Training in progress, step 1393, 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 35237104
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7141fb35b38e73dc0cd6dcb1d2ca0f764a1c83165e27bb41b5614bd1649d857c
|
3 |
size 35237104
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 18810356
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2c2882e51a35fea7cbdfdd7ee97855f918d60480d735b75e4494346a62b87e66
|
3 |
size 18810356
|
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:e0448c0fce72c98b55ce7a959bda7fd70d820e3a465eadeb8aa4e968c89c0b07
|
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 sha256:480cd5995da9a488171cc4b50e2fc2b05048d0e285481447079dda7adec75f16
|
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":
|
5 |
"eval_steps": 2000,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -7015,6 +7015,2757 @@
|
|
7015 |
"learning_rate": 1.8635595197024886e-05,
|
7016 |
"loss": 0.0,
|
7017 |
"step": 1000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7018 |
}
|
7019 |
],
|
7020 |
"logging_steps": 1,
|
@@ -7029,12 +9780,12 @@
|
|
7029 |
"should_evaluate": false,
|
7030 |
"should_log": false,
|
7031 |
"should_save": true,
|
7032 |
-
"should_training_stop":
|
7033 |
},
|
7034 |
"attributes": {}
|
7035 |
}
|
7036 |
},
|
7037 |
-
"total_flos":
|
7038 |
"train_batch_size": 4,
|
7039 |
"trial_name": null,
|
7040 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 10.00359066427289,
|
5 |
"eval_steps": 2000,
|
6 |
+
"global_step": 1393,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
7015 |
"learning_rate": 1.8635595197024886e-05,
|
7016 |
"loss": 0.0,
|
7017 |
"step": 1000
|
7018 |
+
},
|
7019 |
+
{
|
7020 |
+
"epoch": 7.18850987432675,
|
7021 |
+
"grad_norm": 0.008335919119417667,
|
7022 |
+
"learning_rate": 1.8547222382572837e-05,
|
7023 |
+
"loss": 0.0001,
|
7024 |
+
"step": 1001
|
7025 |
+
},
|
7026 |
+
{
|
7027 |
+
"epoch": 7.195691202872531,
|
7028 |
+
"grad_norm": 0.004448807332664728,
|
7029 |
+
"learning_rate": 1.8459011866507452e-05,
|
7030 |
+
"loss": 0.0001,
|
7031 |
+
"step": 1002
|
7032 |
+
},
|
7033 |
+
{
|
7034 |
+
"epoch": 7.202872531418312,
|
7035 |
+
"grad_norm": 0.002166690304875374,
|
7036 |
+
"learning_rate": 1.837096410400078e-05,
|
7037 |
+
"loss": 0.0,
|
7038 |
+
"step": 1003
|
7039 |
+
},
|
7040 |
+
{
|
7041 |
+
"epoch": 7.210053859964093,
|
7042 |
+
"grad_norm": 0.007959997281432152,
|
7043 |
+
"learning_rate": 1.8283079549385e-05,
|
7044 |
+
"loss": 0.0001,
|
7045 |
+
"step": 1004
|
7046 |
+
},
|
7047 |
+
{
|
7048 |
+
"epoch": 7.217235188509874,
|
7049 |
+
"grad_norm": 0.011398511938750744,
|
7050 |
+
"learning_rate": 1.819535865615018e-05,
|
7051 |
+
"loss": 0.0001,
|
7052 |
+
"step": 1005
|
7053 |
+
},
|
7054 |
+
{
|
7055 |
+
"epoch": 7.224416517055655,
|
7056 |
+
"grad_norm": 0.0013214838691055775,
|
7057 |
+
"learning_rate": 1.810780187694185e-05,
|
7058 |
+
"loss": 0.0,
|
7059 |
+
"step": 1006
|
7060 |
+
},
|
7061 |
+
{
|
7062 |
+
"epoch": 7.231597845601437,
|
7063 |
+
"grad_norm": 0.0021551784593611956,
|
7064 |
+
"learning_rate": 1.8020409663558685e-05,
|
7065 |
+
"loss": 0.0,
|
7066 |
+
"step": 1007
|
7067 |
+
},
|
7068 |
+
{
|
7069 |
+
"epoch": 7.238779174147218,
|
7070 |
+
"grad_norm": 0.00151022057980299,
|
7071 |
+
"learning_rate": 1.7933182466950233e-05,
|
7072 |
+
"loss": 0.0,
|
7073 |
+
"step": 1008
|
7074 |
+
},
|
7075 |
+
{
|
7076 |
+
"epoch": 7.2459605026929985,
|
7077 |
+
"grad_norm": 0.004553725477308035,
|
7078 |
+
"learning_rate": 1.784612073721454e-05,
|
7079 |
+
"loss": 0.0001,
|
7080 |
+
"step": 1009
|
7081 |
+
},
|
7082 |
+
{
|
7083 |
+
"epoch": 7.253141831238779,
|
7084 |
+
"grad_norm": 0.0017206427874043584,
|
7085 |
+
"learning_rate": 1.77592249235958e-05,
|
7086 |
+
"loss": 0.0,
|
7087 |
+
"step": 1010
|
7088 |
+
},
|
7089 |
+
{
|
7090 |
+
"epoch": 7.26032315978456,
|
7091 |
+
"grad_norm": 0.009418018162250519,
|
7092 |
+
"learning_rate": 1.767249547448208e-05,
|
7093 |
+
"loss": 0.0001,
|
7094 |
+
"step": 1011
|
7095 |
+
},
|
7096 |
+
{
|
7097 |
+
"epoch": 7.267504488330341,
|
7098 |
+
"grad_norm": 0.006585447583347559,
|
7099 |
+
"learning_rate": 1.7585932837403024e-05,
|
7100 |
+
"loss": 0.0001,
|
7101 |
+
"step": 1012
|
7102 |
+
},
|
7103 |
+
{
|
7104 |
+
"epoch": 7.274685816876122,
|
7105 |
+
"grad_norm": 0.002000662963837385,
|
7106 |
+
"learning_rate": 1.749953745902746e-05,
|
7107 |
+
"loss": 0.0,
|
7108 |
+
"step": 1013
|
7109 |
+
},
|
7110 |
+
{
|
7111 |
+
"epoch": 7.281867145421903,
|
7112 |
+
"grad_norm": 0.007135197054594755,
|
7113 |
+
"learning_rate": 1.741330978516122e-05,
|
7114 |
+
"loss": 0.0001,
|
7115 |
+
"step": 1014
|
7116 |
+
},
|
7117 |
+
{
|
7118 |
+
"epoch": 7.289048473967684,
|
7119 |
+
"grad_norm": 0.005208285059779882,
|
7120 |
+
"learning_rate": 1.73272502607447e-05,
|
7121 |
+
"loss": 0.0001,
|
7122 |
+
"step": 1015
|
7123 |
+
},
|
7124 |
+
{
|
7125 |
+
"epoch": 7.296229802513465,
|
7126 |
+
"grad_norm": 0.0017541215056553483,
|
7127 |
+
"learning_rate": 1.7241359329850698e-05,
|
7128 |
+
"loss": 0.0,
|
7129 |
+
"step": 1016
|
7130 |
+
},
|
7131 |
+
{
|
7132 |
+
"epoch": 7.303411131059246,
|
7133 |
+
"grad_norm": 0.0037172988522797823,
|
7134 |
+
"learning_rate": 1.715563743568199e-05,
|
7135 |
+
"loss": 0.0001,
|
7136 |
+
"step": 1017
|
7137 |
+
},
|
7138 |
+
{
|
7139 |
+
"epoch": 7.310592459605027,
|
7140 |
+
"grad_norm": 0.0013541750377044082,
|
7141 |
+
"learning_rate": 1.7070085020569192e-05,
|
7142 |
+
"loss": 0.0,
|
7143 |
+
"step": 1018
|
7144 |
+
},
|
7145 |
+
{
|
7146 |
+
"epoch": 7.317773788150808,
|
7147 |
+
"grad_norm": 0.0008809631108306348,
|
7148 |
+
"learning_rate": 1.6984702525968326e-05,
|
7149 |
+
"loss": 0.0,
|
7150 |
+
"step": 1019
|
7151 |
+
},
|
7152 |
+
{
|
7153 |
+
"epoch": 7.3249551166965885,
|
7154 |
+
"grad_norm": 0.0033160988241434097,
|
7155 |
+
"learning_rate": 1.6899490392458627e-05,
|
7156 |
+
"loss": 0.0,
|
7157 |
+
"step": 1020
|
7158 |
+
},
|
7159 |
+
{
|
7160 |
+
"epoch": 7.332136445242369,
|
7161 |
+
"grad_norm": 0.006269481498748064,
|
7162 |
+
"learning_rate": 1.6814449059740306e-05,
