Training in progress, step 1104, 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 144805440
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c14477c439a945f86211657f7e4349f974df07c5853128cffa0a6abbc103b4f5
|
3 |
size 144805440
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 74292308
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c5ab6beef80cd7fb59a9c90acc5cce59b00f24fbaa249f3e569717faffa7b653
|
3 |
size 74292308
|
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:c8efa930a9c9cb068a49f0d425053395984a2e165a94288033fccee6e5cde4d8
|
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:cde2d36e902d746027cc06d5b473736c7fe110d40a364346bbeac42ea4778a3d
|
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": 500,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -5159,6 +5159,2582 @@
|
|
5159 |
"learning_rate": 5.026768725679726e-05,
|
5160 |
"loss": 1.5804,
|
5161 |
"step": 736
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5162 |
}
|
5163 |
],
|
5164 |
"logging_steps": 1,
|
@@ -5178,7 +7754,7 @@
|
|
5178 |
"attributes": {}
|
5179 |
}
|
5180 |
},
|
5181 |
-
"total_flos":
|
5182 |
"train_batch_size": 4,
|
5183 |
"trial_name": null,
|
5184 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.07263516291922299,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 1104,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
5159 |
"learning_rate": 5.026768725679726e-05,
|
5160 |
"loss": 1.5804,
|
5161 |
"step": 736
|
5162 |
+
},
|
5163 |
+
{
|
5164 |
+
"epoch": 0.04848923466618418,
|
5165 |
+
"grad_norm": 0.7959031462669373,
|
5166 |
+
"learning_rate": 5.0160612845131414e-05,
|
5167 |
+
"loss": 1.761,
|
5168 |
+
"step": 737
|
5169 |
+
},
|
5170 |
+
{
|
5171 |
+
"epoch": 0.04855502738621972,
|
5172 |
+
"grad_norm": 0.7459606528282166,
|
5173 |
+
"learning_rate": 5.005353769688611e-05,
|
5174 |
+
"loss": 1.4302,
|
5175 |
+
"step": 738
|
5176 |
+
},
|
5177 |
+
{
|
5178 |
+
"epoch": 0.04862082010625524,
|
5179 |
+
"grad_norm": 0.7083699703216553,
|
5180 |
+
"learning_rate": 4.994646230311391e-05,
|
5181 |
+
"loss": 1.5262,
|
5182 |
+
"step": 739
|
5183 |
+
},
|
5184 |
+
{
|
5185 |
+
"epoch": 0.04868661282629077,
|
5186 |
+
"grad_norm": 0.7352794408798218,
|
5187 |
+
"learning_rate": 4.9839387154868584e-05,
|
5188 |
+
"loss": 1.5299,
|
5189 |
+
"step": 740
|
5190 |
+
},
|
5191 |
+
{
|
5192 |
+
"epoch": 0.0487524055463263,
|
5193 |
+
"grad_norm": 0.7063877582550049,
|
5194 |
+
"learning_rate": 4.973231274320276e-05,
|
5195 |
+
"loss": 1.6491,
|
5196 |
+
"step": 741
|
5197 |
+
},
|
5198 |
+
{
|
5199 |
+
"epoch": 0.04881819826636183,
|
5200 |
+
"grad_norm": 0.7635167837142944,
|
5201 |
+
"learning_rate": 4.962523955916569e-05,
|
5202 |
+
"loss": 1.677,
|
5203 |
+
"step": 742
|
5204 |
+
},
|
5205 |
+
{
|
5206 |
+
"epoch": 0.048883990986397355,
|
5207 |
+
"grad_norm": 0.756692111492157,
|
5208 |
+
"learning_rate": 4.951816809380097e-05,
|
5209 |
+
"loss": 1.4049,
|
5210 |
+
"step": 743
|
5211 |
+
},
|
5212 |
+
{
|
5213 |
+
"epoch": 0.04894978370643288,
|
5214 |
+
"grad_norm": 0.7604605555534363,
|
5215 |
+
"learning_rate": 4.9411098838144346e-05,
|
5216 |
+
"loss": 1.6448,
|
5217 |
+
"step": 744
|
5218 |
+
},
|
5219 |
+
{
|
5220 |
+
"epoch": 0.049015576426468414,
|
5221 |
+
"grad_norm": 0.8182808756828308,
|
5222 |
+
"learning_rate": 4.9304032283221405e-05,
|
5223 |
+
"loss": 1.5077,
|
5224 |
+
"step": 745
|
5225 |
+
},
|
5226 |
+
{
|
5227 |
+
"epoch": 0.04908136914650394,
|
5228 |
+
"grad_norm": 0.7334157824516296,
|
5229 |
+
"learning_rate": 4.919696892004539e-05,
|
5230 |
+
"loss": 1.4763,
|
5231 |
+
"step": 746
|
5232 |
+
},
|
5233 |
+
{
|
5234 |
+
"epoch": 0.04914716186653947,
|
5235 |
+
"grad_norm": 0.8388804197311401,
|
5236 |
+
"learning_rate": 4.908990923961488e-05,
|
5237 |
+
"loss": 1.4618,
|
5238 |
+
"step": 747
|
5239 |
+
},
|
5240 |
+
{
|
5241 |
+
"epoch": 0.04921295458657499,
|
5242 |
+
"grad_norm": 0.8400937914848328,
|
5243 |
+
"learning_rate": 4.898285373291152e-05,
|
5244 |
+
"loss": 1.3205,
|
5245 |
+
"step": 748
|
5246 |
+
},
|
5247 |
+
{
|
5248 |
+
"epoch": 0.049278747306610526,
|
5249 |
+
"grad_norm": 0.9195184111595154,
|
5250 |
+
"learning_rate": 4.887580289089787e-05,
|
5251 |
+
"loss": 1.3075,
|
5252 |
+
"step": 749
|
5253 |
+
},
|
5254 |
+
{
|
5255 |
+
"epoch": 0.04934454002664605,
|
5256 |
+
"grad_norm": 1.210403561592102,
|
5257 |
+
"learning_rate": 4.876875720451511e-05,
|
5258 |
+
"loss": 1.3114,
|
5259 |
+
"step": 750
|
5260 |
+
},
|
5261 |
+
{
|
5262 |
+
"epoch": 0.04941033274668158,
|
5263 |
+
"grad_norm": 0.5409945845603943,
|
5264 |
+
"learning_rate": 4.866171716468074e-05,
|
5265 |
+
"loss": 1.5787,
|
5266 |
+
"step": 751
|
5267 |
+
},
|
5268 |
+
{
|
5269 |
+
"epoch": 0.049476125466717105,
|
5270 |
+
"grad_norm": 0.5661841034889221,
|
5271 |
+
"learning_rate": 4.855468326228638e-05,
|
5272 |
+
"loss": 1.5597,
|
5273 |
+
"step": 752
|
5274 |
+
},
|
5275 |
+
{
|
5276 |
+
"epoch": 0.04954191818675264,
|
5277 |
+
"grad_norm": 0.5858348608016968,
|
5278 |
+
"learning_rate": 4.8447655988195464e-05,
|
5279 |
+
"loss": 1.6057,
|
5280 |
+
"step": 753
|
5281 |
+
},
|
5282 |
+
{
|
5283 |
+
"epoch": 0.049607710906788165,
|
5284 |
+
"grad_norm": 0.5190932154655457,
|
5285 |
+
"learning_rate": 4.834063583324111e-05,
|
5286 |
+
"loss": 1.5158,
|
5287 |
+
"step": 754
|
5288 |
+
},
|
5289 |
+
{
|
5290 |
+
"epoch": 0.04967350362682369,
|
5291 |
+
"grad_norm": 0.6089287996292114,
|
5292 |
+
"learning_rate": 4.8233623288223704e-05,
|
5293 |
+
"loss": 1.7857,
|
5294 |
+
"step": 755
|
5295 |
+
},
|
5296 |
+
{
|
5297 |
+
"epoch": 0.04973929634685922,
|
5298 |
+
"grad_norm": 0.6427363753318787,
|
5299 |
+
"learning_rate": 4.8126618843908775e-05,
|
5300 |
+
"loss": 1.8208,
|
5301 |
+
"step": 756
|
5302 |
+
},
|
5303 |
+
{
|
5304 |
+
"epoch": 0.04980508906689475,
|
5305 |
+
"grad_norm": 0.5550252795219421,
|
5306 |
+
"learning_rate": 4.801962299102471e-05,
|
5307 |
+
"loss": 1.6078,
|
5308 |
+
"step": 757
|
5309 |
+
},
|
5310 |
+
{
|
5311 |
+
"epoch": 0.04987088178693028,
|
5312 |
+
"grad_norm": 0.6121551990509033,
|
5313 |
+
"learning_rate": 4.7912636220260473e-05,
|
5314 |
+
"loss": 1.725,
|
5315 |
+
"step": 758
|
5316 |
+
},
|
5317 |
+
{
|
5318 |
+
"epoch": 0.0499366745069658,
|
5319 |
+
"grad_norm": 0.6313748955726624,
|
5320 |
+
"learning_rate": 4.780565902226338e-05,
|
5321 |
+
"loss": 1.8692,
|
5322 |
+
"step": 759
|
5323 |
+
},
|
5324 |
+
{
|
5325 |
+
"epoch": 0.05000246722700133,
|
5326 |
+
"grad_norm": 0.5803863406181335,
|
5327 |
+
"learning_rate": 4.7698691887636854e-05,
|
5328 |
+
"loss": 1.5714,
|
5329 |
+
"step": 760
|
5330 |
+
},
|
5331 |
+
{
|
5332 |
+
"epoch": 0.05006825994703686,
|
5333 |
+
"grad_norm": 0.5760961174964905,
|
5334 |
+
"learning_rate": 4.759173530693814e-05,
|
5335 |
+
"loss": 1.6356,
|
5336 |
+
"step": 761
|
5337 |
+
},
|
5338 |
+
{
|
5339 |
+
"epoch": 0.05013405266707239,
|
5340 |
+
"grad_norm": 0.5896910429000854,
|
5341 |
+
"learning_rate": 4.748478977067614e-05,
|
5342 |
+
"loss": 1.5632,
|
5343 |
+
"step": 762
|
5344 |
+
},
|
5345 |
+
{
|
5346 |
+
"epoch": 0.050199845387107915,
|
5347 |
+
"grad_norm": 0.5783917903900146,
|
5348 |
+
"learning_rate": 4.737785576930908e-05,
|
5349 |
+
"loss": 1.531,
|
5350 |
+
"step": 763
|
5351 |
+
},
|
5352 |
+
{
|
5353 |
+
"epoch": 0.05026563810714344,
|
5354 |
+
"grad_norm": 0.6067524552345276,
|
5355 |
+
"learning_rate": 4.727093379324222e-05,
|
5356 |
+
"loss": 1.6955,
|
5357 |
+
"step": 764
|
5358 |
+
},
|
5359 |
+
{
|
5360 |
+
"epoch": 0.050331430827178975,
|
5361 |
+
"grad_norm": 0.5850797295570374,
|
5362 |
+
"learning_rate": 4.716402433282575e-05,
|
5363 |
+
"loss": 1.6401,
|
5364 |
+
"step": 765
|
5365 |
+
},
|
5366 |
+
{
|
5367 |
+
"epoch": 0.0503972235472145,
|
5368 |
+
"grad_norm": 0.6247797012329102,
|
5369 |
+
"learning_rate": 4.705712787835247e-05,
|
5370 |
+
"loss": 1.7098,
|
5371 |
+
"step": 766
|
5372 |
+
},
|
5373 |
+
{
|
5374 |
+
"epoch": 0.05046301626725003,
|
5375 |
+
"grad_norm": 0.5667417049407959,
|
5376 |
+
"learning_rate": 4.695024492005548e-05,
|
5377 |
+
"loss": 1.5379,
|
5378 |
+
"step": 767
|
5379 |
+
},
|
5380 |
+
{
|
5381 |
+
"epoch": 0.05052880898728556,
|
5382 |
+
"grad_norm": 0.5960890650749207,
|
5383 |
+
"learning_rate": 4.684337594810602e-05,
|
5384 |
+
"loss": 1.7614,
|
5385 |
+
"step": 768
|
5386 |
+
},
|
5387 |
+
{
|
5388 |
+
"epoch": 0.05059460170732109,
|
5389 |
+
"grad_norm": 0.6005357503890991,
|
5390 |
+
"learning_rate": 4.673652145261116e-05,
|
5391 |
+
"loss": 1.5532,
|
5392 |
+
"step": 769
|
5393 |
+
},
|
5394 |
+
{
|
5395 |
+
"epoch": 0.05066039442735661,
|
5396 |
+
"grad_norm": 0.6175205111503601,
|
5397 |
+
"learning_rate": 4.6629681923611603e-05,
|
5398 |
+
"loss": 1.6746,
|
5399 |
+
"step": 770
|
5400 |
+
},
|
5401 |
+
{
|
5402 |
+
"epoch": 0.05072618714739214,
|
5403 |
+
"grad_norm": 0.6387405395507812,
|
5404 |
+
"learning_rate": 4.652285785107943e-05,
|
5405 |
+
"loss": 1.6399,
|
5406 |
+
"step": 771
|
5407 |
+
},
|
5408 |
+
{
|
5409 |
+
"epoch": 0.05079197986742767,
|
5410 |
+
"grad_norm": 0.5912647247314453,
|
5411 |
+
"learning_rate": 4.64160497249158e-05,
|
5412 |
+
"loss": 1.7165,
|
5413 |
+
"step": 772
|
5414 |
+
},
|
5415 |
+
{
|
5416 |
+
"epoch": 0.0508577725874632,
|
5417 |
+
"grad_norm": 0.6159881353378296,
|
5418 |
+
"learning_rate": 4.630925803494877e-05,
|
5419 |
+
"loss": 1.6739,
|
5420 |
+
"step": 773
|
5421 |
+
},
|
5422 |
+
{
|
5423 |
+
"epoch": 0.050923565307498725,
|
5424 |
+
"grad_norm": 0.6138644218444824,
|
5425 |
+
"learning_rate": 4.6202483270931e-05,
|
5426 |
+
"loss": 1.535,
|
5427 |
+
"step": 774
|
5428 |
+
},
|
5429 |
+
{
|
5430 |
+
"epoch": 0.05098935802753425,
|
5431 |
+
"grad_norm": 0.6410723924636841,
|
5432 |
+
"learning_rate": 4.6095725922537533e-05,
|
5433 |
+
"loss": 1.6135,
|
5434 |
+
"step": 775
|
5435 |
+
},
|
5436 |
+
{
|
5437 |
+
"epoch": 0.051055150747569784,
|
5438 |
+
"grad_norm": 0.6804289817810059,
|
5439 |
+
"learning_rate": 4.598898647936354e-05,
|
5440 |
+
"loss": 1.639,
|
5441 |
+
"step": 776
|
5442 |
+
},
|
5443 |
+
{
|
5444 |
+
