6060 |
+
"step": 861
|
6061 |
+
},
|
6062 |
+
{
|
6063 |
+
"epoch": 0.1454668185461756,
|
6064 |
+
"grad_norm": 0.001292189583182335,
|
6065 |
+
"learning_rate": 7.246177319732543e-05,
|
6066 |
+
"loss": 46.0,
|
6067 |
+
"step": 862
|
6068 |
+
},
|
6069 |
+
{
|
6070 |
+
"epoch": 0.14563557355608994,
|
6071 |
+
"grad_norm": 0.000383058562874794,
|
6072 |
+
"learning_rate": 7.225312213771722e-05,
|
6073 |
+
"loss": 46.0,
|
6074 |
+
"step": 863
|
6075 |
+
},
|
6076 |
+
{
|
6077 |
+
"epoch": 0.1458043285660043,
|
6078 |
+
"grad_norm": 0.0009881592122837901,
|
6079 |
+
"learning_rate": 7.204460186873995e-05,
|
6080 |
+
"loss": 46.0,
|
6081 |
+
"step": 864
|
6082 |
+
},
|
6083 |
+
{
|
6084 |
+
"epoch": 0.14597308357591865,
|
6085 |
+
"grad_norm": 0.0010527949780225754,
|
6086 |
+
"learning_rate": 7.183621337329703e-05,
|
6087 |
+
"loss": 46.0,
|
6088 |
+
"step": 865
|
6089 |
+
},
|
6090 |
+
{
|
6091 |
+
"epoch": 0.14614183858583302,
|
6092 |
+
"grad_norm": 0.001268843188881874,
|
6093 |
+
"learning_rate": 7.162795763367049e-05,
|
6094 |
+
"loss": 46.0,
|
6095 |
+
"step": 866
|
6096 |
+
},
|
6097 |
+
{
|
6098 |
+
"epoch": 0.14631059359574738,
|
6099 |
+
"grad_norm": 0.003271568100899458,
|
6100 |
+
"learning_rate": 7.141983563151677e-05,
|
6101 |
+
"loss": 46.0,
|
6102 |
+
"step": 867
|
6103 |
+
},
|
6104 |
+
{
|
6105 |
+
"epoch": 0.14647934860566172,
|
6106 |
+
"grad_norm": 0.0008496344089508057,
|
6107 |
+
"learning_rate": 7.121184834786188e-05,
|
6108 |
+
"loss": 46.0,
|
6109 |
+
"step": 868
|
6110 |
+
},
|
6111 |
+
{
|
6112 |
+
"epoch": 0.1466481036155761,
|
6113 |
+
"grad_norm": 0.001987769966945052,
|
6114 |
+
"learning_rate": 7.10039967630967e-05,
|
6115 |
+
"loss": 46.0,
|
6116 |
+
"step": 869
|
6117 |
+
},
|
6118 |
+
{
|
6119 |
+
"epoch": 0.14681685862549043,
|
6120 |
+
"grad_norm": 0.0011683752527460456,
|
6121 |
+
"learning_rate": 7.079628185697258e-05,
|
6122 |
+
"loss": 46.0,
|
6123 |
+
"step": 870
|
6124 |
+
},
|
6125 |
+
{
|
6126 |
+
"epoch": 0.1469856136354048,
|
6127 |
+
"grad_norm": 0.000853047997225076,
|
6128 |
+
"learning_rate": 7.058870460859656e-05,
|
6129 |
+
"loss": 46.0,
|
6130 |
+
"step": 871
|
6131 |
+
},
|
6132 |
+
{
|
6133 |
+
"epoch": 0.14715436864531917,
|
6134 |
+
"grad_norm": 0.001560655073262751,
|
6135 |
+
"learning_rate": 7.03812659964268e-05,
|
6136 |
+
"loss": 46.0,
|
6137 |
+
"step": 872
|
6138 |
+
},
|
6139 |
+
{
|
6140 |
+
"epoch": 0.1473231236552335,
|
6141 |
+
"grad_norm": 0.0007004750659689307,
|
6142 |
+
"learning_rate": 7.017396699826803e-05,
|
6143 |
+
"loss": 46.0,
|
6144 |
+
"step": 873
|
6145 |
+
},
|
6146 |
+
{
|
6147 |
+
"epoch": 0.14749187866514787,
|
6148 |
+
"grad_norm": 0.0006875868421047926,
|
6149 |
+
"learning_rate": 6.996680859126677e-05,
|
6150 |
+
"loss": 46.0,
|
6151 |
+
"step": 874
|
6152 |
+
},
|
6153 |
+
{
|
6154 |
+
"epoch": 0.14766063367506224,
|
6155 |
+
"grad_norm": 0.0009068532381206751,
|
6156 |
+
"learning_rate": 6.975979175190688e-05,
|
6157 |
+
"loss": 46.0,
|
6158 |
+
"step": 875
|
6159 |
+
},
|
6160 |
+
{
|
6161 |
+
"epoch": 0.14782938868497658,
|
6162 |
+
"grad_norm": 0.0005159789579920471,
|
6163 |
+
"learning_rate": 6.955291745600498e-05,
|
6164 |
+
"loss": 46.0,
|
6165 |
+
"step": 876
|
6166 |
+
},
|
6167 |
+
{
|
6168 |
+
"epoch": 0.14799814369489095,
|
6169 |
+
"grad_norm": 0.0014988022157922387,
|
6170 |
+
"learning_rate": 6.934618667870567e-05,
|
6171 |
+
"loss": 46.0,
|
6172 |
+
"step": 877
|
6173 |
+
},
|
6174 |
+
{
|
6175 |
+
"epoch": 0.1481668987048053,
|
6176 |
+
"grad_norm": 0.0007760879816487432,
|
6177 |
+
"learning_rate": 6.913960039447718e-05,
|
6178 |
+
"loss": 46.0,
|
6179 |
+
"step": 878
|
6180 |
+
},
|
6181 |
+
{
|
6182 |
+
"epoch": 0.14833565371471966,
|
6183 |
+
"grad_norm": 0.0010894398437812924,
|
6184 |
+
"learning_rate": 6.893315957710649e-05,
|
6185 |
+
"loss": 46.0,
|
6186 |
+
"step": 879
|
6187 |
+
},
|
6188 |
+
{
|
6189 |
+
"epoch": 0.14850440872463402,
|
6190 |
+
"grad_norm": 0.0004840297333430499,
|
6191 |
+
"learning_rate": 6.8726865199695e-05,
|
6192 |
+
"loss": 46.0,
|
6193 |
+
"step": 880
|
6194 |
+
},
|
6195 |
+
{
|
6196 |
+
"epoch": 0.14867316373454836,
|
6197 |
+
"grad_norm": 0.0005790110444650054,
|
6198 |
+
"learning_rate": 6.852071823465383e-05,
|
6199 |
+
"loss": 46.0,
|
6200 |
+
"step": 881
|
6201 |
+
},
|
6202 |
+
{
|
6203 |
+
"epoch": 0.14884191874446273,
|
6204 |
+
"grad_norm": 0.002223587827757001,
|
6205 |
+
"learning_rate": 6.831471965369914e-05,
|
6206 |
+
"loss": 46.0,
|
6207 |
+
"step": 882
|
6208 |
+
},
|
6209 |
+
{
|
6210 |
+
"epoch": 0.14901067375437707,
|
6211 |
+
"grad_norm": 0.002203379524871707,
|
6212 |
+
"learning_rate": 6.810887042784777e-05,
|
6213 |
+
"loss": 46.0,
|
6214 |
+
"step": 883
|
6215 |
+
},
|
6216 |
+
{
|
6217 |
+
"epoch": 0.14917942876429144,
|
6218 |
+
"grad_norm": 0.002671119524165988,
|
6219 |
+
"learning_rate": 6.790317152741249e-05,
|
6220 |
+
"loss": 46.0,
|
6221 |
+
"step": 884
|
6222 |
+
},
|
6223 |
+
{
|
6224 |
+
"epoch": 0.1493481837742058,
|
6225 |
+
"grad_norm": 0.001122485613450408,
|
6226 |
+
"learning_rate": 6.769762392199748e-05,
|
6227 |
+
"loss": 46.0,
|
6228 |
+
"step": 885
|
6229 |
+
},
|
6230 |
+
{
|
6231 |
+
"epoch": 0.14951693878412015,
|
6232 |
+
"grad_norm": 0.0007207071175798774,
|
6233 |
+
"learning_rate": 6.749222858049382e-05,
|
6234 |
+
"loss": 46.0,
|
6235 |
+
"step": 886
|
6236 |
+
},
|
6237 |
+
{
|
6238 |
+
"epoch": 0.1496856937940345,
|
6239 |
+
"grad_norm": 0.0007145823910832405,
|
6240 |
+
"learning_rate": 6.728698647107475e-05,
|
6241 |
+
"loss": 46.0,
|
6242 |
+
"step": 887
|
6243 |
+
},
|
6244 |
+
{
|
6245 |
+
"epoch": 0.14985444880394888,
|
6246 |
+
"grad_norm": 0.0008989453199319541,
|
6247 |
+
"learning_rate": 6.708189856119128e-05,
|
6248 |
+
"loss": 46.0,
|
6249 |
+
"step": 888
|
6250 |
+
},
|
6251 |
+
{
|
6252 |
+
"epoch": 0.15002320381386322,
|
6253 |
+
"grad_norm": 0.0011213996913284063,
|
6254 |
+
"learning_rate": 6.687696581756763e-05,
|
6255 |
+
"loss": 46.0,
|
6256 |
+
"step": 889
|
6257 |
+
},
|
6258 |
+
{
|
6259 |
+
"epoch": 0.1501919588237776,
|
6260 |
+
"grad_norm": 0.0005325007368810475,
|
6261 |
+
"learning_rate": 6.667218920619649e-05,
|
6262 |
+
"loss": 46.0,
|
6263 |
+
"step": 890
|
6264 |
+
},
|
6265 |
+
{
|
6266 |
+
"epoch": 0.15036071383369193,
|
6267 |
+
"grad_norm": 0.0008855744963511825,
|
6268 |