|
7163 |
+
"loss": 0.0001,
|
7164 |
+
"step": 1021
|
7165 |
+
},
|
7166 |
+
{
|
7167 |
+
"epoch": 7.339317773788151,
|
7168 |
+
"grad_norm": 0.003873833455145359,
|
7169 |
+
"learning_rate": 1.6729578966632153e-05,
|
7170 |
+
"loss": 0.0,
|
7171 |
+
"step": 1022
|
7172 |
+
},
|
7173 |
+
{
|
7174 |
+
"epoch": 7.346499102333932,
|
7175 |
+
"grad_norm": 0.013830773532390594,
|
7176 |
+
"learning_rate": 1.664488055106942e-05,
|
7177 |
+
"loss": 0.0001,
|
7178 |
+
"step": 1023
|
7179 |
+
},
|
7180 |
+
{
|
7181 |
+
"epoch": 7.353680430879713,
|
7182 |
+
"grad_norm": 0.006302582565695047,
|
7183 |
+
"learning_rate": 1.6560354250101462e-05,
|
7184 |
+
"loss": 0.0001,
|
7185 |
+
"step": 1024
|
7186 |
+
},
|
7187 |
+
{
|
7188 |
+
"epoch": 7.360861759425494,
|
7189 |
+
"grad_norm": 0.008939807303249836,
|
7190 |
+
"learning_rate": 1.6476000499889517e-05,
|
7191 |
+
"loss": 0.0001,
|
7192 |
+
"step": 1025
|
7193 |
+
},
|
7194 |
+
{
|
7195 |
+
"epoch": 7.368043087971275,
|
7196 |
+
"grad_norm": 0.004149202723056078,
|
7197 |
+
"learning_rate": 1.639181973570441e-05,
|
7198 |
+
"loss": 0.0001,
|
7199 |
+
"step": 1026
|
7200 |
+
},
|
7201 |
+
{
|
7202 |
+
"epoch": 7.375224416517056,
|
7203 |
+
"grad_norm": 0.005756485741585493,
|
7204 |
+
"learning_rate": 1.6307812391924428e-05,
|
7205 |
+
"loss": 0.0001,
|
7206 |
+
"step": 1027
|
7207 |
+
},
|
7208 |
+
{
|
7209 |
+
"epoch": 7.382405745062837,
|
7210 |
+
"grad_norm": 0.003365996992215514,
|
7211 |
+
"learning_rate": 1.622397890203292e-05,
|
7212 |
+
"loss": 0.0,
|
7213 |
+
"step": 1028
|
7214 |
+
},
|
7215 |
+
{
|
7216 |
+
"epoch": 7.3895870736086176,
|
7217 |
+
"grad_norm": 0.0007422404014505446,
|
7218 |
+
"learning_rate": 1.6140319698616174e-05,
|
7219 |
+
"loss": 0.0,
|
7220 |
+
"step": 1029
|
7221 |
+
},
|
7222 |
+
{
|
7223 |
+
"epoch": 7.3967684021543985,
|
7224 |
+
"grad_norm": 0.0014552025822922587,
|
7225 |
+
"learning_rate": 1.605683521336116e-05,
|
7226 |
+
"loss": 0.0,
|
7227 |
+
"step": 1030
|
7228 |
+
},
|
7229 |
+
{
|
7230 |
+
"epoch": 7.403949730700179,
|
7231 |
+
"grad_norm": 0.0042173066176474094,
|
7232 |
+
"learning_rate": 1.5973525877053242e-05,
|
7233 |
+
"loss": 0.0,
|
7234 |
+
"step": 1031
|
7235 |
+
},
|
7236 |
+
{
|
7237 |
+
"epoch": 7.41113105924596,
|
7238 |
+
"grad_norm": 0.005891093984246254,
|
7239 |
+
"learning_rate": 1.589039211957406e-05,
|
7240 |
+
"loss": 0.0001,
|
7241 |
+
"step": 1032
|
7242 |
+
},
|
7243 |
+
{
|
7244 |
+
"epoch": 7.418312387791741,
|
7245 |
+
"grad_norm": 0.004144870676100254,
|
7246 |
+
"learning_rate": 1.5807434369899248e-05,
|
7247 |
+
"loss": 0.0001,
|
7248 |
+
"step": 1033
|
7249 |
+
},
|
7250 |
+
{
|
7251 |
+
"epoch": 7.425493716337522,
|
7252 |
+
"grad_norm": 0.011198299936950207,
|
7253 |
+
"learning_rate": 1.572465305609619e-05,
|
7254 |
+
"loss": 0.0001,
|
7255 |
+
"step": 1034
|
7256 |
+
},
|
7257 |
+
{
|
7258 |
+
"epoch": 7.432675044883303,
|
7259 |
+
"grad_norm": 0.002214165637269616,
|
7260 |
+
"learning_rate": 1.564204860532186e-05,
|
7261 |
+
"loss": 0.0,
|
7262 |
+
"step": 1035
|
7263 |
+
},
|
7264 |
+
{
|
7265 |
+
"epoch": 7.439856373429085,
|
7266 |
+
"grad_norm": 0.0012972512049600482,
|
7267 |
+
"learning_rate": 1.5559621443820643e-05,
|
7268 |
+
"loss": 0.0,
|
7269 |
+
"step": 1036
|
7270 |
+
},
|
7271 |
+
{
|
7272 |
+
"epoch": 7.447037701974866,
|
7273 |
+
"grad_norm": 0.0006055528065189719,
|
7274 |
+
"learning_rate": 1.5477371996922067e-05,
|
7275 |
+
"loss": 0.0,
|
7276 |
+
"step": 1037
|
7277 |
+
},
|
7278 |
+
{
|
7279 |
+
"epoch": 7.454219030520647,
|
7280 |
+
"grad_norm": 0.0029738997109234333,
|
7281 |
+
"learning_rate": 1.5395300689038637e-05,
|
7282 |
+
"loss": 0.0,
|
7283 |
+
"step": 1038
|
7284 |
+
},
|
7285 |
+
{
|
7286 |
+
"epoch": 7.4614003590664275,
|
7287 |
+
"grad_norm": 0.0036454282235354185,
|
7288 |
+
"learning_rate": 1.531340794366369e-05,
|
7289 |
+
"loss": 0.0,
|
7290 |
+
"step": 1039
|
7291 |
+
},
|
7292 |
+
{
|
7293 |
+
"epoch": 7.468581687612208,
|
7294 |
+
"grad_norm": 0.00483043072745204,
|
7295 |
+
"learning_rate": 1.5231694183369105e-05,
|
7296 |
+
"loss": 0.0001,
|
7297 |
+
"step": 1040
|
7298 |
+
},
|
7299 |
+
{
|
7300 |
+
"epoch": 7.475763016157989,
|
7301 |
+
"grad_norm": 0.0017489743186160922,
|
7302 |
+
"learning_rate": 1.5150159829803246e-05,
|
7303 |
+
"loss": 0.0,
|
7304 |
+
"step": 1041
|
7305 |
+
},
|
7306 |
+
{
|
7307 |
+
"epoch": 7.48294434470377,
|
7308 |
+
"grad_norm": 0.0027065305039286613,
|
7309 |
+
"learning_rate": 1.5068805303688704e-05,
|
7310 |
+
"loss": 0.0001,
|
7311 |
+
"step": 1042
|
7312 |
+
},
|
7313 |
+
{
|
7314 |
+
"epoch": 7.490125673249551,
|
7315 |
+
"grad_norm": 0.007340044714510441,
|
7316 |
+
"learning_rate": 1.4987631024820147e-05,
|
7317 |
+
"loss": 0.0001,
|
7318 |
+
"step": 1043
|
7319 |
+
},
|
7320 |
+
{
|
7321 |
+
"epoch": 7.497307001795332,
|
7322 |
+
"grad_norm": 0.0028522072825580835,
|
7323 |
+
"learning_rate": 1.490663741206213e-05,
|
7324 |
+
"loss": 0.0,
|
7325 |
+
"step": 1044
|
7326 |
+
},
|
7327 |
+
{
|
7328 |
+
"epoch": 7.504488330341113,
|
7329 |
+
"grad_norm": 0.0007877356256358325,
|
7330 |
+
"learning_rate": 1.4825824883347017e-05,
|
7331 |
+
"loss": 0.0,
|
7332 |
+
"step": 1045
|
7333 |
+
},
|
7334 |
+
{
|
7335 |
+
"epoch": 7.511669658886894,
|
7336 |
+
"grad_norm": 0.0024650641717016697,
|
7337 |
+
"learning_rate": 1.4745193855672707e-05,
|
7338 |
+
"loss": 0.0,
|
7339 |
+
"step": 1046
|
7340 |
+
},
|
7341 |
+
{
|
7342 |
+
"epoch": 7.518850987432675,
|
7343 |
+
"grad_norm": 0.0008501393022015691,
|
7344 |
+
"learning_rate": 1.46647447451006e-05,
|
7345 |
+
"loss": 0.0,
|
7346 |
+
"step": 1047
|
7347 |
+
},
|
7348 |
+
{
|
7349 |
+
"epoch": 7.526032315978456,
|
7350 |
+
"grad_norm": 0.002214528387412429,
|
7351 |
+
"learning_rate": 1.4584477966753324e-05,
|
7352 |
+
"loss": 0.0,
|
7353 |
+
"step": 1048
|
7354 |
+
},
|
7355 |
+
{
|
7356 |
+
"epoch": 7.533213644524237,
|
7357 |
+
"grad_norm": 0.0006226448458619416,
|
7358 |
+
"learning_rate": 1.4504393934812732e-05,
|
7359 |
+
"loss": 0.0,
|
7360 |
+
"step": 1049
|
7361 |
+