"epoch": 0.05112094346760531,
|
5445 |
+
"grad_norm": 0.6560742259025574,
|
5446 |
+
"learning_rate": 4.588226543092209e-05,
|
5447 |
+
"loss": 1.6589,
|
5448 |
+
"step": 777
|
5449 |
+
},
|
5450 |
+
{
|
5451 |
+
"epoch": 0.05118673618764084,
|
5452 |
+
"grad_norm": 0.6562034487724304,
|
5453 |
+
"learning_rate": 4.5775563266641894e-05,
|
5454 |
+
"loss": 1.5987,
|
5455 |
+
"step": 778
|
5456 |
+
},
|
5457 |
+
{
|
5458 |
+
"epoch": 0.05125252890767636,
|
5459 |
+
"grad_norm": 0.7900398969650269,
|
5460 |
+
"learning_rate": 4.566888047586507e-05,
|
5461 |
+
"loss": 1.8027,
|
5462 |
+
"step": 779
|
5463 |
+
},
|
5464 |
+
{
|
5465 |
+
"epoch": 0.051318321627711896,
|
5466 |
+
"grad_norm": 0.6825142502784729,
|
5467 |
+
"learning_rate": 4.556221754784482e-05,
|
5468 |
+
"loss": 1.5588,
|
5469 |
+
"step": 780
|
5470 |
+
},
|
5471 |
+
{
|
5472 |
+
"epoch": 0.05138411434774742,
|
5473 |
+
"grad_norm": 0.6759592294692993,
|
5474 |
+
"learning_rate": 4.545557497174331e-05,
|
5475 |
+
"loss": 1.535,
|
5476 |
+
"step": 781
|
5477 |
+
},
|
5478 |
+
{
|
5479 |
+
"epoch": 0.05144990706778295,
|
5480 |
+
"grad_norm": 0.7403655052185059,
|
5481 |
+
"learning_rate": 4.5348953236629395e-05,
|
5482 |
+
"loss": 1.6259,
|
5483 |
+
"step": 782
|
5484 |
+
},
|
5485 |
+
{
|
5486 |
+
"epoch": 0.051515699787818475,
|
5487 |
+
"grad_norm": 0.711732804775238,
|
5488 |
+
"learning_rate": 4.52423528314763e-05,
|
5489 |
+
"loss": 1.5692,
|
5490 |
+
"step": 783
|
5491 |
+
},
|
5492 |
+
{
|
5493 |
+
"epoch": 0.05158149250785401,
|
5494 |
+
"grad_norm": 0.6846985220909119,
|
5495 |
+
"learning_rate": 4.5135774245159454e-05,
|
5496 |
+
"loss": 1.5152,
|
5497 |
+
"step": 784
|
5498 |
+
},
|
5499 |
+
{
|
5500 |
+
"epoch": 0.051647285227889535,
|
5501 |
+
"grad_norm": 0.6845923066139221,
|
5502 |
+
"learning_rate": 4.502921796645424e-05,
|
5503 |
+
"loss": 1.5296,
|
5504 |
+
"step": 785
|
5505 |
+
},
|
5506 |
+
{
|
5507 |
+
"epoch": 0.05171307794792506,
|
5508 |
+
"grad_norm": 0.7074812054634094,
|
5509 |
+
"learning_rate": 4.492268448403369e-05,
|
5510 |
+
"loss": 1.4728,
|
5511 |
+
"step": 786
|
5512 |
+
},
|
5513 |
+
{
|
5514 |
+
"epoch": 0.05177887066796059,
|
5515 |
+
"grad_norm": 1.4424465894699097,
|
5516 |
+
"learning_rate": 4.4816174286466314e-05,
|
5517 |
+
"loss": 1.4922,
|
5518 |
+
"step": 787
|
5519 |
+
},
|
5520 |
+
{
|
5521 |
+
"epoch": 0.05184466338799612,
|
5522 |
+
"grad_norm": 0.7188639640808105,
|
5523 |
+
"learning_rate": 4.4709687862213866e-05,
|
5524 |
+
"loss": 1.6399,
|
5525 |
+
"step": 788
|
5526 |
+
},
|
5527 |
+
{
|
5528 |
+
"epoch": 0.05191045610803165,
|
5529 |
+
"grad_norm": 0.7116633057594299,
|
5530 |
+
"learning_rate": 4.4603225699629037e-05,
|
5531 |
+
"loss": 1.586,
|
5532 |
+
"step": 789
|
5533 |
+
},
|
5534 |
+
{
|
5535 |
+
"epoch": 0.05197624882806717,
|
5536 |
+
"grad_norm": 0.728759229183197,
|
5537 |
+
"learning_rate": 4.4496788286953266e-05,
|
5538 |
+
"loss": 1.4238,
|
5539 |
+
"step": 790
|
5540 |
+
},
|
5541 |
+
{
|
5542 |
+
"epoch": 0.0520420415481027,
|
5543 |
+
"grad_norm": 0.7237952947616577,
|
5544 |
+
"learning_rate": 4.439037611231448e-05,
|
5545 |
+
"loss": 1.514,
|
5546 |
+
"step": 791
|
5547 |
+
},
|
5548 |
+
{
|
5549 |
+
"epoch": 0.05210783426813823,
|
5550 |
+
"grad_norm": 0.7653865218162537,
|
5551 |
+
"learning_rate": 4.4283989663724875e-05,
|
5552 |
+
"loss": 1.6958,
|
5553 |
+
"step": 792
|
5554 |
+
},
|
5555 |
+
{
|
5556 |
+
"epoch": 0.05217362698817376,
|
5557 |
+
"grad_norm": 0.7304019331932068,
|
5558 |
+
"learning_rate": 4.4177629429078635e-05,
|
5559 |
+
"loss": 1.5366,
|
5560 |
+
"step": 793
|
5561 |
+
},
|
5562 |
+
{
|
5563 |
+
"epoch": 0.052239419708209285,
|
5564 |
+
"grad_norm": 0.7114179730415344,
|
5565 |
+
"learning_rate": 4.407129589614979e-05,
|
5566 |
+
"loss": 1.4206,
|
5567 |
+
"step": 794
|
5568 |
+
},
|
5569 |
+
{
|
5570 |
+
"epoch": 0.05230521242824481,
|
5571 |
+
"grad_norm": 0.7419894337654114,
|
5572 |
+
"learning_rate": 4.396498955258989e-05,
|
5573 |
+
"loss": 1.4686,
|
5574 |
+
"step": 795
|
5575 |
+
},
|
5576 |
+
{
|
5577 |
+
"epoch": 0.052371005148280345,
|
5578 |
+
"grad_norm": 0.8355379700660706,
|
5579 |
+
"learning_rate": 4.385871088592571e-05,
|
5580 |
+
"loss": 1.63,
|
5581 |
+
"step": 796
|
5582 |
+
},
|
5583 |
+
{
|
5584 |
+
"epoch": 0.05243679786831587,
|
5585 |
+
"grad_norm": 0.8412326574325562,
|
5586 |
+
"learning_rate": 4.3752460383557195e-05,
|
5587 |
+
"loss": 1.6383,
|
5588 |
+
"step": 797
|
5589 |
+
},
|
5590 |
+
{
|
5591 |
+
"epoch": 0.0525025905883514,
|
5592 |
+
"grad_norm": 0.8090490102767944,
|
5593 |
+
"learning_rate": 4.3646238532755114e-05,
|
5594 |
+
"loss": 1.407,
|
5595 |
+
"step": 798
|
5596 |
+
},
|
5597 |
+
{
|
5598 |
+
"epoch": 0.05256838330838693,
|
5599 |
+
"grad_norm": 0.8998046517372131,
|
5600 |
+
"learning_rate": 4.3540045820658804e-05,
|
5601 |
+
"loss": 1.625,
|
5602 |
+
"step": 799
|
5603 |
+
},
|
5604 |
+
{
|
5605 |
+
"epoch": 0.05263417602842246,
|
5606 |
+
"grad_norm": 1.3124433755874634,
|
5607 |
+
"learning_rate": 4.3433882734274e-05,
|
5608 |
+
"loss": 1.5433,
|
5609 |
+
"step": 800
|
5610 |
+
},
|
5611 |
+
{
|
5612 |
+
"epoch": 0.05269996874845798,
|
5613 |
+
"grad_norm": 0.5437705516815186,
|
5614 |
+
"learning_rate": 4.332774976047055e-05,
|
5615 |
+
"loss": 1.7119,
|
5616 |
+
"step": 801
|
5617 |
+
},
|
5618 |
+
{
|
5619 |
+
"epoch": 0.05276576146849351,
|
5620 |
+
"grad_norm": 0.5468183159828186,
|
5621 |
+
"learning_rate": 4.322164738598022e-05,
|
5622 |
+
"loss": 1.4418,
|
5623 |
+
"step": 802
|
5624 |
+
},
|
5625 |
+
{
|
5626 |
+
"epoch": 0.05283155418852904,
|
5627 |
+
"grad_norm": 0.5732606053352356,
|
5628 |
+
"learning_rate": 4.311557609739442e-05,
|
5629 |
+
"loss": 1.7286,
|
5630 |
+
"step": 803
|
5631 |
+
},
|
5632 |
+
{
|
5633 |
+
"epoch": 0.05289734690856457,
|
5634 |
+
"grad_norm": 0.5762453675270081,
|
5635 |
+
"learning_rate": 4.300953638116204e-05,
|
5636 |
+
"loss": 1.7308,
|
5637 |
+
"step": 804
|
5638 |
+
},
|
5639 |
+
{
|
5640 |
+
"epoch": 0.052963139628600095,
|
5641 |
+
"grad_norm": 0.5938736796379089,
|
5642 |
+
"learning_rate": 4.290352872358714e-05,
|
5643 |
+
"loss": 1.7055,
|
5644 |
+
"step": 805
|
5645 |
+
},
|
5646 |
+
{
|
5647 |
+
"epoch": 0.05302893234863562,
|
5648 |
+
"grad_norm": 0.5911692976951599,
|
5649 |
+
"learning_rate": 4.27975536108268e-05,
|
5650 |
+
"loss": 1.631,
|
5651 |
+
"step": 806
|
5652 |
+
},
|
5653 |
+
{
|
5654 |
+
"epoch": 0.053094725068671154,
|
5655 |
+
"grad_norm": 0.5975512266159058,
|
5656 |
+
"learning_rate": 4.2691611528888775e-05,
|
5657 |
+
"loss": 1.6773,
|
5658 |
+
"step": 807
|
5659 |
+
},
|
5660 |
+
{
|
5661 |
+
"epoch": 0.05316051778870668,
|
5662 |
+
"grad_norm": 0.5482115745544434,
|
5663 |
+
"learning_rate": 4.258570296362942e-05,
|
5664 |
+
"loss": 1.5443,
|
5665 |
+
"step": 808
|
5666 |
+
},
|
5667 |
+
{
|
5668 |
+
"epoch": 0.05322631050874221,
|
5669 |
+
"grad_norm": 0.6074912548065186,
|
5670 |
+
"learning_rate": 4.24798284007513e-05,
|
5671 |
+
"loss": 1.6919,
|
5672 |
+
"step": 809
|
5673 |
+
},
|
5674 |
+
{
|
5675 |
+
"epoch": 0.05329210322877773,
|
5676 |
+
"grad_norm": 0.5921036601066589,
|
5677 |
+
"learning_rate": 4.2373988325801145e-05,
|
5678 |
+
"loss": 1.6611,
|
5679 |
+
"step": 810
|
5680 |
+
},
|
5681 |
+
{
|
5682 |
+
"epoch": 0.053357895948813266,
|
5683 |
+
"grad_norm": 0.641273558139801,
|
5684 |
+
"learning_rate": 4.2268183224167456e-05,
|
5685 |
+
"loss": 1.8859,
|
5686 |
+
"step": 811
|
5687 |
+
},
|
5688 |
+
{
|
5689 |
+
"epoch": 0.05342368866884879,
|
5690 |
+
"grad_norm": 0.610242486000061,
|
5691 |
+
"learning_rate": 4.216241358107831e-05,
|
5692 |
+
"loss": 1.5211,
|
5693 |
+
"step": 812
|
5694 |
+
},
|
5695 |
+
{
|
5696 |
+
"epoch": 0.05348948138888432,
|
5697 |
+
"grad_norm": 0.6088224053382874,
|
5698 |
+
"learning_rate": 4.205667988159921e-05,
|
5699 |
+
"loss": 1.6132,
|
5700 |
+
"step": 813
|
5701 |
+
},
|
5702 |
+
{
|
5703 |
+
"epoch": 0.053555274108919845,
|
5704 |
+
"grad_norm": 0.5857619643211365,
|
5705 |
+
"learning_rate": 4.195098261063087e-05,
|
5706 |
+
"loss": 1.5505,
|
5707 |
+
"step": 814
|
5708 |
+
},
|
5709 |
+
{
|
5710 |
+
"epoch": 0.05362106682895538,
|
5711 |
+
"grad_norm": 0.6356807351112366,
|
5712 |
+
"learning_rate": 4.1845322252906864e-05,
|
5713 |
+
"loss": 1.6472,
|
5714 |
+
"step": 815
|
5715 |
+
},
|
5716 |
+
{
|
5717 |
+
"epoch": 0.053686859548990905,
|
5718 |
+
"grad_norm": 0.5856146216392517,
|
5719 |
+
"learning_rate": 4.173969929299151e-05,
|
5720 |
+
"loss": 1.7503,
|
5721 |
+
"step": 816
|
5722 |
+
},
|
5723 |
+
{
|
5724 |
+
"epoch": 0.05375265226902643,
|
5725 |
+
"grad_norm": 0.60245680809021,
|
5726 |
+
"learning_rate": 4.1634114215277625e-05,
|
5727 |
+
"loss": 1.5542,
|
5728 |
+
"step": 817
|
5729 |
+
},
|
5730 |
+
{
|
5731 |
+
"epoch": 0.05381844498906196,
|
5732 |
+
"grad_norm": 0.6490529179573059,
|
5733 |
+
"learning_rate": 4.152856750398426e-05,
|
5734 |
+
"loss": 1.5974,
|
5735 |
+
"step": 818
|
5736 |
+
},
|
5737 |
+
{
|
5738 |
+
"epoch": 0.05388423770909749,
|
5739 |
+
"grad_norm": 0.6497132778167725,
|
5740 |
+
"learning_rate": 4.1423059643154564e-05,
|
5741 |
+
"loss": 1.603,
|
5742 |
+
"step": 819
|
5743 |
+
},
|
5744 |
+
{
|
5745 |
+
"epoch": 0.05395003042913302,
|
5746 |
+
"grad_norm": 0.5978454351425171,
|
5747 |
+
"learning_rate": 4.131759111665349e-05,
|
5748 |
+
"loss": 1.4747,
|
5749 |
+
"step": 820
|
5750 |
+
},
|
5751 |
+
{
|
5752 |
+
"epoch": 0.05401582314916854,
|
5753 |
+
"grad_norm": 0.6695238947868347,
|
5754 |
+
"learning_rate": 4.1212162408165595e-05,
|
5755 |
+
"loss": 1.6682,
|
5756 |
+
"step": 821
|
5757 |
+
},
|
5758 |
+
{
|
5759 |
+
"epoch": 0.05408161586920407,
|
5760 |
+
"grad_norm": 0.6585442423820496,
|
5761 |
+
"learning_rate": 4.110677400119285e-05,
|
5762 |
+
"loss": 1.72,
|
5763 |
+
"step": 822
|
5764 |
+
},
|
5765 |
+
{
|
5766 |
+
"epoch": 0.0541474085892396,
|
5767 |
+
"grad_norm": 0.6088001132011414,
|
5768 |
+
"learning_rate": 4.100142637905238e-05,
|
5769 |
+
"loss": 1.474,
|
5770 |
+
"step": 823
|
5771 |
+
},
|
5772 |
+
{
|
5773 |
+
"epoch": 0.05421320130927513,