+
"learning_rate": 6.646756969233463e-05,
|
6269 |
+
"loss": 46.0,
|
6270 |
+
"step": 891
|
6271 |
+
},
|
6272 |
+
{
|
6273 |
+
"epoch": 0.1505294688436063,
|
6274 |
+
"grad_norm": 0.0007719965651631355,
|
6275 |
+
"learning_rate": 6.626310824049838e-05,
|
6276 |
+
"loss": 46.0,
|
6277 |
+
"step": 892
|
6278 |
+
},
|
6279 |
+
{
|
6280 |
+
"epoch": 0.15069822385352066,
|
6281 |
+
"grad_norm": 0.001363624120131135,
|
6282 |
+
"learning_rate": 6.605880581445894e-05,
|
6283 |
+
"loss": 46.0,
|
6284 |
+
"step": 893
|
6285 |
+
},
|
6286 |
+
{
|
6287 |
+
"epoch": 0.150866978863435,
|
6288 |
+
"grad_norm": 0.0006050238152965903,
|
6289 |
+
"learning_rate": 6.585466337723796e-05,
|
6290 |
+
"loss": 46.0,
|
6291 |
+
"step": 894
|
6292 |
+
},
|
6293 |
+
{
|
6294 |
+
"epoch": 0.15103573387334937,
|
6295 |
+
"grad_norm": 0.00020488305017352104,
|
6296 |
+
"learning_rate": 6.565068189110286e-05,
|
6297 |
+
"loss": 46.0,
|
6298 |
+
"step": 895
|
6299 |
+
},
|
6300 |
+
{
|
6301 |
+
"epoch": 0.1512044888832637,
|
6302 |
+
"grad_norm": 0.000528375618159771,
|
6303 |
+
"learning_rate": 6.544686231756246e-05,
|
6304 |
+
"loss": 46.0,
|
6305 |
+
"step": 896
|
6306 |
+
},
|
6307 |
+
{
|
6308 |
+
"epoch": 0.15137324389317808,
|
6309 |
+
"grad_norm": 0.0006251371232792735,
|
6310 |
+
"learning_rate": 6.52432056173624e-05,
|
6311 |
+
"loss": 46.0,
|
6312 |
+
"step": 897
|
6313 |
+
},
|
6314 |
+
{
|
6315 |
+
"epoch": 0.15154199890309245,
|
6316 |
+
"grad_norm": 0.000713423069100827,
|
6317 |
+
"learning_rate": 6.503971275048042e-05,
|
6318 |
+
"loss": 46.0,
|
6319 |
+
"step": 898
|
6320 |
+
},
|
6321 |
+
{
|
6322 |
+
"epoch": 0.15171075391300679,
|
6323 |
+
"grad_norm": 0.0008193039102479815,
|
6324 |
+
"learning_rate": 6.483638467612227e-05,
|
6325 |
+
"loss": 46.0,
|
6326 |
+
"step": 899
|
6327 |
+
},
|
6328 |
+
{
|
6329 |
+
"epoch": 0.15187950892292115,
|
6330 |
+
"grad_norm": 0.000522931688465178,
|
6331 |
+
"learning_rate": 6.463322235271666e-05,
|
6332 |
+
"loss": 46.0,
|
6333 |
+
"step": 900
|
6334 |
+
},
|
6335 |
+
{
|
6336 |
+
"epoch": 0.1520482639328355,
|
6337 |
+
"grad_norm": 0.001072798273526132,
|
6338 |
+
"learning_rate": 6.44302267379111e-05,
|
6339 |
+
"loss": 46.0,
|
6340 |
+
"step": 901
|
6341 |
+
},
|
6342 |
+
{
|
6343 |
+
"epoch": 0.15221701894274986,
|
6344 |
+
"grad_norm": 0.0006312826299108565,
|
6345 |
+
"learning_rate": 6.422739878856735e-05,
|
6346 |
+
"loss": 46.0,
|
6347 |
+
"step": 902
|
6348 |
+
},
|
6349 |
+
{
|
6350 |
+
"epoch": 0.15238577395266423,
|
6351 |
+
"grad_norm": 0.002877163467928767,
|
6352 |
+
"learning_rate": 6.402473946075671e-05,
|
6353 |
+
"loss": 46.0,
|
6354 |
+
"step": 903
|
6355 |
+
},
|
6356 |
+
{
|
6357 |
+
"epoch": 0.15255452896257857,
|
6358 |
+
"grad_norm": 0.0008996203541755676,
|
6359 |
+
"learning_rate": 6.382224970975572e-05,
|
6360 |
+
"loss": 46.0,
|
6361 |
+
"step": 904
|
6362 |
+
},
|
6363 |
+
{
|
6364 |
+
"epoch": 0.15272328397249293,
|
6365 |
+
"grad_norm": 0.0006124867359176278,
|
6366 |
+
"learning_rate": 6.361993049004163e-05,
|
6367 |
+
"loss": 46.0,
|
6368 |
+
"step": 905
|
6369 |
+
},
|
6370 |
+
{
|
6371 |
+
"epoch": 0.1528920389824073,
|
6372 |
+
"grad_norm": 0.001687378971837461,
|
6373 |
+
"learning_rate": 6.341778275528773e-05,
|
6374 |
+
"loss": 46.0,
|
6375 |
+
"step": 906
|
6376 |
+
},
|
6377 |
+
{
|
6378 |
+
"epoch": 0.15306079399232164,
|
6379 |
+
"grad_norm": 0.001008613035082817,
|
6380 |
+
"learning_rate": 6.321580745835915e-05,
|
6381 |
+
"loss": 46.0,
|
6382 |
+
"step": 907
|
6383 |
+
},
|
6384 |
+
{
|
6385 |
+
"epoch": 0.153229549002236,
|
6386 |
+
"grad_norm": 0.0006372160278260708,
|
6387 |
+
"learning_rate": 6.301400555130805e-05,
|
6388 |
+
"loss": 46.0,
|
6389 |
+
"step": 908
|
6390 |
+
},
|
6391 |
+
{
|
6392 |
+
"epoch": 0.15339830401215035,
|
6393 |
+
"grad_norm": 0.0005495776422321796,
|
6394 |
+
"learning_rate": 6.281237798536932e-05,
|
6395 |
+
"loss": 46.0,
|
6396 |
+
"step": 909
|
6397 |
+
},
|
6398 |
+
{
|
6399 |
+
"epoch": 0.15356705902206472,
|
6400 |
+
"grad_norm": 0.0008954824879765511,
|
6401 |
+
"learning_rate": 6.261092571095614e-05,
|
6402 |
+
"loss": 46.0,
|
6403 |
+
"step": 910
|
6404 |
+
},
|
6405 |
+
{
|
6406 |
+
"epoch": 0.15373581403197908,
|
6407 |
+
"grad_norm": 0.001377457519993186,
|
6408 |
+
"learning_rate": 6.240964967765528e-05,
|
6409 |
+
"loss": 46.0,
|
6410 |
+
"step": 911
|
6411 |
+
},
|
6412 |
+
{
|
6413 |
+
"epoch": 0.15390456904189342,
|
6414 |
+
"grad_norm": 0.0011930032633244991,
|
6415 |
+
"learning_rate": 6.220855083422285e-05,
|
6416 |
+
"loss": 46.0,
|
6417 |
+
"step": 912
|
6418 |
+
},
|
6419 |
+
{
|
6420 |
+
"epoch": 0.1540733240518078,
|
6421 |
+
"grad_norm": 0.0013531373115256429,
|
6422 |
+
"learning_rate": 6.200763012857973e-05,
|
6423 |
+
"loss": 46.0,
|
6424 |
+
"step": 913
|
6425 |
+
},
|
6426 |
+
{
|
6427 |
+
"epoch": 0.15424207906172213,
|
6428 |
+
"grad_norm": 0.0009069031220860779,
|
6429 |
+
"learning_rate": 6.180688850780711e-05,
|
6430 |
+
"loss": 46.0,
|
6431 |
+
"step": 914
|
6432 |
+
},
|
6433 |
+
{
|
6434 |
+
"epoch": 0.1544108340716365,
|
6435 |
+
"grad_norm": 0.0008781193173490465,
|
6436 |
+
"learning_rate": 6.160632691814203e-05,
|
6437 |
+
"loss": 46.0,
|
6438 |
+
"step": 915
|
6439 |
+
},
|
6440 |
+
{
|
6441 |
+
"epoch": 0.15457958908155087,
|
6442 |
+
"grad_norm": 0.0026150341145694256,
|
6443 |
+
"learning_rate": 6.140594630497287e-05,
|
6444 |
+
"loss": 46.0,
|
6445 |
+
"step": 916
|
6446 |
+
},
|
6447 |
+
{
|
6448 |
+
"epoch": 0.1547483440914652,
|
6449 |
+
"grad_norm": 0.0013742209412157536,
|
6450 |
+
"learning_rate": 6.120574761283497e-05,
|
6451 |
+
"loss": 46.0,
|
6452 |
+
"step": 917
|
6453 |
+
},
|
6454 |
+
{
|
6455 |
+
"epoch": 0.15491709910137957,
|
6456 |
+
"grad_norm": 0.0027837788220494986,
|
6457 |
+
"learning_rate": 6.100573178540619e-05,
|
6458 |
+
"loss": 46.0,
|
6459 |
+
"step": 918
|
6460 |
+
},
|
6461 |
+
{
|
6462 |
+
"epoch": 0.15508585411129394,
|
6463 |
+
"grad_norm": 0.000727921084035188,
|
6464 |
+
"learning_rate": 6.080589976550233e-05,
|
6465 |
+
"loss": 46.0,
|
6466 |
+
"step": 919
|
6467 |
+
},
|
6468 |
+
{
|
6469 |
+
"epoch": 0.15525460912120828,
|
6470 |
+
"grad_norm": 0.0010688056936487556,
|
6471 |
+
"learning_rate": 6.06062524950729e-05,
|
6472 |
+
"loss": 46.0,
|
6473 |
+
"step": 920
|
6474 |
+
},
|
6475 |
+
{
|
6476 |
+