},
|
7362 |
+
{
|
7363 |
+
"epoch": 7.540394973070018,
|
7364 |
+
"grad_norm": 0.0036075678654015064,
|
7365 |
+
"learning_rate": 1.4424493062517625e-05,
|
7366 |
+
"loss": 0.0,
|
7367 |
+
"step": 1050
|
7368 |
+
},
|
7369 |
+
{
|
7370 |
+
"epoch": 7.547576301615799,
|
7371 |
+
"grad_norm": 0.00435080798342824,
|
7372 |
+
"learning_rate": 1.434477576216176e-05,
|
7373 |
+
"loss": 0.0,
|
7374 |
+
"step": 1051
|
7375 |
+
},
|
7376 |
+
{
|
7377 |
+
"epoch": 7.55475763016158,
|
7378 |
+
"grad_norm": 0.0033029005862772465,
|
7379 |
+
"learning_rate": 1.4265242445091587e-05,
|
7380 |
+
"loss": 0.0001,
|
7381 |
+
"step": 1052
|
7382 |
+
},
|
7383 |
+
{
|
7384 |
+
"epoch": 7.561938958707361,
|
7385 |
+
"grad_norm": 0.007004761602729559,
|
7386 |
+
"learning_rate": 1.4185893521704213e-05,
|
7387 |
+
"loss": 0.0001,
|
7388 |
+
"step": 1053
|
7389 |
+
},
|
7390 |
+
{
|
7391 |
+
"epoch": 7.569120287253142,
|
7392 |
+
"grad_norm": 0.008575864136219025,
|
7393 |
+
"learning_rate": 1.410672940144529e-05,
|
7394 |
+
"loss": 0.0001,
|
7395 |
+
"step": 1054
|
7396 |
+
},
|
7397 |
+
{
|
7398 |
+
"epoch": 7.576301615798923,
|
7399 |
+
"grad_norm": 0.0034808525815606117,
|
7400 |
+
"learning_rate": 1.4027750492806818e-05,
|
7401 |
+
"loss": 0.0001,
|
7402 |
+
"step": 1055
|
7403 |
+
},
|
7404 |
+
{
|
7405 |
+
"epoch": 7.583482944344704,
|
7406 |
+
"grad_norm": 0.0015421882271766663,
|
7407 |
+
"learning_rate": 1.3948957203325152e-05,
|
7408 |
+
"loss": 0.0,
|
7409 |
+
"step": 1056
|
7410 |
+
},
|
7411 |
+
{
|
7412 |
+
"epoch": 7.590664272890485,
|
7413 |
+
"grad_norm": 0.0038173808716237545,
|
7414 |
+
"learning_rate": 1.3870349939578787e-05,
|
7415 |
+
"loss": 0.0001,
|
7416 |
+
"step": 1057
|
7417 |
+
},
|
7418 |
+
{
|
7419 |
+
"epoch": 7.597845601436266,
|
7420 |
+
"grad_norm": 0.003961088601499796,
|
7421 |
+
"learning_rate": 1.3791929107186363e-05,
|
7422 |
+
"loss": 0.0001,
|
7423 |
+
"step": 1058
|
7424 |
+
},
|
7425 |
+
{
|
7426 |
+
"epoch": 7.6050269299820465,
|
7427 |
+
"grad_norm": 0.004210003186017275,
|
7428 |
+
"learning_rate": 1.3713695110804464e-05,
|
7429 |
+
"loss": 0.0001,
|
7430 |
+
"step": 1059
|
7431 |
+
},
|
7432 |
+
{
|
7433 |
+
"epoch": 7.6122082585278275,
|
7434 |
+
"grad_norm": 0.000915760057978332,
|
7435 |
+
"learning_rate": 1.3635648354125663e-05,
|
7436 |
+
"loss": 0.0,
|
7437 |
+
"step": 1060
|
7438 |
+
},
|
7439 |
+
{
|
7440 |
+
"epoch": 7.619389587073608,
|
7441 |
+
"grad_norm": 0.005486072972416878,
|
7442 |
+
"learning_rate": 1.3557789239876296e-05,
|
7443 |
+
"loss": 0.0,
|
7444 |
+
"step": 1061
|
7445 |
+
},
|
7446 |
+
{
|
7447 |
+
"epoch": 7.626570915619389,
|
7448 |
+
"grad_norm": 0.007258336525410414,
|
7449 |
+
"learning_rate": 1.3480118169814465e-05,
|
7450 |
+
"loss": 0.0001,
|
7451 |
+
"step": 1062
|
7452 |
+
},
|
7453 |
+
{
|
7454 |
+
"epoch": 7.63375224416517,
|
7455 |
+
"grad_norm": 0.002517531393095851,
|
7456 |
+
"learning_rate": 1.3402635544727993e-05,
|
7457 |
+
"loss": 0.0,
|
7458 |
+
"step": 1063
|
7459 |
+
},
|
7460 |
+
{
|
7461 |
+
"epoch": 7.640933572710951,
|
7462 |
+
"grad_norm": 0.0016349544748663902,
|
7463 |
+
"learning_rate": 1.3325341764432259e-05,
|
7464 |
+
"loss": 0.0,
|
7465 |
+
"step": 1064
|
7466 |
+
},
|
7467 |
+
{
|
7468 |
+
"epoch": 7.648114901256733,
|
7469 |
+
"grad_norm": 0.0012608628021553159,
|
7470 |
+
"learning_rate": 1.3248237227768245e-05,
|
7471 |
+
"loss": 0.0,
|
7472 |
+
"step": 1065
|
7473 |
+
},
|
7474 |
+
{
|
7475 |
+
"epoch": 7.655296229802514,
|
7476 |
+
"grad_norm": 0.002880517393350601,
|
7477 |
+
"learning_rate": 1.3171322332600367e-05,
|
7478 |
+
"loss": 0.0,
|
7479 |
+
"step": 1066
|
7480 |
+
},
|
7481 |
+
{
|
7482 |
+
"epoch": 7.662477558348295,
|
7483 |
+
"grad_norm": 0.0019977663177996874,
|
7484 |
+
"learning_rate": 1.3094597475814536e-05,
|
7485 |
+
"loss": 0.0,
|
7486 |
+
"step": 1067
|
7487 |
+
},
|
7488 |
+
{
|
7489 |
+
"epoch": 7.669658886894076,
|
7490 |
+
"grad_norm": 0.0027919632848352194,
|
7491 |
+
"learning_rate": 1.3018063053315993e-05,
|
7492 |
+
"loss": 0.0,
|
7493 |
+
"step": 1068
|
7494 |
+
},
|
7495 |
+
{
|
7496 |
+
"epoch": 7.6768402154398565,
|
7497 |
+
"grad_norm": 0.006321036256849766,
|
7498 |
+
"learning_rate": 1.294171946002739e-05,
|
7499 |
+
"loss": 0.0001,
|
7500 |
+
"step": 1069
|
7501 |
+
},
|
7502 |
+
{
|
7503 |
+
"epoch": 7.684021543985637,
|
7504 |
+
"grad_norm": 0.011540860868990421,
|
7505 |
+
"learning_rate": 1.2865567089886642e-05,
|
7506 |
+
"loss": 0.0001,
|
7507 |
+
"step": 1070
|
7508 |
+
},
|
7509 |
+
{
|
7510 |
+
"epoch": 7.691202872531418,
|
7511 |
+
"grad_norm": 0.0030074333772063255,
|
7512 |
+
"learning_rate": 1.2789606335844928e-05,
|
7513 |
+
"loss": 0.0,
|
7514 |
+
"step": 1071
|
7515 |
+
},
|
7516 |
+
{
|
7517 |
+
"epoch": 7.698384201077199,
|
7518 |
+
"grad_norm": 0.0005612525856122375,
|
7519 |
+
"learning_rate": 1.2713837589864725e-05,
|
7520 |
+
"loss": 0.0,
|
7521 |
+
"step": 1072
|
7522 |
+
},
|
7523 |
+
{
|
7524 |
+
"epoch": 7.70556552962298,
|
7525 |
+
"grad_norm": 0.0016029819380491972,
|
7526 |
+
"learning_rate": 1.2638261242917715e-05,
|
7527 |
+
"loss": 0.0,
|
7528 |
+
"step": 1073
|
7529 |
+
},
|
7530 |
+
{
|
7531 |
+
"epoch": 7.712746858168761,
|
7532 |
+
"grad_norm": 0.0021067261695861816,
|
7533 |
+
"learning_rate": 1.2562877684982776e-05,
|
7534 |
+
"loss": 0.0,
|
7535 |
+
"step": 1074
|
7536 |
+
},
|
7537 |
+
{
|
7538 |
+
"epoch": 7.719928186714542,
|
7539 |
+
"grad_norm": 0.001783563639037311,
|
7540 |
+
"learning_rate": 1.2487687305043976e-05,
|
7541 |
+
"loss": 0.0,
|
7542 |
+
"step": 1075
|
7543 |
+
},
|
7544 |
+
{
|
7545 |
+
"epoch": 7.727109515260323,
|
7546 |
+
"grad_norm": 0.0015695631736889482,
|
7547 |
+
"learning_rate": 1.2412690491088613e-05,
|
7548 |
+
"loss": 0.0,
|
7549 |
+
"step": 1076
|
7550 |
+
},
|
7551 |
+
{
|
7552 |
+
"epoch": 7.734290843806104,
|
7553 |
+
"grad_norm": 0.005196002312004566,
|
7554 |
+
"learning_rate": 1.2337887630105116e-05,
|
7555 |
+
"loss": 0.0001,
|
7556 |
+
"step": 1077
|
7557 |
+
},
|
7558 |
+
{
|
7559 |
+
"epoch": 7.741472172351886,
|
7560 |
+