|
5774 |
+
"grad_norm": 0.6855148673057556,
|
5775 |
+
"learning_rate": 4.0896120024874286e-05,
|
5776 |
+
"loss": 1.6749,
|
5777 |
+
"step": 824
|
5778 |
+
},
|
5779 |
+
{
|
5780 |
+
"epoch": 0.054278994029310655,
|
5781 |
+
"grad_norm": 0.6727811694145203,
|
5782 |
+
"learning_rate": 4.079085542159938e-05,
|
5783 |
+
"loss": 1.7432,
|
5784 |
+
"step": 825
|
5785 |
+
},
|
5786 |
+
{
|
5787 |
+
"epoch": 0.05434478674934618,
|
5788 |
+
"grad_norm": 0.6346832513809204,
|
5789 |
+
"learning_rate": 4.068563305197706e-05,
|
5790 |
+
"loss": 1.5826,
|
5791 |
+
"step": 826
|
5792 |
+
},
|
5793 |
+
{
|
5794 |
+
"epoch": 0.054410579469381715,
|
5795 |
+
"grad_norm": 0.6359438300132751,
|
5796 |
+
"learning_rate": 4.0580453398563e-05,
|
5797 |
+
"loss": 1.5543,
|
5798 |
+
"step": 827
|
5799 |
+
},
|
5800 |
+
{
|
5801 |
+
"epoch": 0.05447637218941724,
|
5802 |
+
"grad_norm": 0.6643136739730835,
|
5803 |
+
"learning_rate": 4.047531694371695e-05,
|
5804 |
+
"loss": 1.5587,
|
5805 |
+
"step": 828
|
5806 |
+
},
|
5807 |
+
{
|
5808 |
+
"epoch": 0.05454216490945277,
|
5809 |
+
"grad_norm": 0.6467865109443665,
|
5810 |
+
"learning_rate": 4.037022416960058e-05,
|
5811 |
+
"loss": 1.7538,
|
5812 |
+
"step": 829
|
5813 |
+
},
|
5814 |
+
{
|
5815 |
+
"epoch": 0.0546079576294883,
|
5816 |
+
"grad_norm": 0.6873615980148315,
|
5817 |
+
"learning_rate": 4.0265175558175265e-05,
|
5818 |
+
"loss": 1.5784,
|
5819 |
+
"step": 830
|
5820 |
+
},
|
5821 |
+
{
|
5822 |
+
"epoch": 0.05467375034952383,
|
5823 |
+
"grad_norm": 0.6892521977424622,
|
5824 |
+
"learning_rate": 4.01601715911998e-05,
|
5825 |
+
"loss": 1.5759,
|
5826 |
+
"step": 831
|
5827 |
+
},
|
5828 |
+
{
|
5829 |
+
"epoch": 0.05473954306955935,
|
5830 |
+
"grad_norm": 0.6923861503601074,
|
5831 |
+
"learning_rate": 4.005521275022826e-05,
|
5832 |
+
"loss": 1.6928,
|
5833 |
+
"step": 832
|
5834 |
+
},
|
5835 |
+
{
|
5836 |
+
"epoch": 0.05480533578959488,
|
5837 |
+
"grad_norm": 0.9242671132087708,
|
5838 |
+
"learning_rate": 3.9950299516607766e-05,
|
5839 |
+
"loss": 1.7272,
|
5840 |
+
"step": 833
|
5841 |
+
},
|
5842 |
+
{
|
5843 |
+
"epoch": 0.05487112850963041,
|
5844 |
+
"grad_norm": 0.7108779549598694,
|
5845 |
+
"learning_rate": 3.9845432371476264e-05,
|
5846 |
+
"loss": 1.632,
|
5847 |
+
"step": 834
|
5848 |
+
},
|
5849 |
+
{
|
5850 |
+
"epoch": 0.05493692122966594,
|
5851 |
+
"grad_norm": 0.7247679233551025,
|
5852 |
+
"learning_rate": 3.9740611795760376e-05,
|
5853 |
+
"loss": 1.5377,
|
5854 |
+
"step": 835
|
5855 |
+
},
|
5856 |
+
{
|
5857 |
+
"epoch": 0.055002713949701465,
|
5858 |
+
"grad_norm": 0.7193149328231812,
|
5859 |
+
"learning_rate": 3.9635838270173107e-05,
|
5860 |
+
"loss": 1.431,
|
5861 |
+
"step": 836
|
5862 |
+
},
|
5863 |
+
{
|
5864 |
+
"epoch": 0.05506850666973699,
|
5865 |
+
"grad_norm": 0.7430593967437744,
|
5866 |
+
"learning_rate": 3.9531112275211736e-05,
|
5867 |
+
"loss": 1.5027,
|
5868 |
+
"step": 837
|
5869 |
+
},
|
5870 |
+
{
|
5871 |
+
"epoch": 0.055134299389772524,
|
5872 |
+
"grad_norm": 0.7426053285598755,
|
5873 |
+
"learning_rate": 3.9426434291155526e-05,
|
5874 |
+
"loss": 1.6552,
|
5875 |
+
"step": 838
|
5876 |
+
},
|
5877 |
+
{
|
5878 |
+
"epoch": 0.05520009210980805,
|
5879 |
+
"grad_norm": 0.7447395324707031,
|
5880 |
+
"learning_rate": 3.9321804798063565e-05,
|
5881 |
+
"loss": 1.5632,
|
5882 |
+
"step": 839
|
5883 |
+
},
|
5884 |
+
{
|
5885 |
+
"epoch": 0.05526588482984358,
|
5886 |
+
"grad_norm": 0.7917671203613281,
|
5887 |
+
"learning_rate": 3.921722427577257e-05,
|
5888 |
+
"loss": 1.5141,
|
5889 |
+
"step": 840
|
5890 |
+
},
|
5891 |
+
{
|
5892 |
+
"epoch": 0.0553316775498791,
|
5893 |
+
"grad_norm": 0.815054178237915,
|
5894 |
+
"learning_rate": 3.9112693203894664e-05,
|
5895 |
+
"loss": 1.5968,
|
5896 |
+
"step": 841
|
5897 |
+
},
|
5898 |
+
{
|
5899 |
+
"epoch": 0.055397470269914637,
|
5900 |
+
"grad_norm": 0.8161969184875488,
|
5901 |
+
"learning_rate": 3.900821206181521e-05,
|
5902 |
+
"loss": 1.487,
|
5903 |
+
"step": 842
|
5904 |
+
},
|
5905 |
+
{
|
5906 |
+
"epoch": 0.05546326298995016,
|
5907 |
+
"grad_norm": 0.7944288849830627,
|
5908 |
+
"learning_rate": 3.890378132869059e-05,
|
5909 |
+
"loss": 1.6617,
|
5910 |
+
"step": 843
|
5911 |
+
},
|
5912 |
+
{
|
5913 |
+
"epoch": 0.05552905570998569,
|
5914 |
+
"grad_norm": 0.801561176776886,
|
5915 |
+
"learning_rate": 3.879940148344595e-05,
|
5916 |
+
"loss": 1.4835,
|
5917 |
+
"step": 844
|
5918 |
+
},
|
5919 |
+
{
|
5920 |
+
"epoch": 0.055594848430021215,
|
5921 |
+
"grad_norm": 0.8297501802444458,
|
5922 |
+
"learning_rate": 3.8695073004773106e-05,
|
5923 |
+
"loss": 1.3569,
|
5924 |
+
"step": 845
|
5925 |
+
},
|
5926 |
+
{
|
5927 |
+
"epoch": 0.05566064115005675,
|
5928 |
+
"grad_norm": 0.8283421993255615,
|
5929 |
+
"learning_rate": 3.859079637112833e-05,
|
5930 |
+
"loss": 1.4448,
|
5931 |
+
"step": 846
|
5932 |
+
},
|
5933 |
+
{
|
5934 |
+
"epoch": 0.055726433870092275,
|
5935 |
+
"grad_norm": 0.862443745136261,
|
5936 |
+
"learning_rate": 3.84865720607301e-05,
|
5937 |
+
"loss": 1.558,
|
5938 |
+
"step": 847
|
5939 |
+
},
|
5940 |
+
{
|
5941 |
+
"epoch": 0.0557922265901278,
|
5942 |
+
"grad_norm": 0.8980772495269775,
|
5943 |
+
"learning_rate": 3.838240055155692e-05,
|
5944 |
+
"loss": 1.4493,
|
5945 |
+
"step": 848
|
5946 |
+
},
|
5947 |
+
{
|
5948 |
+
"epoch": 0.05585801931016333,
|
5949 |
+
"grad_norm": 0.9152320623397827,
|
5950 |
+
"learning_rate": 3.827828232134519e-05,
|
5951 |
+
"loss": 1.2151,
|
5952 |
+
"step": 849
|
5953 |
+
},
|
5954 |
+
{
|
5955 |
+
"epoch": 0.05592381203019886,
|
5956 |
+
"grad_norm": 1.169055700302124,
|
5957 |
+
"learning_rate": 3.8174217847586904e-05,
|
5958 |
+
"loss": 1.2846,
|
5959 |
+
"step": 850
|
5960 |
+
},
|
5961 |
+
{
|
5962 |
+
"epoch": 0.05598960475023439,
|
5963 |
+
"grad_norm": 0.5100705623626709,
|
5964 |
+
"learning_rate": 3.8070207607527584e-05,
|
5965 |
+
"loss": 1.5764,
|
5966 |
+
"step": 851
|
5967 |
+
},
|
5968 |
+
{
|
5969 |
+
"epoch": 0.05605539747026991,
|
5970 |
+
"grad_norm": 0.5505961179733276,
|
5971 |
+
"learning_rate": 3.796625207816401e-05,
|
5972 |
+
"loss": 1.7925,
|
5973 |
+
"step": 852
|
5974 |
+
},
|
5975 |
+
{
|
5976 |
+
"epoch": 0.05612119019030544,
|
5977 |
+
"grad_norm": 0.6031649708747864,
|
5978 |
+
"learning_rate": 3.7862351736242066e-05,
|
5979 |
+
"loss": 1.7848,
|
5980 |
+
"step": 853
|
5981 |
+
},
|
5982 |
+
{
|
5983 |
+
"epoch": 0.05618698291034097,
|
5984 |
+
"grad_norm": 0.6231903433799744,
|
5985 |
+
"learning_rate": 3.775850705825454e-05,
|
5986 |
+
"loss": 1.7385,
|
5987 |
+
"step": 854
|
5988 |
+
},
|
5989 |
+
{
|
5990 |
+
"epoch": 0.0562527756303765,
|
5991 |
+
"grad_norm": 0.5796992778778076,
|
5992 |
+
"learning_rate": 3.765471852043892e-05,
|
5993 |
+
"loss": 1.6627,
|
5994 |
+
"step": 855
|
5995 |
+
},
|
5996 |
+
{
|
5997 |
+
"epoch": 0.056318568350412025,
|
5998 |
+
"grad_norm": 0.5962229371070862,
|
5999 |
+
"learning_rate": 3.755098659877527e-05,
|
6000 |
+
"loss": 1.7171,
|
6001 |
+
"step": 856
|
6002 |
+
},
|
6003 |
+
{
|
6004 |
+
"epoch": 0.05638436107044756,
|
6005 |
+
"grad_norm": 0.6059638857841492,
|
6006 |
+
"learning_rate": 3.7447311768983964e-05,
|
6007 |
+
"loss": 1.6715,
|
6008 |
+
"step": 857
|
6009 |
+
},
|
6010 |
+
{
|
6011 |
+
"epoch": 0.056450153790483085,
|
6012 |
+
"grad_norm": 0.6489062905311584,
|
6013 |
+
"learning_rate": 3.7343694506523605e-05,
|
6014 |
+
"loss": 1.7469,
|
6015 |
+
"step": 858
|
6016 |
+
},
|
6017 |
+
{
|
6018 |
+
"epoch": 0.05651594651051861,
|
6019 |
+
"grad_norm": 0.6001795530319214,
|
6020 |
+
"learning_rate": 3.7240135286588765e-05,
|
6021 |
+
"loss": 1.6136,
|
6022 |
+
"step": 859
|
6023 |
+
},
|
6024 |
+
{
|
6025 |
+
"epoch": 0.05658173923055414,
|
6026 |
+
"grad_norm": 0.6038196086883545,
|
6027 |
+
"learning_rate": 3.713663458410779e-05,
|
6028 |
+
"loss": 1.6115,
|
6029 |
+
"step": 860
|
6030 |
+
},
|
6031 |
+
{
|
6032 |
+
"epoch": 0.05664753195058967,
|
6033 |
+
"grad_norm": 0.584757924079895,
|
6034 |
+
"learning_rate": 3.703319287374069e-05,
|
6035 |
+
"loss": 1.5293,
|
6036 |
+
"step": 861
|
6037 |
+
},
|
6038 |
+
{
|
6039 |
+
"epoch": 0.0567133246706252,
|
6040 |
+
"grad_norm": 0.6052143573760986,
|
6041 |
+
"learning_rate": 3.692981062987698e-05,
|
6042 |
+
"loss": 1.8357,
|
6043 |
+
"step": 862
|
6044 |
+
},
|
6045 |
+
{
|
6046 |
+
"epoch": 0.05677911739066072,
|
6047 |
+
"grad_norm": 0.6260353326797485,
|
6048 |
+
"learning_rate": 3.682648832663339e-05,
|
6049 |
+
"loss": 1.6054,
|
6050 |
+
"step": 863
|
6051 |
+
},
|
6052 |
+
{
|
6053 |
+
"epoch": 0.05684491011069625,
|
6054 |
+
"grad_norm": 0.643218457698822,
|
6055 |
+
"learning_rate": 3.67232264378518e-05,
|
6056 |
+
"loss": 1.7711,
|
6057 |
+
"step": 864
|
6058 |
+
},
|
6059 |
+
{
|
6060 |
+
"epoch": 0.05691070283073178,
|
6061 |
+
"grad_norm": 0.577084481716156,
|
6062 |
+
"learning_rate": 3.662002543709702e-05,
|
6063 |
+
"loss": 1.582,
|
6064 |
+
"step": 865
|
6065 |
+
},
|
6066 |
+
{
|
6067 |
+
"epoch": 0.05697649555076731,
|
6068 |
+
"grad_norm": 0.7629356384277344,
|
6069 |
+
"learning_rate": 3.6516885797654594e-05,
|
6070 |
+
"loss": 1.6112,
|
6071 |
+
"step": 866
|
6072 |
+
},
|
6073 |
+
{
|
6074 |
+
"epoch": 0.057042288270802835,
|
6075 |
+
"grad_norm": 0.6926255226135254,
|
6076 |
+
"learning_rate": 3.641380799252869e-05,
|
6077 |
+
"loss": 1.537,
|
6078 |
+
"step": 867
|
6079 |
+
},
|
6080 |
+
{
|
6081 |
+
"epoch": 0.05710808099083836,
|
6082 |
+
"grad_norm": 0.6118739247322083,
|
6083 |
+
"learning_rate": 3.6310792494439894e-05,
|
6084 |
+
"loss": 1.6167,
|
6085 |
+
"step": 868
|
6086 |
+
},
|
6087 |
+
{
|
6088 |
+
"epoch": 0.057173873710873895,
|
6089 |
+
"grad_norm": 0.6357191205024719,
|
6090 |
+
"learning_rate": 3.620783977582305e-05,
|
6091 |
+
"loss": 1.4908,
|
6092 |
+
"step": 869
|
6093 |
+
},
|
6094 |
+
{
|
6095 |
+
"epoch": 0.05723966643090942,
|
6096 |
+
"grad_norm": 0.6403725743293762,
|
6097 |
+
"learning_rate": 3.61049503088251e-05,
|
6098 |
+
"loss": 1.469,
|
6099 |
+
"step": 870
|
6100 |
+
},
|
6101 |
+
{
|
6102 |
+
"epoch": 0.05730545915094495,
|
6103 |