"epoch": 0.15542336413112265,
|
6477 |
+
"grad_norm": 0.0006157424068078399,
|
6478 |
+
"learning_rate": 6.040679091519643e-05,
|
6479 |
+
"loss": 46.0,
|
6480 |
+
"step": 921
|
6481 |
+
},
|
6482 |
+
{
|
6483 |
+
"epoch": 0.155592119141037,
|
6484 |
+
"grad_norm": 0.0018731793388724327,
|
6485 |
+
"learning_rate": 6.020751596607621e-05,
|
6486 |
+
"loss": 46.0,
|
6487 |
+
"step": 922
|
6488 |
+
},
|
6489 |
+
{
|
6490 |
+
"epoch": 0.15576087415095136,
|
6491 |
+
"grad_norm": 0.0011395288165658712,
|
6492 |
+
"learning_rate": 6.000842858703585e-05,
|
6493 |
+
"loss": 46.0,
|
6494 |
+
"step": 923
|
6495 |
+
},
|
6496 |
+
{
|
6497 |
+
"epoch": 0.15592962916086572,
|
6498 |
+
"grad_norm": 0.0004488412232603878,
|
6499 |
+
"learning_rate": 5.980952971651472e-05,
|
6500 |
+
"loss": 46.0,
|
6501 |
+
"step": 924
|
6502 |
+
},
|
6503 |
+
{
|
6504 |
+
"epoch": 0.15609838417078006,
|
6505 |
+
"grad_norm": 0.0012818429386243224,
|
6506 |
+
"learning_rate": 5.9610820292063665e-05,
|
6507 |
+
"loss": 46.0,
|
6508 |
+
"step": 925
|
6509 |
+
},
|
6510 |
+
{
|
6511 |
+
"epoch": 0.15626713918069443,
|
6512 |
+
"grad_norm": 0.0011927819577977061,
|
6513 |
+
"learning_rate": 5.9412301250340584e-05,
|
6514 |
+
"loss": 46.0,
|
6515 |
+
"step": 926
|
6516 |
+
},
|
6517 |
+
{
|
6518 |
+
"epoch": 0.15643589419060877,
|
6519 |
+
"grad_norm": 0.0010172206675633788,
|
6520 |
+
"learning_rate": 5.921397352710587e-05,
|
6521 |
+
"loss": 46.0,
|
6522 |
+
"step": 927
|
6523 |
+
},
|
6524 |
+
{
|
6525 |
+
"epoch": 0.15660464920052314,
|
6526 |
+
"grad_norm": 0.001916920649819076,
|
6527 |
+
"learning_rate": 5.9015838057218196e-05,
|
6528 |
+
"loss": 46.0,
|
6529 |
+
"step": 928
|
6530 |
+
},
|
6531 |
+
{
|
6532 |
+
"epoch": 0.1567734042104375,
|
6533 |
+
"grad_norm": 0.00064216268947348,
|
6534 |
+
"learning_rate": 5.881789577462993e-05,
|
6535 |
+
"loss": 46.0,
|
6536 |
+
"step": 929
|
6537 |
+
},
|
6538 |
+
{
|
6539 |
+
"epoch": 0.15694215922035185,
|
6540 |
+
"grad_norm": 0.0006899251602590084,
|
6541 |
+
"learning_rate": 5.862014761238281e-05,
|
6542 |
+
"loss": 46.0,
|
6543 |
+
"step": 930
|
6544 |
+
},
|
6545 |
+
{
|
6546 |
+
"epoch": 0.1571109142302662,
|
6547 |
+
"grad_norm": 0.0008296074229292572,
|
6548 |
+
"learning_rate": 5.842259450260366e-05,
|
6549 |
+
"loss": 46.0,
|
6550 |
+
"step": 931
|
6551 |
+
},
|
6552 |
+
{
|
6553 |
+
"epoch": 0.15727966924018058,
|
6554 |
+
"grad_norm": 0.0014352595899254084,
|
6555 |
+
"learning_rate": 5.822523737649974e-05,
|
6556 |
+
"loss": 46.0,
|
6557 |
+
"step": 932
|
6558 |
+
},
|
6559 |
+
{
|
6560 |
+
"epoch": 0.15744842425009492,
|
6561 |
+
"grad_norm": 0.0011921770637854934,
|
6562 |
+
"learning_rate": 5.80280771643546e-05,
|
6563 |
+
"loss": 46.0,
|
6564 |
+
"step": 933
|
6565 |
+
},
|
6566 |
+
{
|
6567 |
+
"epoch": 0.1576171792600093,
|
6568 |
+
"grad_norm": 0.0011576671386137605,
|
6569 |
+
"learning_rate": 5.7831114795523547e-05,
|
6570 |
+
"loss": 46.0,
|
6571 |
+
"step": 934
|
6572 |
+
},
|
6573 |
+
{
|
6574 |
+
"epoch": 0.15778593426992363,
|
6575 |
+
"grad_norm": 0.000808388867881149,
|
6576 |
+
"learning_rate": 5.7634351198429304e-05,
|
6577 |
+
"loss": 46.0,
|
6578 |
+
"step": 935
|
6579 |
+
},
|
6580 |
+
{
|
6581 |
+
"epoch": 0.157954689279838,
|
6582 |
+
"grad_norm": 0.0016751131042838097,
|
6583 |
+
"learning_rate": 5.74377873005577e-05,
|
6584 |
+
"loss": 46.0,
|
6585 |
+
"step": 936
|
6586 |
+
},
|
6587 |
+
{
|
6588 |
+
"epoch": 0.15812344428975236,
|
6589 |
+
"grad_norm": 0.0008697423618286848,
|
6590 |
+
"learning_rate": 5.724142402845318e-05,
|
6591 |
+
"loss": 46.0,
|
6592 |
+
"step": 937
|
6593 |
+
},
|
6594 |
+
{
|
6595 |
+
"epoch": 0.1582921992996667,
|
6596 |
+
"grad_norm": 0.0005162409506738186,
|
6597 |
+
"learning_rate": 5.7045262307714497e-05,
|
6598 |
+
"loss": 46.0,
|
6599 |
+
"step": 938
|
6600 |
+
},
|
6601 |
+
{
|
6602 |
+
"epoch": 0.15846095430958107,
|
6603 |
+
"grad_norm": 0.0005604674806818366,
|
6604 |
+
"learning_rate": 5.68493030629905e-05,
|
6605 |
+
"loss": 46.0,
|
6606 |
+
"step": 939
|
6607 |
+
},
|
6608 |
+
{
|
6609 |
+
"epoch": 0.1586297093194954,
|
6610 |
+
"grad_norm": 0.0007223184802569449,
|
6611 |
+
"learning_rate": 5.6653547217975354e-05,
|
6612 |
+
"loss": 46.0,
|
6613 |
+
"step": 940
|
6614 |
+
},
|
6615 |
+
{
|
6616 |
+
"epoch": 0.15879846432940978,
|
6617 |
+
"grad_norm": 0.0010802462929859757,
|
6618 |
+
"learning_rate": 5.645799569540463e-05,
|
6619 |
+
"loss": 46.0,
|
6620 |
+
"step": 941
|
6621 |
+
},
|
6622 |
+
{
|
6623 |
+
"epoch": 0.15896721933932415,
|
6624 |
+
"grad_norm": 0.0014821782242506742,
|
6625 |
+
"learning_rate": 5.626264941705086e-05,
|
6626 |
+
"loss": 46.0,
|
6627 |
+
"step": 942
|
6628 |
+
},
|
6629 |
+
{
|
6630 |
+
"epoch": 0.15913597434923848,
|
6631 |
+
"grad_norm": 0.000693553127348423,
|
6632 |
+
"learning_rate": 5.606750930371888e-05,
|
6633 |
+
"loss": 46.0,
|
6634 |
+
"step": 943
|
6635 |
+
},
|
6636 |
+
{
|
6637 |
+
"epoch": 0.15930472935915285,
|
6638 |
+
"grad_norm": 0.000700769480317831,
|
6639 |
+
"learning_rate": 5.587257627524195e-05,
|
6640 |
+
"loss": 46.0,
|
6641 |
+
"step": 944
|
6642 |
+
},
|
6643 |
+
{
|
6644 |
+
"epoch": 0.15947348436906722,
|
6645 |
+
"grad_norm": 0.0012417947873473167,
|
6646 |
+
"learning_rate": 5.567785125047708e-05,
|
6647 |
+
"loss": 46.0,
|
6648 |
+
"step": 945
|
6649 |
+
},
|
6650 |
+
{
|
6651 |
+
"epoch": 0.15964223937898156,
|
6652 |
+
"grad_norm": 0.0007827091030776501,
|
6653 |
+
"learning_rate": 5.548333514730082e-05,
|
6654 |
+
"loss": 46.0,
|
6655 |
+
"step": 946
|
6656 |
+
},
|
6657 |
+
{
|
6658 |
+
"epoch": 0.15981099438889593,
|
6659 |
+
"grad_norm": 0.0004621573316399008,
|
6660 |
+
"learning_rate": 5.528902888260493e-05,
|
6661 |
+
"loss": 46.0,
|
6662 |
+
"step": 947
|
6663 |
+
},
|
6664 |
+
{
|
6665 |
+
"epoch": 0.15997974939881027,
|
6666 |
+
"grad_norm": 0.0009308409062214196,
|
6667 |
+
"learning_rate": 5.509493337229208e-05,
|
6668 |
+
"loss": 46.0,
|
6669 |
+
"step": 948
|
6670 |
+
},
|
6671 |
+
{
|
6672 |
+
"epoch": 0.16014850440872463,
|
6673 |
+
"grad_norm": 0.000546308874618262,
|
6674 |
+
"learning_rate": 5.4901049531271474e-05,
|
6675 |
+
"loss": 46.0,
|
6676 |
+
"step": 949
|
6677 |
+
},
|
6678 |
+
{
|
6679 |
+
"epoch": 0.160317259418639,
|
6680 |
+
"grad_norm": 0.002490605926141143,
|
6681 |
+
"learning_rate": 5.470737827345458e-05,
|
6682 |
+
"loss": 46.0,