"grad_norm": 0.002812385791912675,
|
7561 |
+
"learning_rate": 1.2263279108081161e-05,
|
7562 |
+
"loss": 0.0,
|
7563 |
+
"step": 1078
|
7564 |
+
},
|
7565 |
+
{
|
7566 |
+
"epoch": 7.748653500897666,
|
7567 |
+
"grad_norm": 0.000525031064171344,
|
7568 |
+
"learning_rate": 1.2188865310001584e-05,
|
7569 |
+
"loss": 0.0,
|
7570 |
+
"step": 1079
|
7571 |
+
},
|
7572 |
+
{
|
7573 |
+
"epoch": 7.755834829443447,
|
7574 |
+
"grad_norm": 0.0007386524812318385,
|
7575 |
+
"learning_rate": 1.2114646619846425e-05,
|
7576 |
+
"loss": 0.0,
|
7577 |
+
"step": 1080
|
7578 |
+
},
|
7579 |
+
{
|
7580 |
+
"epoch": 7.763016157989228,
|
7581 |
+
"grad_norm": 0.0042157904244959354,
|
7582 |
+
"learning_rate": 1.2040623420588986e-05,
|
7583 |
+
"loss": 0.0001,
|
7584 |
+
"step": 1081
|
7585 |
+
},
|
7586 |
+
{
|
7587 |
+
"epoch": 7.770197486535009,
|
7588 |
+
"grad_norm": 0.0025070691481232643,
|
7589 |
+
"learning_rate": 1.1966796094193827e-05,
|
7590 |
+
"loss": 0.0,
|
7591 |
+
"step": 1082
|
7592 |
+
},
|
7593 |
+
{
|
7594 |
+
"epoch": 7.77737881508079,
|
7595 |
+
"grad_norm": 0.0028257272206246853,
|
7596 |
+
"learning_rate": 1.189316502161476e-05,
|
7597 |
+
"loss": 0.0001,
|
7598 |
+
"step": 1083
|
7599 |
+
},
|
7600 |
+
{
|
7601 |
+
"epoch": 7.784560143626571,
|
7602 |
+
"grad_norm": 0.0007815873832441866,
|
7603 |
+
"learning_rate": 1.1819730582792915e-05,
|
7604 |
+
"loss": 0.0,
|
7605 |
+
"step": 1084
|
7606 |
+
},
|
7607 |
+
{
|
7608 |
+
"epoch": 7.791741472172352,
|
7609 |
+
"grad_norm": 0.0014989282935857773,
|
7610 |
+
"learning_rate": 1.1746493156654814e-05,
|
7611 |
+
"loss": 0.0,
|
7612 |
+
"step": 1085
|
7613 |
+
},
|
7614 |
+
{
|
7615 |
+
"epoch": 7.798922800718133,
|
7616 |
+
"grad_norm": 0.0028032218106091022,
|
7617 |
+
"learning_rate": 1.1673453121110333e-05,
|
7618 |
+
"loss": 0.0001,
|
7619 |
+
"step": 1086
|
7620 |
+
},
|
7621 |
+
{
|
7622 |
+
"epoch": 7.806104129263914,
|
7623 |
+
"grad_norm": 0.0017906483262777328,
|
7624 |
+
"learning_rate": 1.1600610853050858e-05,
|
7625 |
+
"loss": 0.0,
|
7626 |
+
"step": 1087
|
7627 |
+
},
|
7628 |
+
{
|
7629 |
+
"epoch": 7.813285457809695,
|
7630 |
+
"grad_norm": 0.0009083447512239218,
|
7631 |
+
"learning_rate": 1.1527966728347234e-05,
|
7632 |
+
"loss": 0.0,
|
7633 |
+
"step": 1088
|
7634 |
+
},
|
7635 |
+
{
|
7636 |
+
"epoch": 7.8204667863554755,
|
7637 |
+
"grad_norm": 0.006937344092875719,
|
7638 |
+
"learning_rate": 1.1455521121847873e-05,
|
7639 |
+
"loss": 0.0001,
|
7640 |
+
"step": 1089
|
7641 |
+
},
|
7642 |
+
{
|
7643 |
+
"epoch": 7.8276481149012564,
|
7644 |
+
"grad_norm": 0.006985332351177931,
|
7645 |
+
"learning_rate": 1.1383274407376847e-05,
|
7646 |
+
"loss": 0.0001,
|
7647 |
+
"step": 1090
|
7648 |
+
},
|
7649 |
+
{
|
7650 |
+
"epoch": 7.834829443447037,
|
7651 |
+
"grad_norm": 0.0022957210894674063,
|
7652 |
+
"learning_rate": 1.1311226957731946e-05,
|
7653 |
+
"loss": 0.0001,
|
7654 |
+
"step": 1091
|
7655 |
+
},
|
7656 |
+
{
|
7657 |
+
"epoch": 7.842010771992818,
|
7658 |
+
"grad_norm": 0.0055581931956112385,
|
7659 |
+
"learning_rate": 1.1239379144682677e-05,
|
7660 |
+
"loss": 0.0001,
|
7661 |
+
"step": 1092
|
7662 |
+
},
|
7663 |
+
{
|
7664 |
+
"epoch": 7.8491921005386,
|
7665 |
+
"grad_norm": 0.006552727427333593,
|
7666 |
+
"learning_rate": 1.1167731338968479e-05,
|
7667 |
+
"loss": 0.0001,
|
7668 |
+
"step": 1093
|
7669 |
+
},
|
7670 |
+
{
|
7671 |
+
"epoch": 7.856373429084381,
|
7672 |
+
"grad_norm": 0.004818024579435587,
|
7673 |
+
"learning_rate": 1.1096283910296696e-05,
|
7674 |
+
"loss": 0.0001,
|
7675 |
+
"step": 1094
|
7676 |
+
},
|
7677 |
+
{
|
7678 |
+
"epoch": 7.863554757630162,
|
7679 |
+
"grad_norm": 0.003887078259140253,
|
7680 |
+
"learning_rate": 1.1025037227340712e-05,
|
7681 |
+
"loss": 0.0,
|
7682 |
+
"step": 1095
|
7683 |
+
},
|
7684 |
+
{
|
7685 |
+
"epoch": 7.870736086175943,
|
7686 |
+
"grad_norm": 0.002333530457690358,
|
7687 |
+
"learning_rate": 1.0953991657738088e-05,
|
7688 |
+
"loss": 0.0,
|
7689 |
+
"step": 1096
|
7690 |
+
},
|
7691 |
+
{
|
7692 |
+
"epoch": 7.877917414721724,
|
7693 |
+
"grad_norm": 0.0037893233820796013,
|
7694 |
+
"learning_rate": 1.0883147568088592e-05,
|
7695 |
+
"loss": 0.0,
|
7696 |
+
"step": 1097
|
7697 |
+
},
|
7698 |
+
{
|
7699 |
+
"epoch": 7.885098743267505,
|
7700 |
+
"grad_norm": 0.0038676662370562553,
|
7701 |
+
"learning_rate": 1.0812505323952344e-05,
|
7702 |
+
"loss": 0.0,
|
7703 |
+
"step": 1098
|
7704 |
+
},
|
7705 |
+
{
|
7706 |
+
"epoch": 7.8922800718132855,
|
7707 |
+
"grad_norm": 0.010742923244833946,
|
7708 |
+
"learning_rate": 1.0742065289847942e-05,
|
7709 |
+
"loss": 0.0001,
|
7710 |
+
"step": 1099
|
7711 |
+
},
|
7712 |
+
{
|
7713 |
+
"epoch": 7.899461400359066,
|
7714 |
+
"grad_norm": 0.0047193365171551704,
|
7715 |
+
"learning_rate": 1.0671827829250585e-05,
|
7716 |
+
"loss": 0.0,
|
7717 |
+
"step": 1100
|
7718 |
+
},
|
7719 |
+
{
|
7720 |
+
"epoch": 7.906642728904847,
|
7721 |
+
"grad_norm": 0.0015092804096639156,
|
7722 |
+
"learning_rate": 1.0601793304590124e-05,
|
7723 |
+
"loss": 0.0,
|
7724 |
+
"step": 1101
|
7725 |
+
},
|
7726 |
+
{
|
7727 |
+
"epoch": 7.913824057450628,
|
7728 |
+
"grad_norm": 0.0028412560932338238,
|
7729 |
+
"learning_rate": 1.0531962077249313e-05,
|
7730 |
+
"loss": 0.0,
|
7731 |
+
"step": 1102
|
7732 |
+
},
|
7733 |
+
{
|
7734 |
+
"epoch": 7.921005385996409,
|
7735 |
+
"grad_norm": 0.0088339289650321,
|
7736 |
+
"learning_rate": 1.0462334507561821e-05,
|
7737 |
+
"loss": 0.0001,
|
7738 |
+
"step": 1103
|
7739 |
+
},
|
7740 |
+
{
|
7741 |
+
"epoch": 7.92818671454219,
|
7742 |
+
"grad_norm": 0.0008142027072608471,
|
7743 |
+
"learning_rate": 1.0392910954810454e-05,
|
7744 |
+
"loss": 0.0,
|
7745 |
+
"step": 1104
|
7746 |
+
},
|
7747 |
+
{
|
7748 |
+
"epoch": 7.935368043087971,
|
7749 |
+
"grad_norm": 0.005558829288929701,
|
7750 |
+
"learning_rate": 1.0323691777225285e-05,
|
7751 |
+
"loss": 0.0001,
|
7752 |
+
"step": 1105
|
7753 |
+
},
|
7754 |
+
{
|
7755 |
+
"epoch": 7.942549371633753,
|
7756 |
+
"grad_norm": 0.013948008418083191,
|
7757 |
+