+
"grad_norm": 0.6135770678520203,
|
6104 |
+
"learning_rate": 3.6002124565302875e-05,
|
6105 |
+
"loss": 1.4758,
|
6106 |
+
"step": 871
|
6107 |
+
},
|
6108 |
+
{
|
6109 |
+
"epoch": 0.05737125187098047,
|
6110 |
+
"grad_norm": 0.6306059956550598,
|
6111 |
+
"learning_rate": 3.5899363016821e-05,
|
6112 |
+
"loss": 1.5523,
|
6113 |
+
"step": 872
|
6114 |
+
},
|
6115 |
+
{
|
6116 |
+
"epoch": 0.05743704459101601,
|
6117 |
+
"grad_norm": 0.6534183025360107,
|
6118 |
+
"learning_rate": 3.579666613464968e-05,
|
6119 |
+
"loss": 1.5653,
|
6120 |
+
"step": 873
|
6121 |
+
},
|
6122 |
+
{
|
6123 |
+
"epoch": 0.05750283731105153,
|
6124 |
+
"grad_norm": 0.6130310893058777,
|
6125 |
+
"learning_rate": 3.56940343897626e-05,
|
6126 |
+
"loss": 1.547,
|
6127 |
+
"step": 874
|
6128 |
+
},
|
6129 |
+
{
|
6130 |
+
"epoch": 0.05756863003108706,
|
6131 |
+
"grad_norm": 0.6713092923164368,
|
6132 |
+
"learning_rate": 3.559146825283465e-05,
|
6133 |
+
"loss": 1.6401,
|
6134 |
+
"step": 875
|
6135 |
+
},
|
6136 |
+
{
|
6137 |
+
"epoch": 0.057634422751122585,
|
6138 |
+
"grad_norm": 0.6866581439971924,
|
6139 |
+
"learning_rate": 3.5488968194239926e-05,
|
6140 |
+
"loss": 1.6317,
|
6141 |
+
"step": 876
|
6142 |
+
},
|
6143 |
+
{
|
6144 |
+
"epoch": 0.05770021547115812,
|
6145 |
+
"grad_norm": 0.6544456481933594,
|
6146 |
+
"learning_rate": 3.538653468404936e-05,
|
6147 |
+
"loss": 1.5999,
|
6148 |
+
"step": 877
|
6149 |
+
},
|
6150 |
+
{
|
6151 |
+
"epoch": 0.057766008191193645,
|
6152 |
+
"grad_norm": 0.6867201328277588,
|
6153 |
+
"learning_rate": 3.528416819202881e-05,
|
6154 |
+
"loss": 1.6641,
|
6155 |
+
"step": 878
|
6156 |
+
},
|
6157 |
+
{
|
6158 |
+
"epoch": 0.05783180091122917,
|
6159 |
+
"grad_norm": 0.7083088159561157,
|
6160 |
+
"learning_rate": 3.5181869187636735e-05,
|
6161 |
+
"loss": 1.499,
|
6162 |
+
"step": 879
|
6163 |
+
},
|
6164 |
+
{
|
6165 |
+
"epoch": 0.0578975936312647,
|
6166 |
+
"grad_norm": 0.7546277046203613,
|
6167 |
+
"learning_rate": 3.5079638140022094e-05,
|
6168 |
+
"loss": 1.6342,
|
6169 |
+
"step": 880
|
6170 |
+
},
|
6171 |
+
{
|
6172 |
+
"epoch": 0.05796338635130023,
|
6173 |
+
"grad_norm": 0.7427401542663574,
|
6174 |
+
"learning_rate": 3.497747551802221e-05,
|
6175 |
+
"loss": 1.7578,
|
6176 |
+
"step": 881
|
6177 |
+
},
|
6178 |
+
{
|
6179 |
+
"epoch": 0.05802917907133576,
|
6180 |
+
"grad_norm": 0.7476483583450317,
|
6181 |
+
"learning_rate": 3.487538179016054e-05,
|
6182 |
+
"loss": 1.5637,
|
6183 |
+
"step": 882
|
6184 |
+
},
|
6185 |
+
{
|
6186 |
+
"epoch": 0.05809497179137128,
|
6187 |
+
"grad_norm": 0.7196775078773499,
|
6188 |
+
"learning_rate": 3.4773357424644685e-05,
|
6189 |
+
"loss": 1.5247,
|
6190 |
+
"step": 883
|
6191 |
+
},
|
6192 |
+
{
|
6193 |
+
"epoch": 0.05816076451140681,
|
6194 |
+
"grad_norm": 0.7455043792724609,
|
6195 |
+
"learning_rate": 3.467140288936407e-05,
|
6196 |
+
"loss": 1.5444,
|
6197 |
+
"step": 884
|
6198 |
+
},
|
6199 |
+
{
|
6200 |
+
"epoch": 0.05822655723144234,
|
6201 |
+
"grad_norm": 0.7095162868499756,
|
6202 |
+
"learning_rate": 3.456951865188791e-05,
|
6203 |
+
"loss": 1.6029,
|
6204 |
+
"step": 885
|
6205 |
+
},
|
6206 |
+
{
|
6207 |
+
"epoch": 0.05829234995147787,
|
6208 |
+
"grad_norm": 0.7810092568397522,
|
6209 |
+
"learning_rate": 3.446770517946303e-05,
|
6210 |
+
"loss": 1.5827,
|
6211 |
+
"step": 886
|
6212 |
+
},
|
6213 |
+
{
|
6214 |
+
"epoch": 0.058358142671513395,
|
6215 |
+
"grad_norm": 0.7805618643760681,
|
6216 |
+
"learning_rate": 3.43659629390117e-05,
|
6217 |
+
"loss": 1.562,
|
6218 |
+
"step": 887
|
6219 |
+
},
|
6220 |
+
{
|
6221 |
+
"epoch": 0.05842393539154893,
|
6222 |
+
"grad_norm": 0.7464081048965454,
|
6223 |
+
"learning_rate": 3.426429239712952e-05,
|
6224 |
+
"loss": 1.646,
|
6225 |
+
"step": 888
|
6226 |
+
},
|
6227 |
+
{
|
6228 |
+
"epoch": 0.058489728111584455,
|
6229 |
+
"grad_norm": 0.7130300998687744,
|
6230 |
+
"learning_rate": 3.4162694020083294e-05,
|
6231 |
+
"loss": 1.3213,
|
6232 |
+
"step": 889
|
6233 |
+
},
|
6234 |
+
{
|
6235 |
+
"epoch": 0.05855552083161998,
|
6236 |
+
"grad_norm": 0.7412570118904114,
|
6237 |
+
"learning_rate": 3.406116827380889e-05,
|
6238 |
+
"loss": 1.6022,
|
6239 |
+
"step": 890
|
6240 |
+
},
|
6241 |
+
{
|
6242 |
+
"epoch": 0.05862131355165551,
|
6243 |
+
"grad_norm": 0.7365239262580872,
|
6244 |
+
"learning_rate": 3.395971562390905e-05,
|
6245 |
+
"loss": 1.5959,
|
6246 |
+
"step": 891
|
6247 |
+
},
|
6248 |
+
{
|
6249 |
+
"epoch": 0.05868710627169104,
|
6250 |
+
"grad_norm": 0.8533955812454224,
|
6251 |
+
"learning_rate": 3.385833653565132e-05,
|
6252 |
+
"loss": 1.766,
|
6253 |
+
"step": 892
|
6254 |
+
},
|
6255 |
+
{
|
6256 |
+
"epoch": 0.05875289899172657,
|
6257 |
+
"grad_norm": 0.8184581995010376,
|
6258 |
+
"learning_rate": 3.375703147396583e-05,
|
6259 |
+
"loss": 1.5165,
|
6260 |
+
"step": 893
|
6261 |
+
},
|
6262 |
+
{
|
6263 |
+
"epoch": 0.05881869171176209,
|
6264 |
+
"grad_norm": 0.7530040740966797,
|
6265 |
+
"learning_rate": 3.365580090344329e-05,
|
6266 |
+
"loss": 1.4796,
|
6267 |
+
"step": 894
|
6268 |
+
},
|
6269 |
+
{
|
6270 |
+
"epoch": 0.05888448443179762,
|
6271 |
+
"grad_norm": 0.8272554278373718,
|
6272 |
+
"learning_rate": 3.3554645288332775e-05,
|
6273 |
+
"loss": 1.3817,
|
6274 |
+
"step": 895
|
6275 |
+
},
|
6276 |
+
{
|
6277 |
+
"epoch": 0.05895027715183315,
|
6278 |
+
"grad_norm": 0.8762139678001404,
|
6279 |
+
"learning_rate": 3.345356509253959e-05,
|
6280 |
+
"loss": 1.4379,
|
6281 |
+
"step": 896
|
6282 |
+
},
|
6283 |
+
{
|
6284 |
+
"epoch": 0.05901606987186868,
|
6285 |
+
"grad_norm": 0.9064789414405823,
|
6286 |
+
"learning_rate": 3.335256077962317e-05,
|
6287 |
+
"loss": 1.5902,
|
6288 |
+
"step": 897
|
6289 |
+
},
|
6290 |
+
{
|
6291 |
+
"epoch": 0.059081862591904205,
|
6292 |
+
"grad_norm": 0.8573328852653503,
|
6293 |
+
"learning_rate": 3.325163281279494e-05,
|
6294 |
+
"loss": 1.4975,
|
6295 |
+
"step": 898
|
6296 |
+
},
|
6297 |
+
{
|
6298 |
+
"epoch": 0.05914765531193973,
|
6299 |
+
"grad_norm": 0.9656907320022583,
|
6300 |
+
"learning_rate": 3.315078165491622e-05,
|
6301 |
+
"loss": 1.5053,
|
6302 |
+
"step": 899
|
6303 |
+
},
|
6304 |
+
{
|
6305 |
+
"epoch": 0.059213448031975265,
|
6306 |
+
"grad_norm": 1.2840112447738647,
|
6307 |
+
"learning_rate": 3.305000776849606e-05,
|
6308 |
+
"loss": 1.2657,
|
6309 |
+
"step": 900
|
6310 |
+
},
|
6311 |
+
{
|
6312 |
+
"epoch": 0.05927924075201079,
|
6313 |
+
"grad_norm": 0.5296594500541687,
|
6314 |
+
"learning_rate": 3.294931161568914e-05,
|
6315 |
+
"loss": 1.7877,
|
6316 |
+
"step": 901
|
6317 |
+
},
|
6318 |
+
{
|
6319 |
+
"epoch": 0.05934503347204632,
|
6320 |
+
"grad_norm": 0.5523350834846497,
|
6321 |
+
"learning_rate": 3.2848693658293675e-05,
|
6322 |
+
"loss": 1.6165,
|
6323 |
+
"step": 902
|
6324 |
+
},
|
6325 |
+
{
|
6326 |
+
"epoch": 0.05941082619208184,
|
6327 |
+
"grad_norm": 0.5559424161911011,
|
6328 |
+
"learning_rate": 3.274815435774921e-05,
|
6329 |
+
"loss": 1.6415,
|
6330 |
+
"step": 903
|
6331 |
+
},
|
6332 |
+
{
|
6333 |
+
"epoch": 0.05947661891211738,
|
6334 |
+
"grad_norm": 0.6036515831947327,
|
6335 |
+
"learning_rate": 3.264769417513463e-05,
|
6336 |
+
"loss": 1.6959,
|
6337 |
+
"step": 904
|
6338 |
+
},
|
6339 |
+
{
|
6340 |
+
"epoch": 0.0595424116321529,
|
6341 |
+
"grad_norm": 0.5790659189224243,
|
6342 |
+
"learning_rate": 3.254731357116597e-05,
|
6343 |
+
"loss": 1.6368,
|
6344 |
+
"step": 905
|
6345 |
+
},
|
6346 |
+
{
|
6347 |
+
"epoch": 0.05960820435218843,
|
6348 |
+
"grad_norm": 0.5860310792922974,
|
6349 |
+
"learning_rate": 3.244701300619427e-05,
|
6350 |
+
"loss": 1.5645,
|
6351 |
+
"step": 906
|
6352 |
+
},
|
6353 |
+
{
|
6354 |
+
"epoch": 0.059673997072223955,
|
6355 |
+
"grad_norm": 0.5953836441040039,
|
6356 |
+
"learning_rate": 3.2346792940203554e-05,
|
6357 |
+
"loss": 1.5427,
|
6358 |
+
"step": 907
|
6359 |
+
},
|
6360 |
+
{
|
6361 |
+
"epoch": 0.05973978979225949,
|
6362 |
+
"grad_norm": 0.574989378452301,
|
6363 |
+
"learning_rate": 3.224665383280867e-05,
|
6364 |
+
"loss": 1.7678,
|
6365 |
+
"step": 908
|
6366 |
+
},
|
6367 |
+
{
|
6368 |
+
"epoch": 0.059805582512295015,
|
6369 |
+
"grad_norm": 0.5947942733764648,
|
6370 |
+
"learning_rate": 3.214659614325313e-05,
|
6371 |
+
"loss": 1.5293,
|
6372 |
+
"step": 909
|
6373 |
+
},
|
6374 |
+
{
|
6375 |
+
"epoch": 0.05987137523233054,
|
6376 |
+
"grad_norm": 0.5816536545753479,
|
6377 |
+
"learning_rate": 3.2046620330407116e-05,
|
6378 |
+
"loss": 1.4343,
|
6379 |
+
"step": 910
|
6380 |
+
},
|
6381 |
+
{
|
6382 |
+
"epoch": 0.05993716795236607,
|
6383 |
+
"grad_norm": 0.5763691663742065,
|
6384 |
+
"learning_rate": 3.194672685276532e-05,
|
6385 |
+
"loss": 1.6881,
|
6386 |
+
"step": 911
|
6387 |
+
},
|
6388 |
+
{
|
6389 |
+
"epoch": 0.0600029606724016,
|
6390 |
+
"grad_norm": 0.629294753074646,
|
6391 |
+
"learning_rate": 3.1846916168444806e-05,
|
6392 |
+
"loss": 1.7855,
|
6393 |
+
"step": 912
|
6394 |
+
},
|
6395 |
+
{
|
6396 |
+
"epoch": 0.06006875339243713,
|
6397 |
+
"grad_norm": 0.6405010223388672,
|
6398 |
+
"learning_rate": 3.174718873518298e-05,
|
6399 |
+
"loss": 1.6608,
|
6400 |
+
"step": 913
|
6401 |
+
},
|
6402 |
+
{
|
6403 |
+
"epoch": 0.06013454611247265,
|
6404 |
+
"grad_norm": 0.5852246284484863,
|
6405 |
+
"learning_rate": 3.16475450103354e-05,
|
6406 |
+
"loss": 1.6286,
|
6407 |
+
"step": 914
|
6408 |
+
},
|
6409 |
+
{
|
6410 |
+
"epoch": 0.06020033883250818,
|
6411 |
+
"grad_norm": 0.6519684195518494,
|
6412 |
+
"learning_rate": 3.154798545087379e-05,
|
6413 |
+
"loss": 1.6609,
|
6414 |
+
"step": 915
|
6415 |
+
},
|
6416 |
+
{
|
6417 |
+
"epoch": 0.06026613155254371,
|
6418 |
+
"grad_norm": 0.6138909459114075,
|
6419 |
+
"learning_rate": 3.144851051338385e-05,
|
6420 |
+
"loss": 1.7577,
|
6421 |
+
"step": 916
|
6422 |
+
},
|
6423 |
+
{
|
6424 |
+
"epoch": 0.06033192427257924,
|
6425 |
+
"grad_norm": 0.5902788639068604,
|
6426 |
+
"learning_rate": 3.1349120654063225e-05,
|
6427 |
+
"loss": 1.622,
|
6428 |
+
"step": 917
|
6429 |
+
},
|
6430 |
+
{
|
6431 |
+
"epoch": 0.060397716992614765,
|
6432 |
+