|
6683 |
+
"step": 950
|
6684 |
+
},
|
6685 |
+
{
|
6686 |
+
"epoch": 0.16048601442855334,
|
6687 |
+
"grad_norm": 0.0007810198585502803,
|
6688 |
+
"learning_rate": 5.451392051175079e-05,
|
6689 |
+
"loss": 46.0,
|
6690 |
+
"step": 951
|
6691 |
+
},
|
6692 |
+
{
|
6693 |
+
"epoch": 0.1606547694384677,
|
6694 |
+
"grad_norm": 0.0004202220879960805,
|
6695 |
+
"learning_rate": 5.4320677158063246e-05,
|
6696 |
+
"loss": 46.0,
|
6697 |
+
"step": 952
|
6698 |
+
},
|
6699 |
+
{
|
6700 |
+
"epoch": 0.16082352444838205,
|
6701 |
+
"grad_norm": 0.0005529209738597274,
|
6702 |
+
"learning_rate": 5.4127649123284264e-05,
|
6703 |
+
"loss": 46.0,
|
6704 |
+
"step": 953
|
6705 |
+
},
|
6706 |
+
{
|
6707 |
+
"epoch": 0.16099227945829642,
|
6708 |
+
"grad_norm": 0.000546621042303741,
|
6709 |
+
"learning_rate": 5.3934837317291276e-05,
|
6710 |
+
"loss": 46.0,
|
6711 |
+
"step": 954
|
6712 |
+
},
|
6713 |
+
{
|
6714 |
+
"epoch": 0.16116103446821078,
|
6715 |
+
"grad_norm": 0.0007445806404575706,
|
6716 |
+
"learning_rate": 5.374224264894261e-05,
|
6717 |
+
"loss": 46.0,
|
6718 |
+
"step": 955
|
6719 |
+
},
|
6720 |
+
{
|
6721 |
+
"epoch": 0.16132978947812512,
|
6722 |
+
"grad_norm": 0.0006346721784211695,
|
6723 |
+
"learning_rate": 5.354986602607279e-05,
|
6724 |
+
"loss": 46.0,
|
6725 |
+
"step": 956
|
6726 |
+
},
|
6727 |
+
{
|
6728 |
+
"epoch": 0.1614985444880395,
|
6729 |
+
"grad_norm": 0.0010019588517025113,
|
6730 |
+
"learning_rate": 5.335770835548883e-05,
|
6731 |
+
"loss": 46.0,
|
6732 |
+
"step": 957
|
6733 |
+
},
|
6734 |
+
{
|
6735 |
+
"epoch": 0.16166729949795386,
|
6736 |
+
"grad_norm": 0.001932504354044795,
|
6737 |
+
"learning_rate": 5.316577054296551e-05,
|
6738 |
+
"loss": 46.0,
|
6739 |
+
"step": 958
|
6740 |
+
},
|
6741 |
+
{
|
6742 |
+
"epoch": 0.1618360545078682,
|
6743 |
+
"grad_norm": 0.0013626782456412911,
|
6744 |
+
"learning_rate": 5.2974053493241274e-05,
|
6745 |
+
"loss": 46.0,
|
6746 |
+
"step": 959
|
6747 |
+
},
|
6748 |
+
{
|
6749 |
+
"epoch": 0.16200480951778257,
|
6750 |
+
"grad_norm": 0.0013853182317689061,
|
6751 |
+
"learning_rate": 5.278255811001398e-05,
|
6752 |
+
"loss": 46.0,
|
6753 |
+
"step": 960
|
6754 |
+
},
|
6755 |
+
{
|
6756 |
+
"epoch": 0.1621735645276969,
|
6757 |
+
"grad_norm": 0.00069183245068416,
|
6758 |
+
"learning_rate": 5.259128529593661e-05,
|
6759 |
+
"loss": 46.0,
|
6760 |
+
"step": 961
|
6761 |
+
},
|
6762 |
+
{
|
6763 |
+
"epoch": 0.16234231953761127,
|
6764 |
+
"grad_norm": 0.00028503654175437987,
|
6765 |
+
"learning_rate": 5.240023595261301e-05,
|
6766 |
+
"loss": 46.0,
|
6767 |
+
"step": 962
|
6768 |
+
},
|
6769 |
+
{
|
6770 |
+
"epoch": 0.16251107454752564,
|
6771 |
+
"grad_norm": 0.0003797942481469363,
|
6772 |
+
"learning_rate": 5.2209410980593674e-05,
|
6773 |
+
"loss": 46.0,
|
6774 |
+
"step": 963
|
6775 |
+
},
|
6776 |
+
{
|
6777 |
+
"epoch": 0.16267982955743998,
|
6778 |
+
"grad_norm": 0.0009620683849789202,
|
6779 |
+
"learning_rate": 5.2018811279371416e-05,
|
6780 |
+
"loss": 46.0,
|
6781 |
+
"step": 964
|
6782 |
+
},
|
6783 |
+
{
|
6784 |
+
"epoch": 0.16284858456735435,
|
6785 |
+
"grad_norm": 0.0010808344231918454,
|
6786 |
+
"learning_rate": 5.1828437747377354e-05,
|
6787 |
+
"loss": 46.0,
|
6788 |
+
"step": 965
|
6789 |
+
},
|
6790 |
+
{
|
6791 |
+
"epoch": 0.1630173395772687,
|
6792 |
+
"grad_norm": 0.0012897250708192587,
|
6793 |
+
"learning_rate": 5.163829128197626e-05,
|
6794 |
+
"loss": 46.0,
|
6795 |
+
"step": 966
|
6796 |
+
},
|
6797 |
+
{
|
6798 |
+
"epoch": 0.16318609458718306,
|
6799 |
+
"grad_norm": 0.0009198890766128898,
|
6800 |
+
"learning_rate": 5.144837277946273e-05,
|
6801 |
+
"loss": 46.0,
|
6802 |
+
"step": 967
|
6803 |
+
},
|
6804 |
+
{
|
6805 |
+
"epoch": 0.16335484959709742,
|
6806 |
+
"grad_norm": 0.0015073876129463315,
|
6807 |
+
"learning_rate": 5.125868313505691e-05,
|
6808 |
+
"loss": 46.0,
|
6809 |
+
"step": 968
|
6810 |
+
},
|
6811 |
+
{
|
6812 |
+
"epoch": 0.16352360460701176,
|
6813 |
+
"grad_norm": 0.000782114511821419,
|
6814 |
+
"learning_rate": 5.1069223242899876e-05,
|
6815 |
+
"loss": 46.0,
|
6816 |
+
"step": 969
|
6817 |
+
},
|
6818 |
+
{
|
6819 |
+
"epoch": 0.16369235961692613,
|
6820 |
+
"grad_norm": 0.0009239850332960486,
|
6821 |
+
"learning_rate": 5.087999399605006e-05,
|
6822 |
+
"loss": 46.0,
|
6823 |
+
"step": 970
|
6824 |
+
},
|
6825 |
+
{
|
6826 |
+
"epoch": 0.16386111462684047,
|
6827 |
+
"grad_norm": 0.0014594901585951447,
|
6828 |
+
"learning_rate": 5.0690996286478464e-05,
|
6829 |
+
"loss": 46.0,
|
6830 |
+
"step": 971
|
6831 |
+
},
|
6832 |
+
{
|
6833 |
+
"epoch": 0.16402986963675484,
|
6834 |
+
"grad_norm": 0.0007826307555660605,
|
6835 |
+
"learning_rate": 5.050223100506479e-05,
|
6836 |
+
"loss": 46.0,
|
6837 |
+
"step": 972
|
6838 |
+
},
|
6839 |
+
{
|
6840 |
+
"epoch": 0.1641986246466692,
|
6841 |
+
"grad_norm": 0.0008899096283130348,
|
6842 |
+
"learning_rate": 5.031369904159311e-05,
|
6843 |
+
"loss": 46.0,
|
6844 |
+
"step": 973
|
6845 |
+
},
|
6846 |
+
{
|
6847 |
+
"epoch": 0.16436737965658355,
|
6848 |
+
"grad_norm": 0.0009231562726199627,
|
6849 |
+
"learning_rate": 5.012540128474773e-05,
|
6850 |
+
"loss": 46.0,
|
6851 |
+
"step": 974
|
6852 |
+
},
|
6853 |
+
{
|
6854 |
+
"epoch": 0.1645361346664979,
|
6855 |
+
"grad_norm": 0.0013324058381840587,
|
6856 |
+
"learning_rate": 4.993733862210894e-05,
|
6857 |
+
"loss": 46.0,
|
6858 |
+
"step": 975
|
6859 |
+
},
|
6860 |
+
{
|
6861 |
+
"epoch": 0.16470488967641228,
|
6862 |
+
"grad_norm": 0.0003165770904161036,
|
6863 |
+
"learning_rate": 4.97495119401489e-05,
|
6864 |
+
"loss": 46.0,
|
6865 |
+
"step": 976
|
6866 |
+
},
|
6867 |
+
{
|
6868 |
+
"epoch": 0.16487364468632662,
|
6869 |
+
"grad_norm": 0.0030829953029751778,
|
6870 |
+
"learning_rate": 4.956192212422737e-05,
|
6871 |
+
"loss": 46.0,
|
6872 |
+
"step": 977
|
6873 |
+
},
|
6874 |
+
{
|
6875 |
+
"epoch": 0.165042399696241,
|
6876 |
+
"grad_norm": 0.0007854963187128305,
|
6877 |
+
"learning_rate": 4.9374570058587735e-05,
|
6878 |
+
"loss": 46.0,
|
6879 |
+
"step": 978
|
6880 |
+
},
|
6881 |
+
{
|
6882 |
+
"epoch": 0.16521115470615533,
|
6883 |
+
"grad_norm": 0.0004671328642871231,
|
6884 |
+
"learning_rate": 4.918745662635249e-05,
|
6885 |
+
"loss": 46.0,
|
6886 |
+
"step": 979
|
6887 |
+
},
|
6888 |
+
{
|
6889 |
+
"epoch": 0.1653799097160697,
|
6890 |
+