"learning_rate": 1.0254677331981822e-05,
|
7758 |
+
"loss": 0.0001,
|
7759 |
+
"step": 1106
|
7760 |
+
},
|
7761 |
+
{
|
7762 |
+
"epoch": 7.949730700179533,
|
7763 |
+
"grad_norm": 0.00884335208684206,
|
7764 |
+
"learning_rate": 1.0185867975199047e-05,
|
7765 |
+
"loss": 0.0001,
|
7766 |
+
"step": 1107
|
7767 |
+
},
|
7768 |
+
{
|
7769 |
+
"epoch": 7.956912028725315,
|
7770 |
+
"grad_norm": 0.001679787179455161,
|
7771 |
+
"learning_rate": 1.0117264061937775e-05,
|
7772 |
+
"loss": 0.0,
|
7773 |
+
"step": 1108
|
7774 |
+
},
|
7775 |
+
{
|
7776 |
+
"epoch": 7.9640933572710955,
|
7777 |
+
"grad_norm": 0.00992383249104023,
|
7778 |
+
"learning_rate": 1.0048865946198683e-05,
|
7779 |
+
"loss": 0.0001,
|
7780 |
+
"step": 1109
|
7781 |
+
},
|
7782 |
+
{
|
7783 |
+
"epoch": 7.971274685816876,
|
7784 |
+
"grad_norm": 0.0046890489757061005,
|
7785 |
+
"learning_rate": 9.980673980920491e-06,
|
7786 |
+
"loss": 0.0,
|
7787 |
+
"step": 1110
|
7788 |
+
},
|
7789 |
+
{
|
7790 |
+
"epoch": 7.978456014362657,
|
7791 |
+
"grad_norm": 0.006373502779752016,
|
7792 |
+
"learning_rate": 9.91268851797822e-06,
|
7793 |
+
"loss": 0.0001,
|
7794 |
+
"step": 1111
|
7795 |
+
},
|
7796 |
+
{
|
7797 |
+
"epoch": 7.985637342908438,
|
7798 |
+
"grad_norm": 0.005302777048200369,
|
7799 |
+
"learning_rate": 9.844909908181283e-06,
|
7800 |
+
"loss": 0.0001,
|
7801 |
+
"step": 1112
|
7802 |
+
},
|
7803 |
+
{
|
7804 |
+
"epoch": 7.992818671454219,
|
7805 |
+
"grad_norm": 0.03128412365913391,
|
7806 |
+
"learning_rate": 9.777338501271728e-06,
|
7807 |
+
"loss": 0.0003,
|
7808 |
+
"step": 1113
|
7809 |
+
},
|
7810 |
+
{
|
7811 |
+
"epoch": 8.0,
|
7812 |
+
"grad_norm": 0.0027070040814578533,
|
7813 |
+
"learning_rate": 9.709974645922432e-06,
|
7814 |
+
"loss": 0.0001,
|
7815 |
+
"step": 1114
|
7816 |
+
},
|
7817 |
+
{
|
7818 |
+
"epoch": 8.007181328545782,
|
7819 |
+
"grad_norm": 0.002355156932026148,
|
7820 |
+
"learning_rate": 9.642818689735307e-06,
|
7821 |
+
"loss": 0.0,
|
7822 |
+
"step": 1115
|
7823 |
+
},
|
7824 |
+
{
|
7825 |
+
"epoch": 8.014362657091562,
|
7826 |
+
"grad_norm": 0.0027824167627841234,
|
7827 |
+
"learning_rate": 9.575870979239454e-06,
|
7828 |
+
"loss": 0.0001,
|
7829 |
+
"step": 1116
|
7830 |
+
},
|
7831 |
+
{
|
7832 |
+
"epoch": 8.021543985637344,
|
7833 |
+
"grad_norm": 0.0012428781483322382,
|
7834 |
+
"learning_rate": 9.509131859889425e-06,
|
7835 |
+
"loss": 0.0,
|
7836 |
+
"step": 1117
|
7837 |
+
},
|
7838 |
+
{
|
7839 |
+
"epoch": 8.028725314183124,
|
7840 |
+
"grad_norm": 0.005846510641276836,
|
7841 |
+
"learning_rate": 9.442601676063456e-06,
|
7842 |
+
"loss": 0.0001,
|
7843 |
+
"step": 1118
|
7844 |
+
},
|
7845 |
+
{
|
7846 |
+
"epoch": 8.035906642728905,
|
7847 |
+
"grad_norm": 0.004024711903184652,
|
7848 |
+
"learning_rate": 9.376280771061612e-06,
|
7849 |
+
"loss": 0.0,
|
7850 |
+
"step": 1119
|
7851 |
+
},
|
7852 |
+
{
|
7853 |
+
"epoch": 8.043087971274685,
|
7854 |
+
"grad_norm": 0.004356505814939737,
|
7855 |
+
"learning_rate": 9.310169487104131e-06,
|
7856 |
+
"loss": 0.0,
|
7857 |
+
"step": 1120
|
7858 |
+
},
|
7859 |
+
{
|
7860 |
+
"epoch": 8.050269299820467,
|
7861 |
+
"grad_norm": 0.003750949166715145,
|
7862 |
+
"learning_rate": 9.244268165329533e-06,
|
7863 |
+
"loss": 0.0001,
|
7864 |
+
"step": 1121
|
7865 |
+
},
|
7866 |
+
{
|
7867 |
+
"epoch": 8.057450628366247,
|
7868 |
+
"grad_norm": 0.002422786783427,
|
7869 |
+
"learning_rate": 9.178577145792938e-06,
|
7870 |
+
"loss": 0.0,
|
7871 |
+
"step": 1122
|
7872 |
+
},
|
7873 |
+
{
|
7874 |
+
"epoch": 8.064631956912029,
|
7875 |
+
"grad_norm": 0.004002750385552645,
|
7876 |
+
"learning_rate": 9.113096767464302e-06,
|
7877 |
+
"loss": 0.0001,
|
7878 |
+
"step": 1123
|
7879 |
+
},
|
7880 |
+
{
|
7881 |
+
"epoch": 8.071813285457809,
|
7882 |
+
"grad_norm": 0.005756510887295008,
|
7883 |
+
"learning_rate": 9.047827368226663e-06,
|
7884 |
+
"loss": 0.0001,
|
7885 |
+
"step": 1124
|
7886 |
+
},
|
7887 |
+
{
|
7888 |
+
"epoch": 8.07899461400359,
|
7889 |
+
"grad_norm": 0.0009316102950833738,
|
7890 |
+
"learning_rate": 8.982769284874387e-06,
|
7891 |
+
"loss": 0.0,
|
7892 |
+
"step": 1125
|
7893 |
+
},
|
7894 |
+
{
|
7895 |
+
"epoch": 8.08617594254937,
|
7896 |
+
"grad_norm": 0.0043763830326497555,
|
7897 |
+
"learning_rate": 8.917922853111405e-06,
|
7898 |
+
"loss": 0.0,
|
7899 |
+
"step": 1126
|
7900 |
+
},
|
7901 |
+
{
|
7902 |
+
"epoch": 8.093357271095153,
|
7903 |
+
"grad_norm": 0.0021101904567331076,
|
7904 |
+
"learning_rate": 8.85328840754957e-06,
|
7905 |
+
"loss": 0.0,
|
7906 |
+
"step": 1127
|
7907 |
+
},
|
7908 |
+
{
|
7909 |
+
"epoch": 8.100538599640933,
|
7910 |
+
"grad_norm": 0.0004592133336700499,
|
7911 |
+
"learning_rate": 8.78886628170682e-06,
|
7912 |
+
"loss": 0.0,
|
7913 |
+
"step": 1128
|
7914 |
+
},
|
7915 |
+
{
|
7916 |
+
"epoch": 8.107719928186714,
|
7917 |
+
"grad_norm": 0.00474977632984519,
|
7918 |
+
"learning_rate": 8.724656808005555e-06,
|
7919 |
+
"loss": 0.0001,
|
7920 |
+
"step": 1129
|
7921 |
+
},
|
7922 |
+
{
|
7923 |
+
"epoch": 8.114901256732496,
|
7924 |
+
"grad_norm": 0.0035962453112006187,
|
7925 |
+
"learning_rate": 8.660660317770841e-06,
|
7926 |
+
"loss": 0.0001,
|
7927 |
+
"step": 1130
|
7928 |
+
},
|
7929 |
+
{
|
7930 |
+
"epoch": 8.122082585278276,
|
7931 |
+
"grad_norm": 0.0009593448485247791,
|
7932 |
+
"learning_rate": 8.596877141228732e-06,
|
7933 |
+
"loss": 0.0,
|
7934 |
+
"step": 1131
|
7935 |
+
},
|
7936 |
+
{
|
7937 |
+
"epoch": 8.129263913824058,
|
7938 |
+
"grad_norm": 0.011563223786652088,
|
7939 |
+
"learning_rate": 8.533307607504597e-06,
|
7940 |
+
"loss": 0.0001,
|
7941 |
+
"step": 1132
|
7942 |
+
},
|
7943 |
+
{
|
7944 |
+
"epoch": 8.136445242369838,
|
7945 |
+
"grad_norm": 0.00835861824452877,
|
7946 |
+
"learning_rate": 8.469952044621398e-06,
|
7947 |
+
"loss": 0.0001,
|
7948 |
+
"step": 1133
|
7949 |
+
},
|
7950 |
+
{
|
7951 |
+
"epoch": 8.14362657091562,
|
7952 |
+
"grad_norm": 0.0016383846523240209,
|
7953 |
+
"learning_rate": 8.406810779497965e-06,
|
7954 |
+
"loss": 0.0,
|
7955 |
+
"step": 1134
|
7956 |
+
},
|
7957 |