"grad_norm": 0.6420292854309082,
|
6433 |
+
"learning_rate": 3.124981632871937e-05,
|
6434 |
+
"loss": 1.6161,
|
6435 |
+
"step": 918
|
6436 |
+
},
|
6437 |
+
{
|
6438 |
+
"epoch": 0.0604635097126503,
|
6439 |
+
"grad_norm": 0.6139567494392395,
|
6440 |
+
"learning_rate": 3.115059799276748e-05,
|
6441 |
+
"loss": 1.682,
|
6442 |
+
"step": 919
|
6443 |
+
},
|
6444 |
+
{
|
6445 |
+
"epoch": 0.060529302432685825,
|
6446 |
+
"grad_norm": 0.6504656672477722,
|
6447 |
+
"learning_rate": 3.1051466101228385e-05,
|
6448 |
+
"loss": 1.5961,
|
6449 |
+
"step": 920
|
6450 |
+
},
|
6451 |
+
{
|
6452 |
+
"epoch": 0.06059509515272135,
|
6453 |
+
"grad_norm": 0.6670247912406921,
|
6454 |
+
"learning_rate": 3.09524211087265e-05,
|
6455 |
+
"loss": 1.6288,
|
6456 |
+
"step": 921
|
6457 |
+
},
|
6458 |
+
{
|
6459 |
+
"epoch": 0.06066088787275688,
|
6460 |
+
"grad_norm": 0.6252246499061584,
|
6461 |
+
"learning_rate": 3.08534634694877e-05,
|
6462 |
+
"loss": 1.527,
|
6463 |
+
"step": 922
|
6464 |
+
},
|
6465 |
+
{
|
6466 |
+
"epoch": 0.06072668059279241,
|
6467 |
+
"grad_norm": 0.6155936121940613,
|
6468 |
+
"learning_rate": 3.0754593637337276e-05,
|
6469 |
+
"loss": 1.5781,
|
6470 |
+
"step": 923
|
6471 |
+
},
|
6472 |
+
{
|
6473 |
+
"epoch": 0.06079247331282794,
|
6474 |
+
"grad_norm": 0.6207473278045654,
|
6475 |
+
"learning_rate": 3.065581206569782e-05,
|
6476 |
+
"loss": 1.3617,
|
6477 |
+
"step": 924
|
6478 |
+
},
|
6479 |
+
{
|
6480 |
+
"epoch": 0.06085826603286346,
|
6481 |
+
"grad_norm": 0.6116368174552917,
|
6482 |
+
"learning_rate": 3.055711920758709e-05,
|
6483 |
+
"loss": 1.6183,
|
6484 |
+
"step": 925
|
6485 |
+
},
|
6486 |
+
{
|
6487 |
+
"epoch": 0.06092405875289899,
|
6488 |
+
"grad_norm": 0.646839439868927,
|
6489 |
+
"learning_rate": 3.0458515515616115e-05,
|
6490 |
+
"loss": 1.5567,
|
6491 |
+
"step": 926
|
6492 |
+
},
|
6493 |
+
{
|
6494 |
+
"epoch": 0.06098985147293452,
|
6495 |
+
"grad_norm": 0.6746354103088379,
|
6496 |
+
"learning_rate": 3.036000144198693e-05,
|
6497 |
+
"loss": 1.5467,
|
6498 |
+
"step": 927
|
6499 |
+
},
|
6500 |
+
{
|
6501 |
+
"epoch": 0.06105564419297005,
|
6502 |
+
"grad_norm": 0.639434278011322,
|
6503 |
+
"learning_rate": 3.0261577438490585e-05,
|
6504 |
+
"loss": 1.5425,
|
6505 |
+
"step": 928
|
6506 |
+
},
|
6507 |
+
{
|
6508 |
+
"epoch": 0.061121436913005575,
|
6509 |
+
"grad_norm": 0.685247540473938,
|
6510 |
+
"learning_rate": 3.0163243956505095e-05,
|
6511 |
+
"loss": 1.6157,
|
6512 |
+
"step": 929
|
6513 |
+
},
|
6514 |
+
{
|
6515 |
+
"epoch": 0.0611872296330411,
|
6516 |
+
"grad_norm": 0.6764075756072998,
|
6517 |
+
"learning_rate": 3.006500144699328e-05,
|
6518 |
+
"loss": 1.5882,
|
6519 |
+
"step": 930
|
6520 |
+
},
|
6521 |
+
{
|
6522 |
+
"epoch": 0.061253022353076635,
|
6523 |
+
"grad_norm": 0.6915985345840454,
|
6524 |
+
"learning_rate": 2.99668503605008e-05,
|
6525 |
+
"loss": 1.51,
|
6526 |
+
"step": 931
|
6527 |
+
},
|
6528 |
+
{
|
6529 |
+
"epoch": 0.06131881507311216,
|
6530 |
+
"grad_norm": 0.74942946434021,
|
6531 |
+
"learning_rate": 2.986879114715403e-05,
|
6532 |
+
"loss": 1.4684,
|
6533 |
+
"step": 932
|
6534 |
+
},
|
6535 |
+
{
|
6536 |
+
"epoch": 0.06138460779314769,
|
6537 |
+
"grad_norm": 0.6873716711997986,
|
6538 |
+
"learning_rate": 2.977082425665802e-05,
|
6539 |
+
"loss": 1.6268,
|
6540 |
+
"step": 933
|
6541 |
+
},
|
6542 |
+
{
|
6543 |
+
"epoch": 0.06145040051318321,
|
6544 |
+
"grad_norm": 0.7743644714355469,
|
6545 |
+
"learning_rate": 2.967295013829442e-05,
|
6546 |
+
"loss": 1.564,
|
6547 |
+
"step": 934
|
6548 |
+
},
|
6549 |
+
{
|
6550 |
+
"epoch": 0.06151619323321875,
|
6551 |
+
"grad_norm": 0.702805757522583,
|
6552 |
+
"learning_rate": 2.95751692409194e-05,
|
6553 |
+
"loss": 1.7741,
|
6554 |
+
"step": 935
|
6555 |
+
},
|
6556 |
+
{
|
6557 |
+
"epoch": 0.06158198595325427,
|
6558 |
+
"grad_norm": 0.7666968107223511,
|
6559 |
+
"learning_rate": 2.947748201296163e-05,
|
6560 |
+
"loss": 1.6298,
|
6561 |
+
"step": 936
|
6562 |
+
},
|
6563 |
+
{
|
6564 |
+
"epoch": 0.0616477786732898,
|
6565 |
+
"grad_norm": 0.700221061706543,
|
6566 |
+
"learning_rate": 2.9379888902420215e-05,
|
6567 |
+
"loss": 1.5275,
|
6568 |
+
"step": 937
|
6569 |
+
},
|
6570 |
+
{
|
6571 |
+
"epoch": 0.061713571393325325,
|
6572 |
+
"grad_norm": 0.7477214932441711,
|
6573 |
+
"learning_rate": 2.9282390356862606e-05,
|
6574 |
+
"loss": 1.5412,
|
6575 |
+
"step": 938
|
6576 |
+
},
|
6577 |
+
{
|
6578 |
+
"epoch": 0.06177936411336086,
|
6579 |
+
"grad_norm": 0.7472383975982666,
|
6580 |
+
"learning_rate": 2.9184986823422623e-05,
|
6581 |
+
"loss": 1.547,
|
6582 |
+
"step": 939
|
6583 |
+
},
|
6584 |
+
{
|
6585 |
+
"epoch": 0.061845156833396385,
|
6586 |
+
"grad_norm": 0.7094493508338928,
|
6587 |
+
"learning_rate": 2.9087678748798297e-05,
|
6588 |
+
"loss": 1.4617,
|
6589 |
+
"step": 940
|
6590 |
+
},
|
6591 |
+
{
|
6592 |
+
"epoch": 0.06191094955343191,
|
6593 |
+
"grad_norm": 0.7252383828163147,
|
6594 |
+
"learning_rate": 2.899046657924992e-05,
|
6595 |
+
"loss": 1.3699,
|
6596 |
+
"step": 941
|
6597 |
+
},
|
6598 |
+
{
|
6599 |
+
"epoch": 0.06197674227346744,
|
6600 |
+
"grad_norm": 0.7631134390830994,
|
6601 |
+
"learning_rate": 2.8893350760597924e-05,
|
6602 |
+
"loss": 1.567,
|
6603 |
+
"step": 942
|
6604 |
+
},
|
6605 |
+
{
|
6606 |
+
"epoch": 0.06204253499350297,
|
6607 |
+
"grad_norm": 0.7023096680641174,
|
6608 |
+
"learning_rate": 2.8796331738220912e-05,
|
6609 |
+
"loss": 1.4324,
|
6610 |
+
"step": 943
|
6611 |
+
},
|
6612 |
+
{
|
6613 |
+
"epoch": 0.0621083277135385,
|
6614 |
+
"grad_norm": 0.7988854050636292,
|
6615 |
+
"learning_rate": 2.8699409957053535e-05,
|
6616 |
+
"loss": 1.6788,
|
6617 |
+
"step": 944
|
6618 |
+
},
|
6619 |
+
{
|
6620 |
+
"epoch": 0.06217412043357402,
|
6621 |
+
"grad_norm": 0.8351010084152222,
|
6622 |
+
"learning_rate": 2.860258586158454e-05,
|
6623 |
+
"loss": 1.6451,
|
6624 |
+
"step": 945
|
6625 |
+
},
|
6626 |
+
{
|
6627 |
+
"epoch": 0.062239913153609557,
|
6628 |
+
"grad_norm": 0.8499585390090942,
|
6629 |
+
"learning_rate": 2.8505859895854604e-05,
|
6630 |
+
"loss": 1.6087,
|
6631 |
+
"step": 946
|
6632 |
+
},
|
6633 |
+
{
|
6634 |
+
"epoch": 0.06230570587364508,
|
6635 |
+
"grad_norm": 0.9862839579582214,
|
6636 |
+
"learning_rate": 2.840923250345442e-05,
|
6637 |
+
"loss": 1.5487,
|
6638 |
+
"step": 947
|
6639 |
+
},
|
6640 |
+
{
|
6641 |
+
"epoch": 0.06237149859368061,
|
6642 |
+
"grad_norm": 0.9058715105056763,
|
6643 |
+
"learning_rate": 2.8312704127522655e-05,
|
6644 |
+
"loss": 1.3746,
|
6645 |
+
"step": 948
|
6646 |
+
},
|
6647 |
+
{
|
6648 |
+
"epoch": 0.062437291313716135,
|
6649 |
+
"grad_norm": 0.9685963988304138,
|
6650 |
+
"learning_rate": 2.821627521074383e-05,
|
6651 |
+
"loss": 1.4866,
|
6652 |
+
"step": 949
|
6653 |
+
},
|
6654 |
+
{
|
6655 |
+
"epoch": 0.06250308403375167,
|
6656 |
+
"grad_norm": 1.311897873878479,
|
6657 |
+
"learning_rate": 2.8119946195346375e-05,
|
6658 |
+
"loss": 1.5507,
|
6659 |
+
"step": 950
|
6660 |
+
},
|
6661 |
+
{
|
6662 |
+
"epoch": 0.0625688767537872,
|
6663 |
+
"grad_norm": 0.5097134113311768,
|
6664 |
+
"learning_rate": 2.8023717523100508e-05,
|
6665 |
+
"loss": 1.8342,
|
6666 |
+
"step": 951
|
6667 |
+
},
|
6668 |
+
{
|
6669 |
+
"epoch": 0.06263466947382272,
|
6670 |
+
"grad_norm": 0.5375131368637085,
|
6671 |
+
"learning_rate": 2.792758963531632e-05,
|
6672 |
+
"loss": 1.6841,
|
6673 |
+
"step": 952
|
6674 |
+
},
|
6675 |
+
{
|
6676 |
+
"epoch": 0.06270046219385825,
|
6677 |
+
"grad_norm": 0.6458749771118164,
|
6678 |
+
"learning_rate": 2.7831562972841696e-05,
|
6679 |
+
"loss": 2.0434,
|
6680 |
+
"step": 953
|
6681 |
+
},
|
6682 |
+
{
|
6683 |
+
"epoch": 0.06276625491389377,
|
6684 |
+
"grad_norm": 0.5497531890869141,
|
6685 |
+
"learning_rate": 2.773563797606028e-05,
|
6686 |
+
"loss": 1.6057,
|
6687 |
+
"step": 954
|
6688 |
+
},
|
6689 |
+
{
|
6690 |
+
"epoch": 0.0628320476339293,
|
6691 |
+
"grad_norm": 0.5604535341262817,
|
6692 |
+
"learning_rate": 2.7639815084889476e-05,
|
6693 |
+
"loss": 1.6338,
|
6694 |
+
"step": 955
|
6695 |
+
},
|
6696 |
+
{
|
6697 |
+
"epoch": 0.06289784035396484,
|
6698 |
+
"grad_norm": 0.6276973485946655,
|
6699 |
+
"learning_rate": 2.7544094738778436e-05,
|
6700 |
+
"loss": 1.7273,
|
6701 |
+
"step": 956
|
6702 |
+
},
|
6703 |
+
{
|
6704 |
+
"epoch": 0.06296363307400037,
|
6705 |
+
"grad_norm": 0.6032599210739136,
|
6706 |
+
"learning_rate": 2.744847737670601e-05,
|
6707 |
+
"loss": 1.6246,
|
6708 |
+
"step": 957
|
6709 |
+
},
|
6710 |
+
{
|
6711 |
+
"epoch": 0.06302942579403589,
|
6712 |
+
"grad_norm": 0.6191315650939941,
|
6713 |
+
"learning_rate": 2.7352963437178786e-05,
|
6714 |
+
"loss": 1.6568,
|
6715 |
+
"step": 958
|
6716 |
+
},
|
6717 |
+
{
|
6718 |
+
"epoch": 0.06309521851407142,
|
6719 |
+
"grad_norm": 0.5911609530448914,
|
6720 |
+
"learning_rate": 2.7257553358229034e-05,
|
6721 |
+
"loss": 1.6472,
|
6722 |
+
"step": 959
|
6723 |
+
},
|
6724 |
+
{
|
6725 |
+
"epoch": 0.06316101123410695,
|
6726 |
+
"grad_norm": 0.5993046164512634,
|
6727 |
+
"learning_rate": 2.7162247577412715e-05,
|
6728 |
+
"loss": 1.5908,
|
6729 |
+
"step": 960
|
6730 |
+
},
|
6731 |
+
{
|
6732 |
+
"epoch": 0.06322680395414247,
|
6733 |
+
"grad_norm": 0.6404312252998352,
|
6734 |
+
"learning_rate": 2.7067046531807494e-05,
|
6735 |
+
"loss": 1.6952,
|
6736 |
+
"step": 961
|
6737 |
+
},
|
6738 |
+
{
|
6739 |
+
"epoch": 0.063292596674178,
|
6740 |
+
"grad_norm": 0.6177670359611511,
|
6741 |
+
"learning_rate": 2.6971950658010666e-05,
|
6742 |
+
"loss": 1.5803,
|
6743 |
+
"step": 962
|
6744 |
+
},
|
6745 |
+
{
|
6746 |
+
"epoch": 0.06335838939421352,
|
6747 |
+
"grad_norm": 0.6164083480834961,
|
6748 |
+
"learning_rate": 2.6876960392137217e-05,
|
6749 |
+
"loss": 1.5917,
|
6750 |
+
"step": 963
|
6751 |
+
},
|
6752 |
+
{
|
6753 |
+
"epoch": 0.06342418211424906,
|
6754 |
+
"grad_norm": 0.6648538708686829,
|
6755 |
+
"learning_rate": 2.678207616981787e-05,
|
6756 |
+
"loss": 1.6052,
|
6757 |
+
"step": 964
|
6758 |
+
},
|
6759 |
+
{
|
6760 |
+
"epoch": 0.06348997483428459,
|
6761 |
+
"grad_norm": 0.6657065153121948,