"grad_norm": 0.0011242057662457228,
|
6891 |
+
"learning_rate": 4.900058270951938e-05,
|
6892 |
+
"loss": 46.0,
|
6893 |
+
"step": 980
|
6894 |
+
},
|
6895 |
+
{
|
6896 |
+
"epoch": 0.16554866472598406,
|
6897 |
+
"grad_norm": 0.0012494113761931658,
|
6898 |
+
"learning_rate": 4.881394918895727e-05,
|
6899 |
+
"loss": 46.0,
|
6900 |
+
"step": 981
|
6901 |
+
},
|
6902 |
+
{
|
6903 |
+
"epoch": 0.1657174197358984,
|
6904 |
+
"grad_norm": 0.0012623087968677282,
|
6905 |
+
"learning_rate": 4.862755694440164e-05,
|
6906 |
+
"loss": 46.0,
|
6907 |
+
"step": 982
|
6908 |
+
},
|
6909 |
+
{
|
6910 |
+
"epoch": 0.16588617474581277,
|
6911 |
+
"grad_norm": 0.0015923196915537119,
|
6912 |
+
"learning_rate": 4.8441406854450764e-05,
|
6913 |
+
"loss": 46.0,
|
6914 |
+
"step": 983
|
6915 |
+
},
|
6916 |
+
{
|
6917 |
+
"epoch": 0.1660549297557271,
|
6918 |
+
"grad_norm": 0.0006417816039174795,
|
6919 |
+
"learning_rate": 4.8255499796561564e-05,
|
6920 |
+
"loss": 46.0,
|
6921 |
+
"step": 984
|
6922 |
+
},
|
6923 |
+
{
|
6924 |
+
"epoch": 0.16622368476564148,
|
6925 |
+
"grad_norm": 0.0016659084940329194,
|
6926 |
+
"learning_rate": 4.806983664704525e-05,
|
6927 |
+
"loss": 46.0,
|
6928 |
+
"step": 985
|
6929 |
+
},
|
6930 |
+
{
|
6931 |
+
"epoch": 0.16639243977555584,
|
6932 |
+
"grad_norm": 0.0007746117771603167,
|
6933 |
+
"learning_rate": 4.788441828106338e-05,
|
6934 |
+
"loss": 46.0,
|
6935 |
+
"step": 986
|
6936 |
+
},
|
6937 |
+
{
|
6938 |
+
"epoch": 0.16656119478547018,
|
6939 |
+
"grad_norm": 0.002332525560632348,
|
6940 |
+
"learning_rate": 4.769924557262364e-05,
|
6941 |
+
"loss": 46.0,
|
6942 |
+
"step": 987
|
6943 |
+
},
|
6944 |
+
{
|
6945 |
+
"epoch": 0.16672994979538455,
|
6946 |
+
"grad_norm": 0.0006849498022347689,
|
6947 |
+
"learning_rate": 4.751431939457579e-05,
|
6948 |
+
"loss": 46.0,
|
6949 |
+
"step": 988
|
6950 |
+
},
|
6951 |
+
{
|
6952 |
+
"epoch": 0.16689870480529892,
|
6953 |
+
"grad_norm": 0.0005523057770915329,
|
6954 |
+
"learning_rate": 4.732964061860752e-05,
|
6955 |
+
"loss": 46.0,
|
6956 |
+
"step": 989
|
6957 |
+
},
|
6958 |
+
{
|
6959 |
+
"epoch": 0.16706745981521326,
|
6960 |
+
"grad_norm": 0.000503736431710422,
|
6961 |
+
"learning_rate": 4.71452101152403e-05,
|
6962 |
+
"loss": 46.0,
|
6963 |
+
"step": 990
|
6964 |
+
},
|
6965 |
+
{
|
6966 |
+
"epoch": 0.16723621482512763,
|
6967 |
+
"grad_norm": 0.0019155279733240604,
|
6968 |
+
"learning_rate": 4.6961028753825364e-05,
|
6969 |
+
"loss": 46.0,
|
6970 |
+
"step": 991
|
6971 |
+
},
|
6972 |
+
{
|
6973 |
+
"epoch": 0.16740496983504197,
|
6974 |
+
"grad_norm": 0.00047645490849390626,
|
6975 |
+
"learning_rate": 4.677709740253953e-05,
|
6976 |
+
"loss": 46.0,
|
6977 |
+
"step": 992
|
6978 |
+
},
|
6979 |
+
{
|
6980 |
+
"epoch": 0.16757372484495633,
|
6981 |
+
"grad_norm": 0.00034740299452096224,
|
6982 |
+
"learning_rate": 4.659341692838113e-05,
|
6983 |
+
"loss": 46.0,
|
6984 |
+
"step": 993
|
6985 |
+
},
|
6986 |
+
{
|
6987 |
+
"epoch": 0.1677424798548707,
|
6988 |
+
"grad_norm": 0.0007441159687004983,
|
6989 |
+
"learning_rate": 4.640998819716605e-05,
|
6990 |
+
"loss": 46.0,
|
6991 |
+
"step": 994
|
6992 |
+
},
|
6993 |
+
{
|
6994 |
+
"epoch": 0.16791123486478504,
|
6995 |
+
"grad_norm": 0.0011230899253860116,
|
6996 |
+
"learning_rate": 4.622681207352335e-05,
|
6997 |
+
"loss": 46.0,
|
6998 |
+
"step": 995
|
6999 |
+
},
|
7000 |
+
{
|
7001 |
+
"epoch": 0.1680799898746994,
|
7002 |
+
"grad_norm": 0.0009723291150294244,
|
7003 |
+
"learning_rate": 4.604388942089146e-05,
|
7004 |
+
"loss": 46.0,
|
7005 |
+
"step": 996
|
7006 |
+
},
|
7007 |
+
{
|
7008 |
+
"epoch": 0.16824874488461375,
|
7009 |
+
"grad_norm": 0.0005291839479468763,
|
7010 |
+
"learning_rate": 4.586122110151414e-05,
|
7011 |
+
"loss": 46.0,
|
7012 |
+
"step": 997
|
7013 |
+
},
|
7014 |
+
{
|
7015 |
+
"epoch": 0.16841749989452812,
|
7016 |
+
"grad_norm": 0.0013463158393278718,
|
7017 |
+
"learning_rate": 4.5678807976436034e-05,
|
7018 |
+
"loss": 46.0,
|
7019 |
+
"step": 998
|
7020 |
+
},
|
7021 |
+
{
|
7022 |
+
"epoch": 0.16858625490444248,
|
7023 |
+
"grad_norm": 0.0005321825738064945,
|
7024 |
+
"learning_rate": 4.549665090549913e-05,
|
7025 |
+
"loss": 46.0,
|
7026 |
+
"step": 999
|
7027 |
+
},
|
7028 |
+
{
|
7029 |
+
"epoch": 0.16875500991435682,
|
7030 |
+
"grad_norm": 0.0006381895509548485,
|
7031 |
+
"learning_rate": 4.531475074733832e-05,
|
7032 |
+
"loss": 46.0,
|
7033 |
+
"step": 1000
|
7034 |
+
},
|
7035 |
+
{
|
7036 |
+
"epoch": 0.1689237649242712,
|
7037 |
+
"grad_norm": 0.0013920076889917254,
|
7038 |
+
"learning_rate": 4.5133108359377504e-05,
|
7039 |
+
"loss": 46.0,
|
7040 |
+
"step": 1001
|
7041 |
+
},
|
7042 |
+
{
|
7043 |
+
"epoch": 0.16909251993418556,
|
7044 |
+
"grad_norm": 0.0016773123061284423,
|
7045 |
+
"learning_rate": 4.495172459782556e-05,
|
7046 |
+
"loss": 46.0,
|
7047 |
+
"step": 1002
|
7048 |
+
},
|
7049 |
+
{
|
7050 |
+
"epoch": 0.1692612749440999,
|
7051 |
+
"grad_norm": 0.0011099010007455945,
|
7052 |
+
"learning_rate": 4.477060031767223e-05,
|
7053 |
+
"loss": 46.0,
|
7054 |
+
"step": 1003
|
7055 |
+
},
|
7056 |
+
{
|
7057 |
+
"epoch": 0.16943002995401427,
|
7058 |
+
"grad_norm": 0.0020444022957235575,
|
7059 |
+
"learning_rate": 4.4589736372684166e-05,
|
7060 |
+
"loss": 46.0,
|
7061 |
+
"step": 1004
|
7062 |
+
},
|
7063 |
+
{
|
7064 |
+
"epoch": 0.1695987849639286,
|
7065 |
+
"grad_norm": 0.0008166442858055234,
|
7066 |
+
"learning_rate": 4.440913361540087e-05,
|
7067 |
+
"loss": 46.0,
|
7068 |
+
"step": 1005
|
7069 |
+
},
|
7070 |
+
{
|
7071 |
+
"epoch": 0.16976753997384297,
|
7072 |
+
"grad_norm": 0.0013784863986074924,
|
7073 |
+
"learning_rate": 4.4228792897130654e-05,
|
7074 |
+
"loss": 46.0,
|
7075 |
+
"step": 1006
|
7076 |
+
},
|
7077 |
+
{
|
7078 |
+
"epoch": 0.16993629498375734,
|
7079 |
+
"grad_norm": 0.0005884088459424675,
|
7080 |
+
"learning_rate": 4.40487150679468e-05,
|
7081 |
+
"loss": 46.0,
|
7082 |
+
"step": 1007
|
7083 |
+
},
|
7084 |
+
{
|
7085 |
+
"epoch": 0.17010504999367168,
|
7086 |
+
"grad_norm": 0.000776112952735275,
|
7087 |
+
"learning_rate": 4.386890097668317e-05,
|
7088 |
+
"loss": 46.0,
|
7089 |
+
"step": 1008
|
7090 |
+
},
|
7091 |
+
{
|
7092 |
+
"epoch": 0.17027380500358605,
|
7093 |
+
"grad_norm": 0.0008750237757340074,
|
7094 |
+
"learning_rate": 4.3689351470930574e-05,
|
7095 |
+
"loss": 46.0,
|
7096 |
+