+
{
|
7958 |
+
"epoch": 8.1508078994614,
|
7959 |
+
"grad_norm": 0.0017268586670979857,
|
7960 |
+
"learning_rate": 8.343884137947333e-06,
|
7961 |
+
"loss": 0.0,
|
7962 |
+
"step": 1135
|
7963 |
+
},
|
7964 |
+
{
|
7965 |
+
"epoch": 8.157989228007182,
|
7966 |
+
"grad_norm": 0.000591074931435287,
|
7967 |
+
"learning_rate": 8.281172444675112e-06,
|
7968 |
+
"loss": 0.0,
|
7969 |
+
"step": 1136
|
7970 |
+
},
|
7971 |
+
{
|
7972 |
+
"epoch": 8.165170556552962,
|
7973 |
+
"grad_norm": 0.002751863794401288,
|
7974 |
+
"learning_rate": 8.218676023277705e-06,
|
7975 |
+
"loss": 0.0,
|
7976 |
+
"step": 1137
|
7977 |
+
},
|
7978 |
+
{
|
7979 |
+
"epoch": 8.172351885098744,
|
7980 |
+
"grad_norm": 0.0022041266784071922,
|
7981 |
+
"learning_rate": 8.15639519624075e-06,
|
7982 |
+
"loss": 0.0,
|
7983 |
+
"step": 1138
|
7984 |
+
},
|
7985 |
+
{
|
7986 |
+
"epoch": 8.179533213644524,
|
7987 |
+
"grad_norm": 0.0013457705499604344,
|
7988 |
+
"learning_rate": 8.094330284937384e-06,
|
7989 |
+
"loss": 0.0,
|
7990 |
+
"step": 1139
|
7991 |
+
},
|
7992 |
+
{
|
7993 |
+
"epoch": 8.186714542190305,
|
7994 |
+
"grad_norm": 0.013541612774133682,
|
7995 |
+
"learning_rate": 8.032481609626574e-06,
|
7996 |
+
"loss": 0.0001,
|
7997 |
+
"step": 1140
|
7998 |
+
},
|
7999 |
+
{
|
8000 |
+
"epoch": 8.193895870736085,
|
8001 |
+
"grad_norm": 0.01249165553599596,
|
8002 |
+
"learning_rate": 7.970849489451548e-06,
|
8003 |
+
"loss": 0.0001,
|
8004 |
+
"step": 1141
|
8005 |
+
},
|
8006 |
+
{
|
8007 |
+
"epoch": 8.201077199281867,
|
8008 |
+
"grad_norm": 0.0025624497793614864,
|
8009 |
+
"learning_rate": 7.909434242438079e-06,
|
8010 |
+
"loss": 0.0,
|
8011 |
+
"step": 1142
|
8012 |
+
},
|
8013 |
+
{
|
8014 |
+
"epoch": 8.208258527827649,
|
8015 |
+
"grad_norm": 0.019143685698509216,
|
8016 |
+
"learning_rate": 7.848236185492835e-06,
|
8017 |
+
"loss": 0.0002,
|
8018 |
+
"step": 1143
|
8019 |
+
},
|
8020 |
+
{
|
8021 |
+
"epoch": 8.215439856373429,
|
8022 |
+
"grad_norm": 0.004778710193932056,
|
8023 |
+
"learning_rate": 7.787255634401785e-06,
|
8024 |
+
"loss": 0.0001,
|
8025 |
+
"step": 1144
|
8026 |
+
},
|
8027 |
+
{
|
8028 |
+
"epoch": 8.22262118491921,
|
8029 |
+
"grad_norm": 0.006595983169972897,
|
8030 |
+
"learning_rate": 7.726492903828574e-06,
|
8031 |
+
"loss": 0.0001,
|
8032 |
+
"step": 1145
|
8033 |
+
},
|
8034 |
+
{
|
8035 |
+
"epoch": 8.22980251346499,
|
8036 |
+
"grad_norm": 0.0033770056907087564,
|
8037 |
+
"learning_rate": 7.665948307312837e-06,
|
8038 |
+
"loss": 0.0,
|
8039 |
+
"step": 1146
|
8040 |
+
},
|
8041 |
+
{
|
8042 |
+
"epoch": 8.236983842010773,
|
8043 |
+
"grad_norm": 0.0023582743015140295,
|
8044 |
+
"learning_rate": 7.605622157268655e-06,
|
8045 |
+
"loss": 0.0,
|
8046 |
+
"step": 1147
|
8047 |
+
},
|
8048 |
+
{
|
8049 |
+
"epoch": 8.244165170556553,
|
8050 |
+
"grad_norm": 0.0007562777609564364,
|
8051 |
+
"learning_rate": 7.545514764982931e-06,
|
8052 |
+
"loss": 0.0,
|
8053 |
+
"step": 1148
|
8054 |
+
},
|
8055 |
+
{
|
8056 |
+
"epoch": 8.251346499102334,
|
8057 |
+
"grad_norm": 0.005157983396202326,
|
8058 |
+
"learning_rate": 7.485626440613686e-06,
|
8059 |
+
"loss": 0.0001,
|
8060 |
+
"step": 1149
|
8061 |
+
},
|
8062 |
+
{
|
8063 |
+
"epoch": 8.258527827648114,
|
8064 |
+
"grad_norm": 0.004036571364849806,
|
8065 |
+
"learning_rate": 7.425957493188601e-06,
|
8066 |
+
"loss": 0.0001,
|
8067 |
+
"step": 1150
|
8068 |
+
},
|
8069 |
+
{
|
8070 |
+
"epoch": 8.265709156193896,
|
8071 |
+
"grad_norm": 0.005852349102497101,
|
8072 |
+
"learning_rate": 7.366508230603364e-06,
|
8073 |
+
"loss": 0.0001,
|
8074 |
+
"step": 1151
|
8075 |
+
},
|
8076 |
+
{
|
8077 |
+
"epoch": 8.272890484739676,
|
8078 |
+
"grad_norm": 0.0009574262658134103,
|
8079 |
+
"learning_rate": 7.30727895962004e-06,
|
8080 |
+
"loss": 0.0,
|
8081 |
+
"step": 1152
|
8082 |
+
},
|
8083 |
+
{
|
8084 |
+
"epoch": 8.280071813285458,
|
8085 |
+
"grad_norm": 0.008616548962891102,
|
8086 |
+
"learning_rate": 7.248269985865513e-06,
|
8087 |
+
"loss": 0.0001,
|
8088 |
+
"step": 1153
|
8089 |
+
},
|
8090 |
+
{
|
8091 |
+
"epoch": 8.287253141831238,
|
8092 |
+
"grad_norm": 0.0030366776045411825,
|
8093 |
+
"learning_rate": 7.189481613829979e-06,
|
8094 |
+
"loss": 0.0,
|
8095 |
+
"step": 1154
|
8096 |
+
},
|
8097 |
+
{
|
8098 |
+
"epoch": 8.29443447037702,
|
8099 |
+
"grad_norm": 0.003713611513376236,
|
8100 |
+
"learning_rate": 7.130914146865248e-06,
|
8101 |
+
"loss": 0.0,
|
8102 |
+
"step": 1155
|
8103 |
+
},
|
8104 |
+
{
|
8105 |
+
"epoch": 8.3016157989228,
|
8106 |
+
"grad_norm": 0.004273141734302044,
|
8107 |
+
"learning_rate": 7.072567887183279e-06,
|
8108 |
+
"loss": 0.0001,
|
8109 |
+
"step": 1156
|
8110 |
+
},
|
8111 |
+
{
|
8112 |
+
"epoch": 8.308797127468582,
|
8113 |
+
"grad_norm": 0.002239127177745104,
|
8114 |
+
"learning_rate": 7.014443135854615e-06,
|
8115 |
+
"loss": 0.0,
|
8116 |
+
"step": 1157
|
8117 |
+
},
|
8118 |
+
{
|
8119 |
+
"epoch": 8.315978456014363,
|
8120 |
+
"grad_norm": 0.00268027326092124,
|
8121 |
+
"learning_rate": 6.956540192806721e-06,
|
8122 |
+
"loss": 0.0,
|
8123 |
+
"step": 1158
|
8124 |
+
},
|
8125 |
+
{
|
8126 |
+
"epoch": 8.323159784560143,
|
8127 |
+
"grad_norm": 0.0018624209333211184,
|
8128 |
+
"learning_rate": 6.898859356822585e-06,
|
8129 |
+
"loss": 0.0,
|
8130 |
+
"step": 1159
|
8131 |
+
},
|
8132 |
+
{
|
8133 |
+
"epoch": 8.330341113105925,
|
8134 |
+
"grad_norm": 0.0014236210845410824,
|
8135 |
+
"learning_rate": 6.841400925539104e-06,
|
8136 |
+
"loss": 0.0,
|
8137 |
+
"step": 1160
|
8138 |
+
},
|
8139 |
+
{
|
8140 |
+
"epoch": 8.337522441651705,
|
8141 |
+
"grad_norm": 0.0037910134997218847,
|
8142 |
+
"learning_rate": 6.784165195445518e-06,
|
8143 |
+
"loss": 0.0001,
|
8144 |
+
"step": 1161
|
8145 |
+
},
|
8146 |
+
{
|
8147 |
+
"epoch": 8.344703770197487,
|
8148 |
+
"grad_norm": 0.0016732742078602314,
|
8149 |
+
"learning_rate": 6.727152461881925e-06,
|
8150 |
+
"loss": 0.0,
|
8151 |
+
"step": 1162
|
8152 |
+
},
|
8153 |
+
{
|
8154 |
+
"epoch": 8.351885098743267,
|
8155 |
+
"grad_norm": 0.0035439447965472937,
|
8156 |
+