|
6762 |
+
"learning_rate": 2.6687298426196973e-05,
|
6763 |
+
"loss": 1.7399,
|
6764 |
+
"step": 965
|
6765 |
+
},
|
6766 |
+
{
|
6767 |
+
"epoch": 0.06355576755432012,
|
6768 |
+
"grad_norm": 0.6626632213592529,
|
6769 |
+
"learning_rate": 2.6592627595930542e-05,
|
6770 |
+
"loss": 1.6635,
|
6771 |
+
"step": 966
|
6772 |
+
},
|
6773 |
+
{
|
6774 |
+
"epoch": 0.06362156027435564,
|
6775 |
+
"grad_norm": 0.6235553622245789,
|
6776 |
+
"learning_rate": 2.6498064113184338e-05,
|
6777 |
+
"loss": 1.6467,
|
6778 |
+
"step": 967
|
6779 |
+
},
|
6780 |
+
{
|
6781 |
+
"epoch": 0.06368735299439117,
|
6782 |
+
"grad_norm": 0.6630668044090271,
|
6783 |
+
"learning_rate": 2.6403608411631742e-05,
|
6784 |
+
"loss": 1.6262,
|
6785 |
+
"step": 968
|
6786 |
+
},
|
6787 |
+
{
|
6788 |
+
"epoch": 0.0637531457144267,
|
6789 |
+
"grad_norm": 0.6901178956031799,
|
6790 |
+
"learning_rate": 2.6309260924451907e-05,
|
6791 |
+
"loss": 1.687,
|
6792 |
+
"step": 969
|
6793 |
+
},
|
6794 |
+
{
|
6795 |
+
"epoch": 0.06381893843446222,
|
6796 |
+
"grad_norm": 0.6554874777793884,
|
6797 |
+
"learning_rate": 2.62150220843277e-05,
|
6798 |
+
"loss": 1.7402,
|
6799 |
+
"step": 970
|
6800 |
+
},
|
6801 |
+
{
|
6802 |
+
"epoch": 0.06388473115449775,
|
6803 |
+
"grad_norm": 0.6539959907531738,
|
6804 |
+
"learning_rate": 2.612089232344371e-05,
|
6805 |
+
"loss": 1.7455,
|
6806 |
+
"step": 971
|
6807 |
+
},
|
6808 |
+
{
|
6809 |
+
"epoch": 0.06395052387453329,
|
6810 |
+
"grad_norm": 0.6740013360977173,
|
6811 |
+
"learning_rate": 2.602687207348429e-05,
|
6812 |
+
"loss": 1.6105,
|
6813 |
+
"step": 972
|
6814 |
+
},
|
6815 |
+
{
|
6816 |
+
"epoch": 0.06401631659456881,
|
6817 |
+
"grad_norm": 0.6463841199874878,
|
6818 |
+
"learning_rate": 2.593296176563157e-05,
|
6819 |
+
"loss": 1.5821,
|
6820 |
+
"step": 973
|
6821 |
+
},
|
6822 |
+
{
|
6823 |
+
"epoch": 0.06408210931460434,
|
6824 |
+
"grad_norm": 0.6570295691490173,
|
6825 |
+
"learning_rate": 2.5839161830563474e-05,
|
6826 |
+
"loss": 1.6883,
|
6827 |
+
"step": 974
|
6828 |
+
},
|
6829 |
+
{
|
6830 |
+
"epoch": 0.06414790203463987,
|
6831 |
+
"grad_norm": 0.6713987588882446,
|
6832 |
+
"learning_rate": 2.5745472698451767e-05,
|
6833 |
+
"loss": 1.5053,
|
6834 |
+
"step": 975
|
6835 |
+
},
|
6836 |
+
{
|
6837 |
+
"epoch": 0.0642136947546754,
|
6838 |
+
"grad_norm": 0.6906765103340149,
|
6839 |
+
"learning_rate": 2.565189479896005e-05,
|
6840 |
+
"loss": 1.5105,
|
6841 |
+
"step": 976
|
6842 |
+
},
|
6843 |
+
{
|
6844 |
+
"epoch": 0.06427948747471092,
|
6845 |
+
"grad_norm": 0.6981975436210632,
|
6846 |
+
"learning_rate": 2.555842856124182e-05,
|
6847 |
+
"loss": 1.5976,
|
6848 |
+
"step": 977
|
6849 |
+
},
|
6850 |
+
{
|
6851 |
+
"epoch": 0.06434528019474645,
|
6852 |
+
"grad_norm": 0.7114136815071106,
|
6853 |
+
"learning_rate": 2.546507441393845e-05,
|
6854 |
+
"loss": 1.6393,
|
6855 |
+
"step": 978
|
6856 |
+
},
|
6857 |
+
{
|
6858 |
+
"epoch": 0.06441107291478199,
|
6859 |
+
"grad_norm": 0.6741265058517456,
|
6860 |
+
"learning_rate": 2.5371832785177273e-05,
|
6861 |
+
"loss": 1.6601,
|
6862 |
+
"step": 979
|
6863 |
+
},
|
6864 |
+
{
|
6865 |
+
"epoch": 0.06447686563481751,
|
6866 |
+
"grad_norm": 0.7347429990768433,
|
6867 |
+
"learning_rate": 2.5278704102569662e-05,
|
6868 |
+
"loss": 1.6784,
|
6869 |
+
"step": 980
|
6870 |
+
},
|
6871 |
+
{
|
6872 |
+
"epoch": 0.06454265835485304,
|
6873 |
+
"grad_norm": 0.6604593992233276,
|
6874 |
+
"learning_rate": 2.518568879320895e-05,
|
6875 |
+
"loss": 1.5858,
|
6876 |
+
"step": 981
|
6877 |
+
},
|
6878 |
+
{
|
6879 |
+
"epoch": 0.06460845107488856,
|
6880 |
+
"grad_norm": 0.6967578530311584,
|
6881 |
+
"learning_rate": 2.509278728366855e-05,
|
6882 |
+
"loss": 1.5591,
|
6883 |
+
"step": 982
|
6884 |
+
},
|
6885 |
+
{
|
6886 |
+
"epoch": 0.06467424379492409,
|
6887 |
+
"grad_norm": 0.744243323802948,
|
6888 |
+
"learning_rate": 2.500000000000001e-05,
|
6889 |
+
"loss": 1.5185,
|
6890 |
+
"step": 983
|
6891 |
+
},
|
6892 |
+
{
|
6893 |
+
"epoch": 0.06474003651495962,
|
6894 |
+
"grad_norm": 0.7131791114807129,
|
6895 |
+
"learning_rate": 2.4907327367730963e-05,
|
6896 |
+
"loss": 1.5966,
|
6897 |
+
"step": 984
|
6898 |
+
},
|
6899 |
+
{
|
6900 |
+
"epoch": 0.06480582923499514,
|
6901 |
+
"grad_norm": 0.745911180973053,
|
6902 |
+
"learning_rate": 2.4814769811863313e-05,
|
6903 |
+
"loss": 1.6642,
|
6904 |
+
"step": 985
|
6905 |
+
},
|
6906 |
+
{
|
6907 |
+
"epoch": 0.06487162195503067,
|
6908 |
+
"grad_norm": 0.7129960656166077,
|
6909 |
+
"learning_rate": 2.472232775687119e-05,
|
6910 |
+
"loss": 1.4803,
|
6911 |
+
"step": 986
|
6912 |
+
},
|
6913 |
+
{
|
6914 |
+
"epoch": 0.06493741467506621,
|
6915 |
+
"grad_norm": 0.7604677677154541,
|
6916 |
+
"learning_rate": 2.463000162669903e-05,
|
6917 |
+
"loss": 1.4905,
|
6918 |
+
"step": 987
|
6919 |
+
},
|
6920 |
+
{
|
6921 |
+
"epoch": 0.06500320739510174,
|
6922 |
+
"grad_norm": 0.7828547358512878,
|
6923 |
+
"learning_rate": 2.453779184475964e-05,
|
6924 |
+
"loss": 1.5679,
|
6925 |
+
"step": 988
|
6926 |
+
},
|
6927 |
+
{
|
6928 |
+
"epoch": 0.06506900011513726,
|
6929 |
+
"grad_norm": 0.7184737324714661,
|
6930 |
+
"learning_rate": 2.4445698833932234e-05,
|
6931 |
+
"loss": 1.4162,
|
6932 |
+
"step": 989
|
6933 |
+
},
|
6934 |
+
{
|
6935 |
+
"epoch": 0.06513479283517279,
|
6936 |
+
"grad_norm": 0.7497789263725281,
|
6937 |
+
"learning_rate": 2.4353723016560527e-05,
|
6938 |
+
"loss": 1.5783,
|
6939 |
+
"step": 990
|
6940 |
+
},
|
6941 |
+
{
|
6942 |
+
"epoch": 0.06520058555520832,
|
6943 |
+
"grad_norm": 0.7946657538414001,
|
6944 |
+
"learning_rate": 2.4261864814450758e-05,
|
6945 |
+
"loss": 1.5937,
|
6946 |
+
"step": 991
|
6947 |
+
},
|
6948 |
+
{
|
6949 |
+
"epoch": 0.06526637827524384,
|
6950 |
+
"grad_norm": 0.7872624397277832,
|
6951 |
+
"learning_rate": 2.417012464886978e-05,
|
6952 |
+
"loss": 1.5904,
|
6953 |
+
"step": 992
|
6954 |
+
},
|
6955 |
+
{
|
6956 |
+
"epoch": 0.06533217099527937,
|
6957 |
+
"grad_norm": 0.8094744682312012,
|
6958 |
+
"learning_rate": 2.4078502940543157e-05,
|
6959 |
+
"loss": 1.5454,
|
6960 |
+
"step": 993
|
6961 |
+
},
|
6962 |
+
{
|
6963 |
+
"epoch": 0.0653979637153149,
|
6964 |
+
"grad_norm": 0.9983581304550171,
|
6965 |
+
"learning_rate": 2.3987000109653134e-05,
|
6966 |
+
"loss": 1.4996,
|
6967 |
+
"step": 994
|
6968 |
+
},
|
6969 |
+
{
|
6970 |
+
"epoch": 0.06546375643535043,
|
6971 |
+
"grad_norm": 0.826250433921814,
|
6972 |
+
"learning_rate": 2.389561657583681e-05,
|
6973 |
+
"loss": 1.5753,
|
6974 |
+
"step": 995
|
6975 |
+
},
|
6976 |
+
{
|
6977 |
+
"epoch": 0.06552954915538596,
|
6978 |
+
"grad_norm": 0.8517383337020874,
|
6979 |
+
"learning_rate": 2.3804352758184223e-05,
|
6980 |
+
"loss": 1.5688,
|
6981 |
+
"step": 996
|
6982 |
+
},
|
6983 |
+
{
|
6984 |
+
"epoch": 0.06559534187542149,
|
6985 |
+
"grad_norm": 0.8434426784515381,
|
6986 |
+
"learning_rate": 2.3713209075236343e-05,
|
6987 |
+
"loss": 1.3069,
|
6988 |
+
"step": 997
|
6989 |
+
},
|
6990 |
+
{
|
6991 |
+
"epoch": 0.06566113459545701,
|
6992 |
+
"grad_norm": 0.9146838784217834,
|
6993 |
+
"learning_rate": 2.3622185944983188e-05,
|
6994 |
+
"loss": 1.4057,
|
6995 |
+
"step": 998
|
6996 |
+
},
|
6997 |
+
{
|
6998 |
+
"epoch": 0.06572692731549254,
|
6999 |
+
"grad_norm": 1.0109113454818726,
|
7000 |
+
"learning_rate": 2.3531283784861952e-05,
|
7001 |
+
"loss": 1.6269,
|
7002 |
+
"step": 999
|
7003 |
+
},
|
7004 |
+
{
|
7005 |
+
"epoch": 0.06579272003552807,
|
7006 |
+
"grad_norm": 1.2136893272399902,
|
7007 |
+
"learning_rate": 2.3440503011755e-05,
|
7008 |
+
"loss": 1.5136,
|
7009 |
+
"step": 1000
|
7010 |
+
},
|
7011 |
+
{
|
7012 |
+
"epoch": 0.06585851275556359,
|
7013 |
+
"grad_norm": 0.5091771483421326,
|
7014 |
+
"learning_rate": 2.3349844041988045e-05,
|
7015 |
+
"loss": 1.7648,
|
7016 |
+
"step": 1001
|
7017 |
+
},
|
7018 |
+
{
|
7019 |
+
"epoch": 0.06592430547559912,
|
7020 |
+
"grad_norm": 0.5724332928657532,
|
7021 |
+
"learning_rate": 2.325930729132821e-05,
|
7022 |
+
"loss": 1.8152,
|
7023 |
+
"step": 1002
|
7024 |
+
},
|
7025 |
+
{
|
7026 |
+
"epoch": 0.06599009819563466,
|
7027 |
+
"grad_norm": 0.5698429942131042,
|
7028 |
+
"learning_rate": 2.3168893174982098e-05,
|
7029 |
+
"loss": 1.6787,
|
7030 |
+
"step": 1003
|
7031 |
+
},
|
7032 |
+
{
|
7033 |
+
"epoch": 0.06605589091567018,
|
7034 |
+
"grad_norm": 0.5834808945655823,
|
7035 |
+
"learning_rate": 2.30786021075939e-05,
|
7036 |
+
"loss": 1.6803,
|
7037 |
+
"step": 1004
|
7038 |
+
},
|
7039 |
+
{
|
7040 |
+
"epoch": 0.06612168363570571,
|
7041 |
+
"grad_norm": 0.588674008846283,
|
7042 |
+
"learning_rate": 2.2988434503243516e-05,
|
7043 |
+
"loss": 1.5799,
|
7044 |
+
"step": 1005
|
7045 |
+
},
|
7046 |
+
{
|
7047 |
+
"epoch": 0.06618747635574124,
|
7048 |
+
"grad_norm": 0.6044074296951294,
|
7049 |
+
"learning_rate": 2.289839077544463e-05,
|
7050 |
+
"loss": 1.7156,
|
7051 |
+
"step": 1006
|
7052 |
+
},
|
7053 |
+
{
|
7054 |
+
"epoch": 0.06625326907577676,
|
7055 |
+
"grad_norm": 0.6402976512908936,
|
7056 |
+
"learning_rate": 2.28084713371428e-05,
|
7057 |
+
"loss": 1.7041,
|
7058 |
+
"step": 1007
|
7059 |
+
},
|
7060 |
+
{
|
7061 |
+
"epoch": 0.06631906179581229,
|
7062 |
+
"grad_norm": 0.5715565085411072,
|
7063 |
+
"learning_rate": 2.2718676600713622e-05,
|
7064 |
+
"loss": 1.5912,
|
7065 |
+
"step": 1008
|
7066 |
+
},
|
7067 |
+
{
|
7068 |
+
"epoch": 0.06638485451584782,
|
7069 |
+
"grad_norm": 0.6274817585945129,
|
7070 |
+
"learning_rate": 2.2629006977960798e-05,
|
7071 |
+
"loss": 1.6121,
|
7072 |
+
"step": 1009
|
7073 |
+
},
|
7074 |
+
{
|
7075 |
+
"epoch": 0.06645064723588336,
|
7076 |
+
"grad_norm": 0.5526905655860901,
|
7077 |
+
"learning_rate": 2.2539462880114194e-05,
|
7078 |
+
"loss": 1.5085,
|
7079 |
+
"step": 1010
|
7080 |
+
},
|
7081 |
+
{
|
7082 |
+
"epoch": 0.06651643995591888,
|
7083 |
+
"grad_norm": 0.6097638010978699,
|
7084 |
+
"learning_rate": 2.245004471782806e-05,
|
7085 |
+
"loss": 1.5009,
|
7086 |
+
"step": 1011
|
7087 |
+
},
|
7088 |
+
{
|
7089 |
+
"epoch": 0.06658223267595441,
|
7090 |
+
"grad_norm": 0.6747389435768127,