"step": 1009
|
7097 |
+
},
|
7098 |
+
{
|
7099 |
+
"epoch": 0.1704425600135004,
|
7100 |
+
"grad_norm": 0.000761337811127305,
|
7101 |
+
"learning_rate": 4.351006739703275e-05,
|
7102 |
+
"loss": 46.0,
|
7103 |
+
"step": 1010
|
7104 |
+
},
|
7105 |
+
{
|
7106 |
+
"epoch": 0.17061131502341476,
|
7107 |
+
"grad_norm": 0.0031786442268639803,
|
7108 |
+
"learning_rate": 4.333104960008203e-05,
|
7109 |
+
"loss": 46.0,
|
7110 |
+
"step": 1011
|
7111 |
+
},
|
7112 |
+
{
|
7113 |
+
"epoch": 0.17078007003332912,
|
7114 |
+
"grad_norm": 0.004400862380862236,
|
7115 |
+
"learning_rate": 4.3152298923915734e-05,
|
7116 |
+
"loss": 46.0,
|
7117 |
+
"step": 1012
|
7118 |
+
},
|
7119 |
+
{
|
7120 |
+
"epoch": 0.17094882504324346,
|
7121 |
+
"grad_norm": 0.0022119544446468353,
|
7122 |
+
"learning_rate": 4.2973816211112095e-05,
|
7123 |
+
"loss": 46.0,
|
7124 |
+
"step": 1013
|
7125 |
+
},
|
7126 |
+
{
|
7127 |
+
"epoch": 0.17111758005315783,
|
7128 |
+
"grad_norm": 0.000868487695697695,
|
7129 |
+
"learning_rate": 4.279560230298616e-05,
|
7130 |
+
"loss": 46.0,
|
7131 |
+
"step": 1014
|
7132 |
+
},
|
7133 |
+
{
|
7134 |
+
"epoch": 0.1712863350630722,
|
7135 |
+
"grad_norm": 0.0006466138293035328,
|
7136 |
+
"learning_rate": 4.261765803958594e-05,
|
7137 |
+
"loss": 46.0,
|
7138 |
+
"step": 1015
|
7139 |
+
},
|
7140 |
+
{
|
7141 |
+
"epoch": 0.17145509007298654,
|
7142 |
+
"grad_norm": 0.0007957179914228618,
|
7143 |
+
"learning_rate": 4.243998425968841e-05,
|
7144 |
+
"loss": 46.0,
|
7145 |
+
"step": 1016
|
7146 |
+
},
|
7147 |
+
{
|
7148 |
+
"epoch": 0.1716238450829009,
|
7149 |
+
"grad_norm": 0.00033654290018603206,
|
7150 |
+
"learning_rate": 4.2262581800795586e-05,
|
7151 |
+
"loss": 46.0,
|
7152 |
+
"step": 1017
|
7153 |
+
},
|
7154 |
+
{
|
7155 |
+
"epoch": 0.17179260009281525,
|
7156 |
+
"grad_norm": 0.003317353082820773,
|
7157 |
+
"learning_rate": 4.2085451499130524e-05,
|
7158 |
+
"loss": 46.0,
|
7159 |
+
"step": 1018
|
7160 |
+
},
|
7161 |
+
{
|
7162 |
+
"epoch": 0.1719613551027296,
|
7163 |
+
"grad_norm": 0.00047639801050536335,
|
7164 |
+
"learning_rate": 4.190859418963341e-05,
|
7165 |
+
"loss": 46.0,
|
7166 |
+
"step": 1019
|
7167 |
+
},
|
7168 |
+
{
|
7169 |
+
"epoch": 0.17213011011264398,
|
7170 |
+
"grad_norm": 0.002305058529600501,
|
7171 |
+
"learning_rate": 4.173201070595775e-05,
|
7172 |
+
"loss": 46.0,
|
7173 |
+
"step": 1020
|
7174 |
+
},
|
7175 |
+
{
|
7176 |
+
"epoch": 0.17229886512255832,
|
7177 |
+
"grad_norm": 0.0012476850533857942,
|
7178 |
+
"learning_rate": 4.1555701880466105e-05,
|
7179 |
+
"loss": 46.0,
|
7180 |
+
"step": 1021
|
7181 |
+
},
|
7182 |
+
{
|
7183 |
+
"epoch": 0.1724676201324727,
|
7184 |
+
"grad_norm": 0.004061322659254074,
|
7185 |
+
"learning_rate": 4.137966854422647e-05,
|
7186 |
+
"loss": 46.0,
|
7187 |
+
"step": 1022
|
7188 |
+
},
|
7189 |
+
{
|
7190 |
+
"epoch": 0.17263637514238703,
|
7191 |
+
"grad_norm": 0.0006147538078948855,
|
7192 |
+
"learning_rate": 4.120391152700841e-05,
|
7193 |
+
"loss": 46.0,
|
7194 |
+
"step": 1023
|
7195 |
+
},
|
7196 |
+
{
|
7197 |
+
"epoch": 0.1728051301523014,
|
7198 |
+
"grad_norm": 0.0011847203131765127,
|
7199 |
+
"learning_rate": 4.102843165727873e-05,
|
7200 |
+
"loss": 46.0,
|
7201 |
+
"step": 1024
|
7202 |
+
},
|
7203 |
+
{
|
7204 |
+
"epoch": 0.17297388516221576,
|
7205 |
+
"grad_norm": 0.0005954731605015695,
|
7206 |
+
"learning_rate": 4.085322976219802e-05,
|
7207 |
+
"loss": 46.0,
|
7208 |
+
"step": 1025
|
7209 |
+
},
|
7210 |
+
{
|
7211 |
+
"epoch": 0.1731426401721301,
|
7212 |
+
"grad_norm": 0.0009544877684675157,
|
7213 |
+
"learning_rate": 4.0678306667616606e-05,
|
7214 |
+
"loss": 46.0,
|
7215 |
+
"step": 1026
|
7216 |
+
},
|
7217 |
+
{
|
7218 |
+
"epoch": 0.17331139518204447,
|
7219 |
+
"grad_norm": 0.0006944774650037289,
|
7220 |
+
"learning_rate": 4.050366319807052e-05,
|
7221 |
+
"loss": 46.0,
|
7222 |
+
"step": 1027
|
7223 |
+
},
|
7224 |
+
{
|
7225 |
+
"epoch": 0.17348015019195884,
|
7226 |
+
"grad_norm": 0.002084943698719144,
|
7227 |
+
"learning_rate": 4.0329300176777775e-05,
|
7228 |
+
"loss": 46.0,
|
7229 |
+
"step": 1028
|
7230 |
+
},
|
7231 |
+
{
|
7232 |
+
"epoch": 0.17364890520187318,
|
7233 |
+
"grad_norm": 0.002356353448703885,
|
7234 |
+
"learning_rate": 4.015521842563444e-05,
|
7235 |
+
"loss": 46.0,
|
7236 |
+
"step": 1029
|
7237 |
+
},
|
7238 |
+
{
|
7239 |
+
"epoch": 0.17381766021178754,
|
7240 |
+
"grad_norm": 0.0005482649430632591,
|
7241 |
+
"learning_rate": 3.998141876521074e-05,
|
7242 |
+
"loss": 46.0,
|
7243 |
+
"step": 1030
|
7244 |
+
},
|
7245 |
+
{
|
7246 |
+
"epoch": 0.17398641522170188,
|
7247 |
+
"grad_norm": 0.0008993589435704052,
|
7248 |
+
"learning_rate": 3.980790201474723e-05,
|
7249 |
+
"loss": 46.0,
|
7250 |
+
"step": 1031
|
7251 |
+
},
|
7252 |
+
{
|
7253 |
+
"epoch": 0.17415517023161625,
|
7254 |
+
"grad_norm": 0.0011524191359058022,
|
7255 |
+
"learning_rate": 3.9634668992150895e-05,
|
7256 |
+
"loss": 46.0,
|
7257 |
+
"step": 1032
|
7258 |
+
},
|
7259 |
+
{
|
7260 |
+
"epoch": 0.17432392524153062,
|
7261 |
+
"grad_norm": 0.0006604307563975453,
|
7262 |
+
"learning_rate": 3.9461720513991285e-05,
|
7263 |
+
"loss": 46.0,
|
7264 |
+
"step": 1033
|
7265 |
+
},
|
7266 |
+
{
|
7267 |
+
"epoch": 0.17449268025144496,
|
7268 |
+
"grad_norm": 0.0013253578217700124,
|
7269 |
+
"learning_rate": 3.928905739549675e-05,
|
7270 |
+
"loss": 46.0,
|
7271 |
+
"step": 1034
|
7272 |
+
},
|
7273 |
+
{
|
7274 |
+
"epoch": 0.17466143526135933,
|
7275 |
+
"grad_norm": 0.0005015510250814259,
|
7276 |
+
"learning_rate": 3.911668045055047e-05,
|
7277 |
+
"loss": 46.0,
|
7278 |
+
"step": 1035
|
7279 |
+
},
|
7280 |
+
{
|
7281 |
+
"epoch": 0.17483019027127367,
|
7282 |
+
"grad_norm": 0.0005467745941132307,
|
7283 |
+
"learning_rate": 3.8944590491686816e-05,
|
7284 |
+
"loss": 46.0,
|
7285 |
+
"step": 1036
|
7286 |
+
},
|
7287 |
+
{
|
7288 |
+
"epoch": 0.17499894528118803,
|
7289 |
+
"grad_norm": 0.0007600914686918259,
|
7290 |
+
"learning_rate": 3.87727883300872e-05,
|
7291 |
+
"loss": 46.0,
|
7292 |
+
"step": 1037
|
7293 |
+
},
|
7294 |
+
{
|
7295 |
+
"epoch": 0.1751677002911024,
|
7296 |
+
"grad_norm": 0.0011487057199701667,
|
7297 |
+
"learning_rate": 3.860127477557654e-05,
|
7298 |
+
"loss": 46.0,
|
7299 |
+
"step": 1038
|
7300 |
+
},
|
7301 |
+
{
|
7302 |
+
"epoch": 0.17533645530101674,
|
7303 |
+