"learning_rate": 6.670363019037779e-06,
|
8157 |
+
"loss": 0.0,
|
8158 |
+
"step": 1163
|
8159 |
+
},
|
8160 |
+
{
|
8161 |
+
"epoch": 8.359066427289049,
|
8162 |
+
"grad_norm": 0.0007749520009383559,
|
8163 |
+
"learning_rate": 6.613797159950286e-06,
|
8164 |
+
"loss": 0.0,
|
8165 |
+
"step": 1164
|
8166 |
+
},
|
8167 |
+
{
|
8168 |
+
"epoch": 8.366247755834829,
|
8169 |
+
"grad_norm": 0.0013621584512293339,
|
8170 |
+
"learning_rate": 6.5574551765029855e-06,
|
8171 |
+
"loss": 0.0,
|
8172 |
+
"step": 1165
|
8173 |
+
},
|
8174 |
+
{
|
8175 |
+
"epoch": 8.37342908438061,
|
8176 |
+
"grad_norm": 0.007024025544524193,
|
8177 |
+
"learning_rate": 6.501337359424214e-06,
|
8178 |
+
"loss": 0.0001,
|
8179 |
+
"step": 1166
|
8180 |
+
},
|
8181 |
+
{
|
8182 |
+
"epoch": 8.38061041292639,
|
8183 |
+
"grad_norm": 0.0021991136018186808,
|
8184 |
+
"learning_rate": 6.445443998285533e-06,
|
8185 |
+
"loss": 0.0,
|
8186 |
+
"step": 1167
|
8187 |
+
},
|
8188 |
+
{
|
8189 |
+
"epoch": 8.387791741472173,
|
8190 |
+
"grad_norm": 0.0009170992416329682,
|
8191 |
+
"learning_rate": 6.389775381500351e-06,
|
8192 |
+
"loss": 0.0,
|
8193 |
+
"step": 1168
|
8194 |
+
},
|
8195 |
+
{
|
8196 |
+
"epoch": 8.394973070017953,
|
8197 |
+
"grad_norm": 0.0004729504871647805,
|
8198 |
+
"learning_rate": 6.334331796322384e-06,
|
8199 |
+
"loss": 0.0,
|
8200 |
+
"step": 1169
|
8201 |
+
},
|
8202 |
+
{
|
8203 |
+
"epoch": 8.402154398563734,
|
8204 |
+
"grad_norm": 0.0016273808432742953,
|
8205 |
+
"learning_rate": 6.279113528844127e-06,
|
8206 |
+
"loss": 0.0,
|
8207 |
+
"step": 1170
|
8208 |
+
},
|
8209 |
+
{
|
8210 |
+
"epoch": 8.409335727109514,
|
8211 |
+
"grad_norm": 0.0019783461466431618,
|
8212 |
+
"learning_rate": 6.22412086399547e-06,
|
8213 |
+
"loss": 0.0,
|
8214 |
+
"step": 1171
|
8215 |
+
},
|
8216 |
+
{
|
8217 |
+
"epoch": 8.416517055655296,
|
8218 |
+
"grad_norm": 0.00046266100252978504,
|
8219 |
+
"learning_rate": 6.169354085542145e-06,
|
8220 |
+
"loss": 0.0,
|
8221 |
+
"step": 1172
|
8222 |
+
},
|
8223 |
+
{
|
8224 |
+
"epoch": 8.423698384201078,
|
8225 |
+
"grad_norm": 0.004233096726238728,
|
8226 |
+
"learning_rate": 6.114813476084291e-06,
|
8227 |
+
"loss": 0.0,
|
8228 |
+
"step": 1173
|
8229 |
+
},
|
8230 |
+
{
|
8231 |
+
"epoch": 8.430879712746858,
|
8232 |
+
"grad_norm": 0.007232322357594967,
|
8233 |
+
"learning_rate": 6.060499317055024e-06,
|
8234 |
+
"loss": 0.0001,
|
8235 |
+
"step": 1174
|
8236 |
+
},
|
8237 |
+
{
|
8238 |
+
"epoch": 8.43806104129264,
|
8239 |
+
"grad_norm": 0.0054048034362494946,
|
8240 |
+
"learning_rate": 6.0064118887189825e-06,
|
8241 |
+
"loss": 0.0001,
|
8242 |
+
"step": 1175
|
8243 |
+
},
|
8244 |
+
{
|
8245 |
+
"epoch": 8.44524236983842,
|
8246 |
+
"grad_norm": 0.012967641465365887,
|
8247 |
+
"learning_rate": 5.95255147017078e-06,
|
8248 |
+
"loss": 0.0001,
|
8249 |
+
"step": 1176
|
8250 |
+
},
|
8251 |
+
{
|
8252 |
+
"epoch": 8.452423698384202,
|
8253 |
+
"grad_norm": 0.003502080449834466,
|
8254 |
+
"learning_rate": 5.898918339333714e-06,
|
8255 |
+
"loss": 0.0001,
|
8256 |
+
"step": 1177
|
8257 |
+
},
|
8258 |
+
{
|
8259 |
+
"epoch": 8.459605026929982,
|
8260 |
+
"grad_norm": 0.0027490684296935797,
|
8261 |
+
"learning_rate": 5.845512772958256e-06,
|
8262 |
+
"loss": 0.0,
|
8263 |
+
"step": 1178
|
8264 |
+
},
|
8265 |
+
{
|
8266 |
+
"epoch": 8.466786355475763,
|
8267 |
+
"grad_norm": 0.002882092259824276,
|
8268 |
+
"learning_rate": 5.792335046620601e-06,
|
8269 |
+
"loss": 0.0,
|
8270 |
+
"step": 1179
|
8271 |
+
},
|
8272 |
+
{
|
8273 |
+
"epoch": 8.473967684021543,
|
8274 |
+
"grad_norm": 0.0011095460504293442,
|
8275 |
+
"learning_rate": 5.739385434721295e-06,
|
8276 |
+
"loss": 0.0,
|
8277 |
+
"step": 1180
|
8278 |
+
},
|
8279 |
+
{
|
8280 |
+
"epoch": 8.481149012567325,
|
8281 |
+
"grad_norm": 0.01054337527602911,
|
8282 |
+
"learning_rate": 5.686664210483794e-06,
|
8283 |
+
"loss": 0.0001,
|
8284 |
+
"step": 1181
|
8285 |
+
},
|
8286 |
+
{
|
8287 |
+
"epoch": 8.488330341113105,
|
8288 |
+
"grad_norm": 0.0006395714590325952,
|
8289 |
+
"learning_rate": 5.6341716459530516e-06,
|
8290 |
+
"loss": 0.0,
|
8291 |
+
"step": 1182
|
8292 |
+
},
|
8293 |
+
{
|
8294 |
+
"epoch": 8.495511669658887,
|
8295 |
+
"grad_norm": 0.0024796442594379187,
|
8296 |
+
"learning_rate": 5.581908011994131e-06,
|
8297 |
+
"loss": 0.0,
|
8298 |
+
"step": 1183
|
8299 |
+
},
|
8300 |
+
{
|
8301 |
+
"epoch": 8.502692998204667,
|
8302 |
+
"grad_norm": 0.009811460971832275,
|
8303 |
+
"learning_rate": 5.529873578290818e-06,
|
8304 |
+
"loss": 0.0001,
|
8305 |
+
"step": 1184
|
8306 |
+
},
|
8307 |
+
{
|
8308 |
+
"epoch": 8.509874326750449,
|
8309 |
+
"grad_norm": 0.0020709994714707136,
|
8310 |
+
"learning_rate": 5.478068613344151e-06,
|
8311 |
+
"loss": 0.0,
|
8312 |
+
"step": 1185
|
8313 |
+
},
|
8314 |
+
{
|
8315 |
+
"epoch": 8.517055655296229,
|
8316 |
+
"grad_norm": 0.003330863080918789,
|
8317 |
+
"learning_rate": 5.426493384471154e-06,
|
8318 |
+
"loss": 0.0001,
|
8319 |
+
"step": 1186
|
8320 |
+
},
|
8321 |
+
{
|
8322 |
+
"epoch": 8.52423698384201,
|
8323 |
+
"grad_norm": 0.005057830363512039,
|
8324 |
+
"learning_rate": 5.375148157803383e-06,
|
8325 |
+
"loss": 0.0001,
|
8326 |
+
"step": 1187
|
8327 |
+
},
|
8328 |
+
{
|
8329 |
+
"epoch": 8.531418312387792,
|
8330 |
+
"grad_norm": 0.002186673227697611,
|
8331 |
+
"learning_rate": 5.324033198285544e-06,
|
8332 |
+
"loss": 0.0,
|
8333 |
+
"step": 1188
|
8334 |
+
},
|
8335 |
+
{
|
8336 |
+
"epoch": 8.538599640933572,
|
8337 |
+
"grad_norm": 0.014916039071977139,
|
8338 |
+
"learning_rate": 5.2731487696741854e-06,
|
8339 |
+
"loss": 0.0001,
|
8340 |
+
"step": 1189
|
8341 |
+
},
|
8342 |
+
{
|
8343 |
+
"epoch": 8.545780969479354,
|
8344 |
+
"grad_norm": 0.0025849922094494104,
|
8345 |
+
"learning_rate": 5.222495134536271e-06,
|
8346 |
+
"loss": 0.0,
|
8347 |
+
"step": 1190
|
8348 |
+
},
|
8349 |
+
{
|
8350 |
+
"epoch": 8.552962298025134,
|
8351 |
+
"grad_norm": 0.003023435128852725,
|
8352 |
+
"learning_rate": 5.172072554247859e-06,
|
8353 |
+
"loss": 0.0,
|
8354 |
+
"step": 1191
|