|
7091 |
+
"learning_rate": 2.236075290117914e-05,
|
7092 |
+
"loss": 1.6578,
|
7093 |
+
"step": 1012
|
7094 |
+
},
|
7095 |
+
{
|
7096 |
+
"epoch": 0.06664802539598993,
|
7097 |
+
"grad_norm": 0.6369504928588867,
|
7098 |
+
"learning_rate": 2.2271587839664672e-05,
|
7099 |
+
"loss": 1.5528,
|
7100 |
+
"step": 1013
|
7101 |
+
},
|
7102 |
+
{
|
7103 |
+
"epoch": 0.06671381811602546,
|
7104 |
+
"grad_norm": 0.6106559038162231,
|
7105 |
+
"learning_rate": 2.218254994220062e-05,
|
7106 |
+
"loss": 1.5907,
|
7107 |
+
"step": 1014
|
7108 |
+
},
|
7109 |
+
{
|
7110 |
+
"epoch": 0.06677961083606099,
|
7111 |
+
"grad_norm": 0.5885758399963379,
|
7112 |
+
"learning_rate": 2.2093639617119794e-05,
|
7113 |
+
"loss": 1.513,
|
7114 |
+
"step": 1015
|
7115 |
+
},
|
7116 |
+
{
|
7117 |
+
"epoch": 0.06684540355609651,
|
7118 |
+
"grad_norm": 0.6991572380065918,
|
7119 |
+
"learning_rate": 2.2004857272169876e-05,
|
7120 |
+
"loss": 1.823,
|
7121 |
+
"step": 1016
|
7122 |
+
},
|
7123 |
+
{
|
7124 |
+
"epoch": 0.06691119627613204,
|
7125 |
+
"grad_norm": 0.6160642504692078,
|
7126 |
+
"learning_rate": 2.1916203314511692e-05,
|
7127 |
+
"loss": 1.6389,
|
7128 |
+
"step": 1017
|
7129 |
+
},
|
7130 |
+
{
|
7131 |
+
"epoch": 0.06697698899616758,
|
7132 |
+
"grad_norm": 0.6292829513549805,
|
7133 |
+
"learning_rate": 2.1827678150717256e-05,
|
7134 |
+
"loss": 1.6663,
|
7135 |
+
"step": 1018
|
7136 |
+
},
|
7137 |
+
{
|
7138 |
+
"epoch": 0.0670427817162031,
|
7139 |
+
"grad_norm": 0.6121276617050171,
|
7140 |
+
"learning_rate": 2.1739282186767923e-05,
|
7141 |
+
"loss": 1.5885,
|
7142 |
+
"step": 1019
|
7143 |
+
},
|
7144 |
+
{
|
7145 |
+
"epoch": 0.06710857443623863,
|
7146 |
+
"grad_norm": 0.6440593004226685,
|
7147 |
+
"learning_rate": 2.165101582805254e-05,
|
7148 |
+
"loss": 1.6319,
|
7149 |
+
"step": 1020
|
7150 |
+
},
|
7151 |
+
{
|
7152 |
+
"epoch": 0.06717436715627416,
|
7153 |
+
"grad_norm": 0.636544942855835,
|
7154 |
+
"learning_rate": 2.1562879479365556e-05,
|
7155 |
+
"loss": 1.644,
|
7156 |
+
"step": 1021
|
7157 |
+
},
|
7158 |
+
{
|
7159 |
+
"epoch": 0.06724015987630969,
|
7160 |
+
"grad_norm": 0.6443448662757874,
|
7161 |
+
"learning_rate": 2.1474873544905205e-05,
|
7162 |
+
"loss": 1.5995,
|
7163 |
+
"step": 1022
|
7164 |
+
},
|
7165 |
+
{
|
7166 |
+
"epoch": 0.06730595259634521,
|
7167 |
+
"grad_norm": 0.6514659523963928,
|
7168 |
+
"learning_rate": 2.1386998428271633e-05,
|
7169 |
+
"loss": 1.554,
|
7170 |
+
"step": 1023
|
7171 |
+
},
|
7172 |
+
{
|
7173 |
+
"epoch": 0.06737174531638074,
|
7174 |
+
"grad_norm": 0.6635521650314331,
|
7175 |
+
"learning_rate": 2.129925453246504e-05,
|
7176 |
+
"loss": 1.7461,
|
7177 |
+
"step": 1024
|
7178 |
+
},
|
7179 |
+
{
|
7180 |
+
"epoch": 0.06743753803641626,
|
7181 |
+
"grad_norm": 0.6211679577827454,
|
7182 |
+
"learning_rate": 2.1211642259883867e-05,
|
7183 |
+
"loss": 1.4678,
|
7184 |
+
"step": 1025
|
7185 |
+
},
|
7186 |
+
{
|
7187 |
+
"epoch": 0.0675033307564518,
|
7188 |
+
"grad_norm": 0.668057918548584,
|
7189 |
+
"learning_rate": 2.1124162012322862e-05,
|
7190 |
+
"loss": 1.6624,
|
7191 |
+
"step": 1026
|
7192 |
+
},
|
7193 |
+
{
|
7194 |
+
"epoch": 0.06756912347648733,
|
7195 |
+
"grad_norm": 0.6626843810081482,
|
7196 |
+
"learning_rate": 2.1036814190971333e-05,
|
7197 |
+
"loss": 1.5244,
|
7198 |
+
"step": 1027
|
7199 |
+
},
|
7200 |
+
{
|
7201 |
+
"epoch": 0.06763491619652286,
|
7202 |
+
"grad_norm": 0.6613166332244873,
|
7203 |
+
"learning_rate": 2.0949599196411325e-05,
|
7204 |
+
"loss": 1.497,
|
7205 |
+
"step": 1028
|
7206 |
+
},
|
7207 |
+
{
|
7208 |
+
"epoch": 0.06770070891655838,
|
7209 |
+
"grad_norm": 0.6597977876663208,
|
7210 |
+
"learning_rate": 2.086251742861565e-05,
|
7211 |
+
"loss": 1.711,
|
7212 |
+
"step": 1029
|
7213 |
+
},
|
7214 |
+
{
|
7215 |
+
"epoch": 0.06776650163659391,
|
7216 |
+
"grad_norm": 0.7360920310020447,
|
7217 |
+
"learning_rate": 2.077556928694619e-05,
|
7218 |
+
"loss": 1.6281,
|
7219 |
+
"step": 1030
|
7220 |
+
},
|
7221 |
+
{
|
7222 |
+
"epoch": 0.06783229435662944,
|
7223 |
+
"grad_norm": 0.6604809761047363,
|
7224 |
+
"learning_rate": 2.0688755170151996e-05,
|
7225 |
+
"loss": 1.6061,
|
7226 |
+
"step": 1031
|
7227 |
+
},
|
7228 |
+
{
|
7229 |
+
"epoch": 0.06789808707666496,
|
7230 |
+
"grad_norm": 0.6945058703422546,
|
7231 |
+
"learning_rate": 2.0602075476367432e-05,
|
7232 |
+
"loss": 1.6064,
|
7233 |
+
"step": 1032
|
7234 |
+
},
|
7235 |
+
{
|
7236 |
+
"epoch": 0.06796387979670049,
|
7237 |
+
"grad_norm": 0.71868896484375,
|
7238 |
+
"learning_rate": 2.051553060311045e-05,
|
7239 |
+
"loss": 1.56,
|
7240 |
+
"step": 1033
|
7241 |
+
},
|
7242 |
+
{
|
7243 |
+
"epoch": 0.06802967251673603,
|
7244 |
+
"grad_norm": 0.6665169596672058,
|
7245 |
+
"learning_rate": 2.0429120947280678e-05,
|
7246 |
+
"loss": 1.5953,
|
7247 |
+
"step": 1034
|
7248 |
+
},
|
7249 |
+
{
|
7250 |
+
"epoch": 0.06809546523677155,
|
7251 |
+
"grad_norm": 0.6772186756134033,
|
7252 |
+
"learning_rate": 2.0342846905157636e-05,
|
7253 |
+
"loss": 1.5658,
|
7254 |
+
"step": 1035
|
7255 |
+
},
|
7256 |
+
{
|
7257 |
+
"epoch": 0.06816125795680708,
|
7258 |
+
"grad_norm": 0.7677963376045227,
|
7259 |
+
"learning_rate": 2.0256708872398915e-05,
|
7260 |
+
"loss": 1.7299,
|
7261 |
+
"step": 1036
|
7262 |
+
},
|
7263 |
+
{
|
7264 |
+
"epoch": 0.06822705067684261,
|
7265 |
+
"grad_norm": 0.7391518950462341,
|
7266 |
+
"learning_rate": 2.017070724403835e-05,
|
7267 |
+
"loss": 1.6122,
|
7268 |
+
"step": 1037
|
7269 |
+
},
|
7270 |
+
{
|
7271 |
+
"epoch": 0.06829284339687813,
|
7272 |
+
"grad_norm": 0.8133260607719421,
|
7273 |
+
"learning_rate": 2.0084842414484222e-05,
|
7274 |
+
"loss": 1.5689,
|
7275 |
+
"step": 1038
|
7276 |
+
},
|
7277 |
+
{
|
7278 |
+
"epoch": 0.06835863611691366,
|
7279 |
+
"grad_norm": 0.7881767749786377,
|
7280 |
+
"learning_rate": 1.999911477751746e-05,
|
7281 |
+
"loss": 1.673,
|
7282 |
+
"step": 1039
|
7283 |
+
},
|
7284 |
+
{
|
7285 |
+
"epoch": 0.06842442883694919,
|
7286 |
+
"grad_norm": 0.7791821360588074,
|
7287 |
+
"learning_rate": 1.9913524726289784e-05,
|
7288 |
+
"loss": 1.7744,
|
7289 |
+
"step": 1040
|
7290 |
+
},
|
7291 |
+
{
|
7292 |
+
"epoch": 0.06849022155698473,
|
7293 |
+
"grad_norm": 0.7703235745429993,
|
7294 |
+
"learning_rate": 1.9828072653321978e-05,
|
7295 |
+
"loss": 1.4325,
|
7296 |
+
"step": 1041
|
7297 |
+
},
|
7298 |
+
{
|
7299 |
+
"epoch": 0.06855601427702025,
|
7300 |
+
"grad_norm": 0.7927573323249817,
|
7301 |
+
"learning_rate": 1.9742758950502045e-05,
|
7302 |
+
"loss": 1.6792,
|
7303 |
+
"step": 1042
|
7304 |
+
},
|
7305 |
+
{
|
7306 |
+
"epoch": 0.06862180699705578,
|
7307 |
+
"grad_norm": 0.7753776907920837,
|
7308 |
+
"learning_rate": 1.965758400908334e-05,
|
7309 |
+
"loss": 1.4926,
|
7310 |
+
"step": 1043
|
7311 |
+
},
|
7312 |
+
{
|
7313 |
+
"epoch": 0.0686875997170913,
|
7314 |
+
"grad_norm": 0.8299908638000488,
|
7315 |
+
"learning_rate": 1.957254821968298e-05,
|
7316 |
+
"loss": 1.743,
|
7317 |
+
"step": 1044
|
7318 |
+
},
|
7319 |
+
{
|
7320 |
+
"epoch": 0.06875339243712683,
|
7321 |
+
"grad_norm": 0.754217267036438,
|
7322 |
+
"learning_rate": 1.948765197227983e-05,
|
7323 |
+
"loss": 1.4112,
|
7324 |
+
"step": 1045
|
7325 |
+
},
|
7326 |
+
{
|
7327 |
+
"epoch": 0.06881918515716236,
|
7328 |
+
"grad_norm": 0.7843993902206421,
|
7329 |
+
"learning_rate": 1.9402895656212833e-05,
|
7330 |
+
"loss": 1.4893,
|
7331 |
+
"step": 1046
|
7332 |
+
},
|
7333 |
+
{
|
7334 |
+
"epoch": 0.06888497787719788,
|
7335 |
+
"grad_norm": 0.8628965616226196,
|
7336 |
+
"learning_rate": 1.9318279660179217e-05,
|
7337 |
+
"loss": 1.5237,
|
7338 |
+
"step": 1047
|
7339 |
+
},
|
7340 |
+
{
|
7341 |
+
"epoch": 0.06895077059723341,
|
7342 |
+
"grad_norm": 0.9835182428359985,
|
7343 |
+
"learning_rate": 1.9233804372232643e-05,
|
7344 |
+
"loss": 1.5732,
|
7345 |
+
"step": 1048
|
7346 |
+
},
|
7347 |
+
{
|
7348 |
+
"epoch": 0.06901656331726895,
|
7349 |
+
"grad_norm": 1.099034309387207,
|
7350 |
+
"learning_rate": 1.914947017978153e-05,
|
7351 |
+
"loss": 1.5774,
|
7352 |
+
"step": 1049
|
7353 |
+
},
|
7354 |
+
{
|
7355 |
+
"epoch": 0.06908235603730448,
|
7356 |
+
"grad_norm": 1.265041470527649,
|
7357 |
+
"learning_rate": 1.9065277469587204e-05,
|
7358 |
+
"loss": 1.3189,
|
7359 |
+
"step": 1050
|
7360 |
+
},
|
7361 |
+
{
|
7362 |
+
"epoch": 0.06914814875734,
|
7363 |
+
"grad_norm": 0.5150356888771057,
|
7364 |
+
"learning_rate": 1.8981226627762143e-05,
|
7365 |
+
"loss": 1.8932,
|
7366 |
+
"step": 1051
|
7367 |
+
},
|
7368 |
+
{
|
7369 |
+
"epoch": 0.06921394147737553,
|
7370 |
+
"grad_norm": 0.5652185678482056,
|
7371 |
+
"learning_rate": 1.889731803976822e-05,
|
7372 |
+
"loss": 1.8346,
|
7373 |
+
"step": 1052
|
7374 |
+
},
|
7375 |
+
{
|
7376 |
+
"epoch": 0.06927973419741106,
|
7377 |
+
"grad_norm": 0.5585607886314392,
|
7378 |
+
"learning_rate": 1.881355209041491e-05,
|
7379 |
+
"loss": 1.7214,
|
7380 |
+
"step": 1053
|
7381 |
+
},
|
7382 |
+
{
|
7383 |
+
"epoch": 0.06934552691744658,
|
7384 |
+
"grad_norm": 0.5780192613601685,
|
7385 |
+
"learning_rate": 1.8729929163857552e-05,
|
7386 |
+
"loss": 1.6242,
|
7387 |
+
"step": 1054
|
7388 |
+
},
|
7389 |
+
{
|
7390 |
+
"epoch": 0.06941131963748211,
|
7391 |
+
"grad_norm": 0.5689457654953003,
|
7392 |
+
"learning_rate": 1.8646449643595565e-05,
|
7393 |
+
"loss": 1.6942,
|
7394 |
+
"step": 1055
|
7395 |
+
},
|
7396 |
+
{
|
7397 |
+
"epoch": 0.06947711235751763,
|
7398 |
+
"grad_norm": 0.5897836685180664,
|
7399 |
+
"learning_rate": 1.8563113912470702e-05,
|
7400 |
+
"loss": 1.6164,
|
7401 |
+
"step": 1056
|
7402 |
+
},
|
7403 |
+
{
|
7404 |
+
"epoch": 0.06954290507755317,
|
7405 |
+
"grad_norm": 0.5599413514137268,
|
7406 |
+
"learning_rate": 1.847992235266529e-05,
|
7407 |
+
"loss": 1.5614,
|
7408 |
+
"step": 1057
|
7409 |
+
},
|
7410 |
+
{
|
7411 |
+
"epoch": 0.0696086977975887,
|
7412 |
+
"grad_norm": 0.5642629861831665,
|
7413 |
+
"learning_rate": 1.8396875345700497e-05,
|
7414 |
+
"loss": 1.54,
|
7415 |
+
"step": 1058
|
7416 |
+
},
|
7417 |
+
{
|
7418 |
+
"epoch": 0.06967449051762423,