"grad_norm": 0.003172243246808648,
|
7304 |
+
"learning_rate": 3.8430050636619406e-05,
|
7305 |
+
"loss": 46.0,
|
7306 |
+
"step": 1039
|
7307 |
+
},
|
7308 |
+
{
|
7309 |
+
"epoch": 0.1755052103109311,
|
7310 |
+
"grad_norm": 0.0005271739210002124,
|
7311 |
+
"learning_rate": 3.825911672031611e-05,
|
7312 |
+
"loss": 46.0,
|
7313 |
+
"step": 1040
|
7314 |
+
},
|
7315 |
+
{
|
7316 |
+
"epoch": 0.17567396532084545,
|
7317 |
+
"grad_norm": 0.001015481655485928,
|
7318 |
+
"learning_rate": 3.8088473832398806e-05,
|
7319 |
+
"loss": 46.0,
|
7320 |
+
"step": 1041
|
7321 |
+
},
|
7322 |
+
{
|
7323 |
+
"epoch": 0.17584272033075982,
|
7324 |
+
"grad_norm": 0.0004703805025201291,
|
7325 |
+
"learning_rate": 3.7918122777228026e-05,
|
7326 |
+
"loss": 46.0,
|
7327 |
+
"step": 1042
|
7328 |
+
},
|
7329 |
+
{
|
7330 |
+
"epoch": 0.17601147534067418,
|
7331 |
+
"grad_norm": 0.001103199552744627,
|
7332 |
+
"learning_rate": 3.774806435778858e-05,
|
7333 |
+
"loss": 46.0,
|
7334 |
+
"step": 1043
|
7335 |
+
},
|
7336 |
+
{
|
7337 |
+
"epoch": 0.17618023035058852,
|
7338 |
+
"grad_norm": 0.0005269388784654438,
|
7339 |
+
"learning_rate": 3.757829937568591e-05,
|
7340 |
+
"loss": 46.0,
|
7341 |
+
"step": 1044
|
7342 |
+
},
|
7343 |
+
{
|
7344 |
+
"epoch": 0.1763489853605029,
|
7345 |
+
"grad_norm": 0.0017238155705854297,
|
7346 |
+
"learning_rate": 3.7408828631142237e-05,
|
7347 |
+
"loss": 46.0,
|
7348 |
+
"step": 1045
|
7349 |
+
},
|
7350 |
+
{
|
7351 |
+
"epoch": 0.17651774037041726,
|
7352 |
+
"grad_norm": 0.0006927169160917401,
|
7353 |
+
"learning_rate": 3.7239652922992894e-05,
|
7354 |
+
"loss": 46.0,
|
7355 |
+
"step": 1046
|
7356 |
+
},
|
7357 |
+
{
|
7358 |
+
"epoch": 0.1766864953803316,
|
7359 |
+
"grad_norm": 0.002759503899142146,
|
7360 |
+
"learning_rate": 3.707077304868245e-05,
|
7361 |
+
"loss": 46.0,
|
7362 |
+
"step": 1047
|
7363 |
+
},
|
7364 |
+
{
|
7365 |
+
"epoch": 0.17685525039024597,
|
7366 |
+
"grad_norm": 0.0005645275814458728,
|
7367 |
+
"learning_rate": 3.690218980426101e-05,
|
7368 |
+
"loss": 46.0,
|
7369 |
+
"step": 1048
|
7370 |
+
},
|
7371 |
+
{
|
7372 |
+
"epoch": 0.1770240054001603,
|
7373 |
+
"grad_norm": 0.00040731692570261657,
|
7374 |
+
"learning_rate": 3.6733903984380545e-05,
|
7375 |
+
"loss": 46.0,
|
7376 |
+
"step": 1049
|
7377 |
+
},
|
7378 |
+
{
|
7379 |
+
"epoch": 0.17719276041007467,
|
7380 |
+
"grad_norm": 0.0008958344697020948,
|
7381 |
+
"learning_rate": 3.656591638229088e-05,
|
7382 |
+
"loss": 46.0,
|
7383 |
+
"step": 1050
|
7384 |
+
},
|
7385 |
+
{
|
7386 |
+
"epoch": 0.17736151541998904,
|
7387 |
+
"grad_norm": 0.0009003482409752905,
|
7388 |
+
"learning_rate": 3.63982277898362e-05,
|
7389 |
+
"loss": 46.0,
|
7390 |
+
"step": 1051
|
7391 |
+
},
|
7392 |
+
{
|
7393 |
+
"epoch": 0.17753027042990338,
|
7394 |
+
"grad_norm": 0.00033022521529346704,
|
7395 |
+
"learning_rate": 3.6230838997451365e-05,
|
7396 |
+
"loss": 46.0,
|
7397 |
+
"step": 1052
|
7398 |
+
},
|
7399 |
+
{
|
7400 |
+
"epoch": 0.17769902543981775,
|
7401 |
+
"grad_norm": 0.0003871631342917681,
|
7402 |
+
"learning_rate": 3.606375079415797e-05,
|
7403 |
+
"loss": 46.0,
|
7404 |
+
"step": 1053
|
7405 |
+
},
|
7406 |
+
{
|
7407 |
+
"epoch": 0.1778677804497321,
|
7408 |
+
"grad_norm": 0.0018638977780938148,
|
7409 |
+
"learning_rate": 3.589696396756058e-05,
|
7410 |
+
"loss": 46.0,
|
7411 |
+
"step": 1054
|
7412 |
+
},
|
7413 |
+
{
|
7414 |
+
"epoch": 0.17803653545964646,
|
7415 |
+
"grad_norm": 0.0007231778581626713,
|
7416 |
+
"learning_rate": 3.5730479303843446e-05,
|
7417 |
+
"loss": 46.0,
|
7418 |
+
"step": 1055
|
7419 |
+
},
|
7420 |
+
{
|
7421 |
+
"epoch": 0.17820529046956082,
|
7422 |
+
"grad_norm": 0.001085842028260231,
|
7423 |
+
"learning_rate": 3.556429758776629e-05,
|
7424 |
+
"loss": 46.0,
|
7425 |
+
"step": 1056
|
7426 |
+
},
|
7427 |
+
{
|
7428 |
+
"epoch": 0.17837404547947516,
|
7429 |
+
"grad_norm": 0.0005750608397647738,
|
7430 |
+
"learning_rate": 3.5398419602660935e-05,
|
7431 |
+
"loss": 46.0,
|
7432 |
+
"step": 1057
|
7433 |
+
},
|
7434 |
+
{
|
7435 |
+
"epoch": 0.17854280048938953,
|
7436 |
+
"grad_norm": 0.0007418044842779636,
|
7437 |
+
"learning_rate": 3.523284613042745e-05,
|
7438 |
+
"loss": 46.0,
|
7439 |
+
"step": 1058
|
7440 |
+
},
|
7441 |
+
{
|
7442 |
+
"epoch": 0.1787115554993039,
|
7443 |
+
"grad_norm": 0.004355406854301691,
|
7444 |
+
"learning_rate": 3.506757795153056e-05,
|
7445 |
+
"loss": 46.0,
|
7446 |
+
"step": 1059
|
7447 |
+
},
|
7448 |
+
{
|
7449 |
+
"epoch": 0.17888031050921824,
|
7450 |
+
"grad_norm": 0.0003858884156215936,
|
7451 |
+
"learning_rate": 3.49026158449959e-05,
|
7452 |
+
"loss": 46.0,
|
7453 |
+
"step": 1060
|
7454 |
+
},
|
7455 |
+
{
|
7456 |
+
"epoch": 0.1790490655191326,
|
7457 |
+
"grad_norm": 0.0006274758488871157,
|
7458 |
+
"learning_rate": 3.4737960588406374e-05,
|
7459 |
+
"loss": 46.0,
|
7460 |
+
"step": 1061
|
7461 |
+
},
|
7462 |
+
{
|
7463 |
+
"epoch": 0.17921782052904695,
|
7464 |
+
"grad_norm": 0.001342911273241043,
|
7465 |
+
"learning_rate": 3.457361295789849e-05,
|
7466 |
+
"loss": 46.0,
|
7467 |
+
"step": 1062
|
7468 |
+
},
|
7469 |
+
{
|
7470 |
+
"epoch": 0.1793865755389613,
|
7471 |
+
"grad_norm": 0.0006192997680045664,
|
7472 |
+
"learning_rate": 3.440957372815872e-05,
|
7473 |
+
"loss": 46.0,
|
7474 |
+
"step": 1063
|
7475 |
+
},
|
7476 |
+
{
|
7477 |
+
"epoch": 0.17955533054887568,
|
7478 |
+
"grad_norm": 0.0011153030209243298,
|
7479 |
+
"learning_rate": 3.424584367241973e-05,
|
7480 |
+
"loss": 46.0,
|
7481 |
+
"step": 1064
|
7482 |
+
},
|
7483 |
+
{
|
7484 |
+
"epoch": 0.17972408555879002,
|
7485 |
+
"grad_norm": 0.0015017263358458877,
|
7486 |
+
"learning_rate": 3.408242356245701e-05,
|
7487 |
+
"loss": 46.0,
|
7488 |
+
"step": 1065
|
7489 |
+
},
|
7490 |
+
{
|
7491 |
+
"epoch": 0.1798928405687044,
|
7492 |
+
"grad_norm": 0.0004973539616912603,
|
7493 |
+
"learning_rate": 3.3919314168584924e-05,
|
7494 |
+
"loss": 46.0,
|
7495 |
+
"step": 1066
|
7496 |
+
},
|
7497 |
+
{
|
7498 |
+
"epoch": 0.18006159557861873,
|
7499 |
+
"grad_norm": 0.00341922789812088,
|
7500 |
+
"learning_rate": 3.375651625965315e-05,
|
7501 |
+
"loss": 46.0,
|
7502 |
+
"step": 1067
|
7503 |
+
},
|
7504 |
+
{
|
7505 |
+
"epoch": 0.1802303505885331,
|
7506 |
+
"grad_norm": 0.0006390162161551416,
|
7507 |
+
"learning_rate": 3.3594030603043304e-05,
|
7508 |
+
"loss": 46.0,