|
7419 |
+
"grad_norm": 0.6085110902786255,
|
7420 |
+
"learning_rate": 1.8313973272434504e-05,
|
7421 |
+
"loss": 1.6037,
|
7422 |
+
"step": 1059
|
7423 |
+
},
|
7424 |
+
{
|
7425 |
+
"epoch": 0.06974028323765975,
|
7426 |
+
"grad_norm": 0.5913036465644836,
|
7427 |
+
"learning_rate": 1.8231216513060893e-05,
|
7428 |
+
"loss": 1.6223,
|
7429 |
+
"step": 1060
|
7430 |
+
},
|
7431 |
+
{
|
7432 |
+
"epoch": 0.06980607595769528,
|
7433 |
+
"grad_norm": 0.5918750166893005,
|
7434 |
+
"learning_rate": 1.8148605447106797e-05,
|
7435 |
+
"loss": 1.5262,
|
7436 |
+
"step": 1061
|
7437 |
+
},
|
7438 |
+
{
|
7439 |
+
"epoch": 0.0698718686777308,
|
7440 |
+
"grad_norm": 0.5838180184364319,
|
7441 |
+
"learning_rate": 1.8066140453431173e-05,
|
7442 |
+
"loss": 1.6248,
|
7443 |
+
"step": 1062
|
7444 |
+
},
|
7445 |
+
{
|
7446 |
+
"epoch": 0.06993766139776633,
|
7447 |
+
"grad_norm": 0.6557063460350037,
|
7448 |
+
"learning_rate": 1.798382191022313e-05,
|
7449 |
+
"loss": 1.7616,
|
7450 |
+
"step": 1063
|
7451 |
+
},
|
7452 |
+
{
|
7453 |
+
"epoch": 0.07000345411780187,
|
7454 |
+
"grad_norm": 0.6090285181999207,
|
7455 |
+
"learning_rate": 1.7901650195000068e-05,
|
7456 |
+
"loss": 1.6591,
|
7457 |
+
"step": 1064
|
7458 |
+
},
|
7459 |
+
{
|
7460 |
+
"epoch": 0.0700692468378374,
|
7461 |
+
"grad_norm": 0.6101527214050293,
|
7462 |
+
"learning_rate": 1.7819625684606102e-05,
|
7463 |
+
"loss": 1.5755,
|
7464 |
+
"step": 1065
|
7465 |
+
},
|
7466 |
+
{
|
7467 |
+
"epoch": 0.07013503955787292,
|
7468 |
+
"grad_norm": 0.5960958003997803,
|
7469 |
+
"learning_rate": 1.7737748755210214e-05,
|
7470 |
+
"loss": 1.5059,
|
7471 |
+
"step": 1066
|
7472 |
+
},
|
7473 |
+
{
|
7474 |
+
"epoch": 0.07020083227790845,
|
7475 |
+
"grad_norm": 0.631040096282959,
|
7476 |
+
"learning_rate": 1.76560197823046e-05,
|
7477 |
+
"loss": 1.5921,
|
7478 |
+
"step": 1067
|
7479 |
+
},
|
7480 |
+
{
|
7481 |
+
"epoch": 0.07026662499794398,
|
7482 |
+
"grad_norm": 0.6434530019760132,
|
7483 |
+
"learning_rate": 1.7574439140702902e-05,
|
7484 |
+
"loss": 1.5911,
|
7485 |
+
"step": 1068
|
7486 |
+
},
|
7487 |
+
{
|
7488 |
+
"epoch": 0.0703324177179795,
|
7489 |
+
"grad_norm": 0.6554668545722961,
|
7490 |
+
"learning_rate": 1.7493007204538503e-05,
|
7491 |
+
"loss": 1.7435,
|
7492 |
+
"step": 1069
|
7493 |
+
},
|
7494 |
+
{
|
7495 |
+
"epoch": 0.07039821043801503,
|
7496 |
+
"grad_norm": 0.6489726901054382,
|
7497 |
+
"learning_rate": 1.7411724347262824e-05,
|
7498 |
+
"loss": 1.7055,
|
7499 |
+
"step": 1070
|
7500 |
+
},
|
7501 |
+
{
|
7502 |
+
"epoch": 0.07046400315805056,
|
7503 |
+
"grad_norm": 0.628710150718689,
|
7504 |
+
"learning_rate": 1.7330590941643603e-05,
|
7505 |
+
"loss": 1.5092,
|
7506 |
+
"step": 1071
|
7507 |
+
},
|
7508 |
+
{
|
7509 |
+
"epoch": 0.0705297958780861,
|
7510 |
+
"grad_norm": 0.6012537479400635,
|
7511 |
+
"learning_rate": 1.7249607359763174e-05,
|
7512 |
+
"loss": 1.4466,
|
7513 |
+
"step": 1072
|
7514 |
+
},
|
7515 |
+
{
|
7516 |
+
"epoch": 0.07059558859812162,
|
7517 |
+
"grad_norm": 0.6696065664291382,
|
7518 |
+
"learning_rate": 1.7168773973016776e-05,
|
7519 |
+
"loss": 1.5727,
|
7520 |
+
"step": 1073
|
7521 |
+
},
|
7522 |
+
{
|
7523 |
+
"epoch": 0.07066138131815715,
|
7524 |
+
"grad_norm": 0.6296167373657227,
|
7525 |
+
"learning_rate": 1.7088091152110873e-05,
|
7526 |
+
"loss": 1.5393,
|
7527 |
+
"step": 1074
|
7528 |
+
},
|
7529 |
+
{
|
7530 |
+
"epoch": 0.07072717403819268,
|
7531 |
+
"grad_norm": 0.6769093871116638,
|
7532 |
+
"learning_rate": 1.7007559267061334e-05,
|
7533 |
+
"loss": 1.6523,
|
7534 |
+
"step": 1075
|
7535 |
+
},
|
7536 |
+
{
|
7537 |
+
"epoch": 0.0707929667582282,
|
7538 |
+
"grad_norm": 0.6519213914871216,
|
7539 |
+
"learning_rate": 1.692717868719195e-05,
|
7540 |
+
"loss": 1.5566,
|
7541 |
+
"step": 1076
|
7542 |
+
},
|
7543 |
+
{
|
7544 |
+
"epoch": 0.07085875947826373,
|
7545 |
+
"grad_norm": 0.6642208695411682,
|
7546 |
+
"learning_rate": 1.6846949781132548e-05,
|
7547 |
+
"loss": 1.5886,
|
7548 |
+
"step": 1077
|
7549 |
+
},
|
7550 |
+
{
|
7551 |
+
"epoch": 0.07092455219829925,
|
7552 |
+
"grad_norm": 0.6803880333900452,
|
7553 |
+
"learning_rate": 1.6766872916817377e-05,
|
7554 |
+
"loss": 1.6519,
|
7555 |
+
"step": 1078
|
7556 |
+
},
|
7557 |
+
{
|
7558 |
+
"epoch": 0.07099034491833478,
|
7559 |
+
"grad_norm": 0.7287022471427917,
|
7560 |
+
"learning_rate": 1.668694846148343e-05,
|
7561 |
+
"loss": 1.7061,
|
7562 |
+
"step": 1079
|
7563 |
+
},
|
7564 |
+
{
|
7565 |
+
"epoch": 0.07105613763837032,
|
7566 |
+
"grad_norm": 0.6571015119552612,
|
7567 |
+
"learning_rate": 1.660717678166871e-05,
|
7568 |
+
"loss": 1.5716,
|
7569 |
+
"step": 1080
|
7570 |
+
},
|
7571 |
+
{
|
7572 |
+
"epoch": 0.07112193035840585,
|
7573 |
+
"grad_norm": 0.6374773383140564,
|
7574 |
+
"learning_rate": 1.6527558243210623e-05,
|
7575 |
+
"loss": 1.455,
|
7576 |
+
"step": 1081
|
7577 |
+
},
|
7578 |
+
{
|
7579 |
+
"epoch": 0.07118772307844137,
|
7580 |
+
"grad_norm": 0.699365496635437,
|
7581 |
+
"learning_rate": 1.644809321124423e-05,
|
7582 |
+
"loss": 1.6404,
|
7583 |
+
"step": 1082
|
7584 |
+
},
|
7585 |
+
{
|
7586 |
+
"epoch": 0.0712535157984769,
|
7587 |
+
"grad_norm": 0.6866129636764526,
|
7588 |
+
"learning_rate": 1.6368782050200633e-05,
|
7589 |
+
"loss": 1.4377,
|
7590 |
+
"step": 1083
|
7591 |
+
},
|
7592 |
+
{
|
7593 |
+
"epoch": 0.07131930851851243,
|
7594 |
+
"grad_norm": 0.6566864252090454,
|
7595 |
+
"learning_rate": 1.6289625123805245e-05,
|
7596 |
+
"loss": 1.481,
|
7597 |
+
"step": 1084
|
7598 |
+
},
|
7599 |
+
{
|
7600 |
+
"epoch": 0.07138510123854795,
|
7601 |
+
"grad_norm": 0.7273065447807312,
|
7602 |
+
"learning_rate": 1.621062279507617e-05,
|
7603 |
+
"loss": 1.6732,
|
7604 |
+
"step": 1085
|
7605 |
+
},
|
7606 |
+
{
|
7607 |
+
"epoch": 0.07145089395858348,
|
7608 |
+
"grad_norm": 0.7388779520988464,
|
7609 |
+
"learning_rate": 1.6131775426322503e-05,
|
7610 |
+
"loss": 1.5516,
|
7611 |
+
"step": 1086
|
7612 |
+
},
|
7613 |
+
{
|
7614 |
+
"epoch": 0.071516686678619,
|
7615 |
+
"grad_norm": 0.7159716486930847,
|
7616 |
+
"learning_rate": 1.60530833791427e-05,
|
7617 |
+
"loss": 1.5815,
|
7618 |
+
"step": 1087
|
7619 |
+
},
|
7620 |
+
{
|
7621 |
+
"epoch": 0.07158247939865454,
|
7622 |
+
"grad_norm": 0.7883169054985046,
|
7623 |
+
"learning_rate": 1.597454701442288e-05,
|
7624 |
+
"loss": 1.5699,
|
7625 |
+
"step": 1088
|
7626 |
+
},
|
7627 |
+
{
|
7628 |
+
"epoch": 0.07164827211869007,
|
7629 |
+
"grad_norm": 0.7535433769226074,
|
7630 |
+
"learning_rate": 1.589616669233522e-05,
|
7631 |
+
"loss": 1.71,
|
7632 |
+
"step": 1089
|
7633 |
+
},
|
7634 |
+
{
|
7635 |
+
"epoch": 0.0717140648387256,
|
7636 |
+
"grad_norm": 0.7542433142662048,
|
7637 |
+
"learning_rate": 1.581794277233628e-05,
|
7638 |
+
"loss": 1.6412,
|
7639 |
+
"step": 1090
|
7640 |
+
},
|
7641 |
+
{
|
7642 |
+
"epoch": 0.07177985755876112,
|
7643 |
+
"grad_norm": 0.7680163383483887,
|
7644 |
+
"learning_rate": 1.5739875613165283e-05,
|
7645 |
+
"loss": 1.5868,
|
7646 |
+
"step": 1091
|
7647 |
+
},
|
7648 |
+
{
|
7649 |
+
"epoch": 0.07184565027879665,
|
7650 |
+
"grad_norm": 0.8039641380310059,
|
7651 |
+
"learning_rate": 1.566196557284264e-05,
|
7652 |
+
"loss": 1.6165,
|
7653 |
+
"step": 1092
|
7654 |
+
},
|
7655 |
+
{
|
7656 |
+
"epoch": 0.07191144299883218,
|
7657 |
+
"grad_norm": 0.8360242247581482,
|
7658 |
+
"learning_rate": 1.5584213008668147e-05,
|
7659 |
+
"loss": 1.4836,
|
7660 |
+
"step": 1093
|
7661 |
+
},
|
7662 |
+
{
|
7663 |
+
"epoch": 0.0719772357188677,
|
7664 |
+
"grad_norm": 0.9569554328918457,
|
7665 |
+
"learning_rate": 1.550661827721941e-05,
|
7666 |
+
"loss": 1.7255,
|
7667 |
+
"step": 1094
|
7668 |
+
},
|
7669 |
+
{
|
7670 |
+
"epoch": 0.07204302843890324,
|
7671 |
+
"grad_norm": 0.8485451340675354,
|
7672 |
+
"learning_rate": 1.5429181734350236e-05,
|
7673 |
+
"loss": 1.6198,
|
7674 |
+
"step": 1095
|
7675 |
+
},
|
7676 |
+
{
|
7677 |
+
"epoch": 0.07210882115893877,
|
7678 |
+
"grad_norm": 0.8728616237640381,
|
7679 |
+
"learning_rate": 1.53519037351889e-05,
|
7680 |
+
"loss": 1.6094,
|
7681 |
+
"step": 1096
|
7682 |
+
},
|
7683 |
+
{
|
7684 |
+
"epoch": 0.0721746138789743,
|
7685 |
+
"grad_norm": 0.9522159695625305,
|
7686 |
+
"learning_rate": 1.527478463413666e-05,
|
7687 |
+
"loss": 1.3929,
|
7688 |
+
"step": 1097
|
7689 |
+
},
|
7690 |
+
{
|
7691 |
+
"epoch": 0.07224040659900982,
|
7692 |
+
"grad_norm": 1.0558445453643799,
|
7693 |
+
"learning_rate": 1.5197824784866015e-05,
|
7694 |
+
"loss": 1.7511,
|
7695 |
+
"step": 1098
|
7696 |
+
},
|
7697 |
+
{
|
7698 |
+
"epoch": 0.07230619931904535,
|
7699 |
+
"grad_norm": 1.0129352807998657,
|
7700 |
+
"learning_rate": 1.5121024540319161e-05,
|
7701 |
+
"loss": 1.4986,
|
7702 |
+
"step": 1099
|
7703 |
+
},
|
7704 |
+
{
|
7705 |
+
"epoch": 0.07237199203908087,
|
7706 |
+
"grad_norm": 1.232009768486023,
|
7707 |
+
"learning_rate": 1.5044384252706312e-05,
|
7708 |
+
"loss": 1.2485,
|
7709 |
+
"step": 1100
|
7710 |
+
},
|
7711 |
+
{
|
7712 |
+
"epoch": 0.0724377847591164,
|
7713 |
+
"grad_norm": 0.5130254626274109,
|
7714 |
+
"learning_rate": 1.4967904273504113e-05,
|
7715 |
+
"loss": 1.816,
|
7716 |
+
"step": 1101
|
7717 |
+
},
|
7718 |
+
{
|
7719 |
+
"epoch": 0.07250357747915193,
|
7720 |
+
"grad_norm": 0.5423897504806519,
|
7721 |
+
"learning_rate": 1.4891584953454036e-05,
|
7722 |
+
"loss": 1.7201,
|
7723 |
+
"step": 1102
|
7724 |
+
},
|
7725 |
+
{
|
7726 |
+
"epoch": 0.07256937019918747,
|
7727 |
+
"grad_norm": 0.5203692317008972,
|
7728 |
+
"learning_rate": 1.4815426642560754e-05,
|
7729 |
+
"loss": 1.5166,
|
7730 |
+
"step": 1103
|
7731 |
+
},
|
7732 |
+
{
|
7733 |
+
"epoch": 0.07263516291922299,
|
7734 |
+
"grad_norm": 0.5805792212486267,
|
7735 |
+
"learning_rate": 1.4739429690090533e-05,
|
7736 |
+
"loss": 1.6115,
|
7737 |
+
"step": 1104
|
7738 |
}
|
7739 |
],
|
7740 |
"logging_steps": 1,
|
|
|
7754 |
"attributes": {}
|
7755 |
}
|
7756 |
},
|
7757 |
+
"total_flos": 1.430555906373845e+18,
|
7758 |
"train_batch_size": 4,
|
7759 |
"trial_name": null,
|
7760 |
"trial_params": null
|