|
7509 |
+
"step": 1068
|
7510 |
+
},
|
7511 |
+
{
|
7512 |
+
"epoch": 0.18039910559844746,
|
7513 |
+
"grad_norm": 0.00042940647108480334,
|
7514 |
+
"learning_rate": 3.3431857964665003e-05,
|
7515 |
+
"loss": 46.0,
|
7516 |
+
"step": 1069
|
7517 |
+
},
|
7518 |
+
{
|
7519 |
+
"epoch": 0.1805678606083618,
|
7520 |
+
"grad_norm": 0.0022605019621551037,
|
7521 |
+
"learning_rate": 3.326999910895242e-05,
|
7522 |
+
"loss": 46.0,
|
7523 |
+
"step": 1070
|
7524 |
+
},
|
7525 |
+
{
|
7526 |
+
"epoch": 0.18073661561827617,
|
7527 |
+
"grad_norm": 0.0012887527700513601,
|
7528 |
+
"learning_rate": 3.3108454798860656e-05,
|
7529 |
+
"loss": 46.0,
|
7530 |
+
"step": 1071
|
7531 |
+
},
|
7532 |
+
{
|
7533 |
+
"epoch": 0.18090537062819054,
|
7534 |
+
"grad_norm": 0.0010164948180317879,
|
7535 |
+
"learning_rate": 3.2947225795862125e-05,
|
7536 |
+
"loss": 46.0,
|
7537 |
+
"step": 1072
|
7538 |
+
},
|
7539 |
+
{
|
7540 |
+
"epoch": 0.18107412563810488,
|
7541 |
+
"grad_norm": 0.002262418856844306,
|
7542 |
+
"learning_rate": 3.278631285994298e-05,
|
7543 |
+
"loss": 46.0,
|
7544 |
+
"step": 1073
|
7545 |
+
},
|
7546 |
+
{
|
7547 |
+
"epoch": 0.18124288064801924,
|
7548 |
+
"grad_norm": 0.00039437160012312233,
|
7549 |
+
"learning_rate": 3.2625716749599546e-05,
|
7550 |
+
"loss": 46.0,
|
7551 |
+
"step": 1074
|
7552 |
+
},
|
7553 |
+
{
|
7554 |
+
"epoch": 0.18141163565793358,
|
7555 |
+
"grad_norm": 0.0006692737806588411,
|
7556 |
+
"learning_rate": 3.246543822183469e-05,
|
7557 |
+
"loss": 46.0,
|
7558 |
+
"step": 1075
|
7559 |
+
},
|
7560 |
+
{
|
7561 |
+
"epoch": 0.18158039066784795,
|
7562 |
+
"grad_norm": 0.0014356159372255206,
|
7563 |
+
"learning_rate": 3.2305478032154324e-05,
|
7564 |
+
"loss": 46.0,
|
7565 |
+
"step": 1076
|
7566 |
+
},
|
7567 |
+
{
|
7568 |
+
"epoch": 0.18174914567776232,
|
7569 |
+
"grad_norm": 0.001208776026032865,
|
7570 |
+
"learning_rate": 3.2145836934563745e-05,
|
7571 |
+
"loss": 46.0,
|
7572 |
+
"step": 1077
|
7573 |
+
},
|
7574 |
+
{
|
7575 |
+
"epoch": 0.18191790068767666,
|
7576 |
+
"grad_norm": 0.0006577944732271135,
|
7577 |
+
"learning_rate": 3.198651568156426e-05,
|
7578 |
+
"loss": 46.0,
|
7579 |
+
"step": 1078
|
7580 |
+
},
|
7581 |
+
{
|
7582 |
+
"epoch": 0.18208665569759103,
|
7583 |
+
"grad_norm": 0.0010622587287798524,
|
7584 |
+
"learning_rate": 3.1827515024149445e-05,
|
7585 |
+
"loss": 46.0,
|
7586 |
+
"step": 1079
|
7587 |
+
},
|
7588 |
+
{
|
7589 |
+
"epoch": 0.18225541070750537,
|
7590 |
+
"grad_norm": 0.0010634633945301175,
|
7591 |
+
"learning_rate": 3.166883571180159e-05,
|
7592 |
+
"loss": 46.0,
|
7593 |
+
"step": 1080
|
7594 |
+
},
|
7595 |
+
{
|
7596 |
+
"epoch": 0.18242416571741973,
|
7597 |
+
"grad_norm": 0.0010152349714189768,
|
7598 |
+
"learning_rate": 3.151047849248847e-05,
|
7599 |
+
"loss": 46.0,
|
7600 |
+
"step": 1081
|
7601 |
+
},
|
7602 |
+
{
|
7603 |
+
"epoch": 0.1825929207273341,
|
7604 |
+
"grad_norm": 0.0008210540981963277,
|
7605 |
+
"learning_rate": 3.1352444112659484e-05,
|
7606 |
+
"loss": 46.0,
|
7607 |
+
"step": 1082
|
7608 |
+
},
|
7609 |
+
{
|
7610 |
+
"epoch": 0.18276167573724844,
|
7611 |
+
"grad_norm": 0.0006314092315733433,
|
7612 |
+
"learning_rate": 3.119473331724219e-05,
|
7613 |
+
"loss": 46.0,
|
7614 |
+
"step": 1083
|
7615 |
+
},
|
7616 |
+
{
|
7617 |
+
"epoch": 0.1829304307471628,
|
7618 |
+
"grad_norm": 0.0015025155153125525,
|
7619 |
+
"learning_rate": 3.103734684963902e-05,
|
7620 |
+
"loss": 46.0,
|
7621 |
+
"step": 1084
|
7622 |
+
},
|
7623 |
+
{
|
7624 |
+
"epoch": 0.18309918575707718,
|
7625 |
+
"grad_norm": 0.0008753265137784183,
|
7626 |
+
"learning_rate": 3.088028545172352e-05,
|
7627 |
+
"loss": 46.0,
|
7628 |
+
"step": 1085
|
7629 |
+
},
|
7630 |
+
{
|
7631 |
+
"epoch": 0.18326794076699152,
|
7632 |
+
"grad_norm": 0.0014098742976784706,
|
7633 |
+
"learning_rate": 3.0723549863836996e-05,
|
7634 |
+
"loss": 46.0,
|
7635 |
+
"step": 1086
|
7636 |
+
},
|
7637 |
+
{
|
7638 |
+
"epoch": 0.18343669577690588,
|
7639 |
+
"grad_norm": 0.0014926984440535307,
|
7640 |
+
"learning_rate": 3.056714082478496e-05,
|
7641 |
+
"loss": 46.0,
|
7642 |
+
"step": 1087
|
7643 |
+
},
|
7644 |
+
{
|
7645 |
+
"epoch": 0.18360545078682022,
|
7646 |
+
"grad_norm": 0.0008304324583150446,
|
7647 |
+
"learning_rate": 3.0411059071833668e-05,
|
7648 |
+
"loss": 46.0,
|
7649 |
+
"step": 1088
|
7650 |
+
},
|
7651 |
+
{
|
7652 |
+
"epoch": 0.1837742057967346,
|
7653 |
+
"grad_norm": 0.0004860296321567148,
|
7654 |
+
"learning_rate": 3.025530534070664e-05,
|
7655 |
+
"loss": 46.0,
|
7656 |
+
"step": 1089
|
7657 |
+
},
|
7658 |
+
{
|
7659 |
+
"epoch": 0.18394296080664896,
|
7660 |
+
"grad_norm": 0.0006332461489364505,
|
7661 |
+
"learning_rate": 3.00998803655812e-05,
|
7662 |
+
"loss": 46.0,
|
7663 |
+
"step": 1090
|
7664 |
+
},
|
7665 |
+
{
|
7666 |
+
"epoch": 0.1841117158165633,
|
7667 |
+
"grad_norm": 0.0013554253382608294,
|
7668 |
+
"learning_rate": 2.9944784879085065e-05,
|
7669 |
+
"loss": 46.0,
|
7670 |
+
"step": 1091
|
7671 |
+
},
|
7672 |
+
{
|
7673 |
+
"epoch": 0.18428047082647767,
|
7674 |
+
"grad_norm": 0.0008776708273217082,
|
7675 |
+
"learning_rate": 2.979001961229281e-05,
|
7676 |
+
"loss": 46.0,
|
7677 |
+
"step": 1092
|
7678 |
+
},
|
7679 |
+
{
|
7680 |
+
"epoch": 0.184449225836392,
|
7681 |
+
"grad_norm": 0.001277380739338696,
|
7682 |
+
"learning_rate": 2.9635585294722336e-05,
|
7683 |
+
"loss": 46.0,
|
7684 |
+
"step": 1093
|
7685 |
+
},
|
7686 |
+
{
|
7687 |
+
"epoch": 0.18461798084630637,
|
7688 |
+
"grad_norm": 0.000868733914103359,
|
7689 |
+
"learning_rate": 2.948148265433174e-05,
|
7690 |
+
"loss": 46.0,
|
7691 |
+
"step": 1094
|
7692 |
+
},
|
7693 |
+
{
|
7694 |
+
"epoch": 0.18478673585622074,
|
7695 |
+
"grad_norm": 0.0004029095871374011,
|
7696 |
+
"learning_rate": 2.9327712417515597e-05,
|
7697 |
+
"loss": 46.0,
|
7698 |
+
"step": 1095
|
7699 |
+
},
|
7700 |
+
{
|
7701 |
+
"epoch": 0.18478673585622074,
|
7702 |
+
"eval_loss": 11.5,
|
7703 |
+
"eval_runtime": 14.1998,
|
7704 |
+
"eval_samples_per_second": 175.707,
|
7705 |
+
"eval_steps_per_second": 87.888,
|
7706 |
+
"step": 1095
|
7707 |
}
|
7708 |
],
|
7709 |
"logging_steps": 1,
|
|
|
7723 |
"attributes": {}
|
7724 |
}
|
7725 |
},
|
7726 |
+
"total_flos": 21815551918080.0,
|
7727 |
"train_batch_size": 2,
|
7728 |
"trial_name": null,
|
7729 |
"trial_params": null
|