8355 |
+
},
|
8356 |
+
{
|
8357 |
+
"epoch": 8.560143626570916,
|
8358 |
+
"grad_norm": 0.001386603806167841,
|
8359 |
+
"learning_rate": 5.121881288992758e-06,
|
8360 |
+
"loss": 0.0,
|
8361 |
+
"step": 1192
|
8362 |
+
},
|
8363 |
+
{
|
8364 |
+
"epoch": 8.567324955116696,
|
8365 |
+
"grad_norm": 0.0004793599364347756,
|
8366 |
+
"learning_rate": 5.071921597761198e-06,
|
8367 |
+
"loss": 0.0,
|
8368 |
+
"step": 1193
|
8369 |
+
},
|
8370 |
+
{
|
8371 |
+
"epoch": 8.574506283662478,
|
8372 |
+
"grad_norm": 0.0028707697056233883,
|
8373 |
+
"learning_rate": 5.0221937383484294e-06,
|
8374 |
+
"loss": 0.0,
|
8375 |
+
"step": 1194
|
8376 |
+
},
|
8377 |
+
{
|
8378 |
+
"epoch": 8.581687612208258,
|
8379 |
+
"grad_norm": 0.0010357374558225274,
|
8380 |
+
"learning_rate": 4.9726979673534454e-06,
|
8381 |
+
"loss": 0.0,
|
8382 |
+
"step": 1195
|
8383 |
+
},
|
8384 |
+
{
|
8385 |
+
"epoch": 8.58886894075404,
|
8386 |
+
"grad_norm": 0.0010934651363641024,
|
8387 |
+
"learning_rate": 4.923434540177679e-06,
|
8388 |
+
"loss": 0.0,
|
8389 |
+
"step": 1196
|
8390 |
+
},
|
8391 |
+
{
|
8392 |
+
"epoch": 8.59605026929982,
|
8393 |
+
"grad_norm": 0.0008548993500880897,
|
8394 |
+
"learning_rate": 4.874403711023612e-06,
|
8395 |
+
"loss": 0.0,
|
8396 |
+
"step": 1197
|
8397 |
+
},
|
8398 |
+
{
|
8399 |
+
"epoch": 8.603231597845602,
|
8400 |
+
"grad_norm": 0.008287246339023113,
|
8401 |
+
"learning_rate": 4.825605732893545e-06,
|
8402 |
+
"loss": 0.0001,
|
8403 |
+
"step": 1198
|
8404 |
+
},
|
8405 |
+
{
|
8406 |
+
"epoch": 8.610412926391383,
|
8407 |
+
"grad_norm": 0.0029302104376256466,
|
8408 |
+
"learning_rate": 4.777040857588222e-06,
|
8409 |
+
"loss": 0.0,
|
8410 |
+
"step": 1199
|
8411 |
+
},
|
8412 |
+
{
|
8413 |
+
"epoch": 8.617594254937163,
|
8414 |
+
"grad_norm": 0.0007794447592459619,
|
8415 |
+
"learning_rate": 4.728709335705561e-06,
|
8416 |
+
"loss": 0.0,
|
8417 |
+
"step": 1200
|
8418 |
+
},
|
8419 |
+
{
|
8420 |
+
"epoch": 8.624775583482945,
|
8421 |
+
"grad_norm": 0.001528155873529613,
|
8422 |
+
"learning_rate": 4.68061141663938e-06,
|
8423 |
+
"loss": 0.0,
|
8424 |
+
"step": 1201
|
8425 |
+
},
|
8426 |
+
{
|
8427 |
+
"epoch": 8.631956912028725,
|
8428 |
+
"grad_norm": 0.003254116512835026,
|
8429 |
+
"learning_rate": 4.632747348578092e-06,
|
8430 |
+
"loss": 0.0001,
|
8431 |
+
"step": 1202
|
8432 |
+
},
|
8433 |
+
{
|
8434 |
+
"epoch": 8.639138240574507,
|
8435 |
+
"grad_norm": 0.0019204439595341682,
|
8436 |
+
"learning_rate": 4.585117378503412e-06,
|
8437 |
+
"loss": 0.0,
|
8438 |
+
"step": 1203
|
8439 |
+
},
|
8440 |
+
{
|
8441 |
+
"epoch": 8.646319569120287,
|
8442 |
+
"grad_norm": 0.0011220823507755995,
|
8443 |
+
"learning_rate": 4.537721752189078e-06,
|
8444 |
+
"loss": 0.0,
|
8445 |
+
"step": 1204
|
8446 |
+
},
|
8447 |
+
{
|
8448 |
+
"epoch": 8.653500897666069,
|
8449 |
+
"grad_norm": 0.009151519276201725,
|
8450 |
+
"learning_rate": 4.4905607141996375e-06,
|
8451 |
+
"loss": 0.0001,
|
8452 |
+
"step": 1205
|
8453 |
+
},
|
8454 |
+
{
|
8455 |
+
"epoch": 8.660682226211849,
|
8456 |
+
"grad_norm": 0.0016937977634370327,
|
8457 |
+
"learning_rate": 4.443634507889105e-06,
|
8458 |
+
"loss": 0.0,
|
8459 |
+
"step": 1206
|
8460 |
+
},
|
8461 |
+
{
|
8462 |
+
"epoch": 8.66786355475763,
|
8463 |
+
"grad_norm": 0.002167623955756426,
|
8464 |
+
"learning_rate": 4.3969433753997975e-06,
|
8465 |
+
"loss": 0.0,
|
8466 |
+
"step": 1207
|
8467 |
+
},
|
8468 |
+
{
|
8469 |
+
"epoch": 8.67504488330341,
|
8470 |
+
"grad_norm": 0.0070151593536138535,
|
8471 |
+
"learning_rate": 4.350487557660993e-06,
|
8472 |
+
"loss": 0.0001,
|
8473 |
+
"step": 1208
|
8474 |
+
},
|
8475 |
+
{
|
8476 |
+
"epoch": 8.682226211849192,
|
8477 |
+
"grad_norm": 0.004574051592499018,
|
8478 |
+
"learning_rate": 4.304267294387737e-06,
|
8479 |
+
"loss": 0.0001,
|
8480 |
+
"step": 1209
|
8481 |
+
},
|
8482 |
+
{
|
8483 |
+
"epoch": 8.689407540394972,
|
8484 |
+
"grad_norm": 0.0009771317709237337,
|
8485 |
+
"learning_rate": 4.258282824079618e-06,
|
8486 |
+
"loss": 0.0,
|
8487 |
+
"step": 1210
|
8488 |
+
},
|
8489 |
+
{
|
8490 |
+
"epoch": 8.696588868940754,
|
8491 |
+
"grad_norm": 0.0029267631471157074,
|
8492 |
+
"learning_rate": 4.212534384019512e-06,
|
8493 |
+
"loss": 0.0001,
|
8494 |
+
"step": 1211
|
8495 |
+
},
|
8496 |
+
{
|
8497 |
+
"epoch": 8.703770197486534,
|
8498 |
+
"grad_norm": 0.0017605145694687963,
|
8499 |
+
"learning_rate": 4.167022210272347e-06,
|
8500 |
+
"loss": 0.0,
|
8501 |
+
"step": 1212
|
8502 |
+
},
|
8503 |
+
{
|
8504 |
+
"epoch": 8.710951526032316,
|
8505 |
+
"grad_norm": 0.008878316730260849,
|
8506 |
+
"learning_rate": 4.121746537683907e-06,
|
8507 |
+
"loss": 0.0001,
|
8508 |
+
"step": 1213
|
8509 |
+
},
|
8510 |
+
{
|
8511 |
+
"epoch": 8.718132854578098,
|
8512 |
+
"grad_norm": 0.010409415699541569,
|
8513 |
+
"learning_rate": 4.0767075998796314e-06,
|
8514 |
+
"loss": 0.0001,
|
8515 |
+
"step": 1214
|
8516 |
+
},
|
8517 |
+
{
|
8518 |
+
"epoch": 8.725314183123878,
|
8519 |
+
"grad_norm": 0.0004229078476782888,
|
8520 |
+
"learning_rate": 4.031905629263372e-06,
|
8521 |
+
"loss": 0.0,
|
8522 |
+
"step": 1215
|
8523 |
+
},
|
8524 |
+
{
|
8525 |
+
"epoch": 8.73249551166966,
|
8526 |
+
"grad_norm": 0.0005739431362599134,
|
8527 |
+
"learning_rate": 3.987340857016225e-06,
|
8528 |
+
"loss": 0.0,
|
8529 |
+
"step": 1216
|
8530 |
+
},
|
8531 |
+
{
|
8532 |
+
"epoch": 8.73967684021544,
|
8533 |
+
"grad_norm": 0.005821234546601772,
|
8534 |
+
"learning_rate": 3.943013513095345e-06,
|
8535 |
+
"loss": 0.0001,
|
8536 |
+
"step": 1217
|
8537 |
+
},
|
8538 |
+
{
|
8539 |
+
"epoch": 8.746858168761221,
|
8540 |
+
"grad_norm": 0.002494215965270996,
|
8541 |
+
"learning_rate": 3.898923826232687e-06,
|
8542 |
+
"loss": 0.0,
|
8543 |
+
"step": 1218
|
8544 |
+
},
|
8545 |
+
{
|
8546 |
+
"epoch": 8.754039497307001,
|
8547 |
+
"grad_norm": 0.005722567439079285,
|
8548 |
+
"learning_rate": 3.855072023933931e-06,
|
8549 |
+
"loss": 0.0001,
|
8550 |
+
"step": 1219
|
8551 |
+
},
|
8552 |
+
{
|
8553 |
+
"epoch": 8.761220825852783,
|
8554 |
+
"grad_norm": 0.0009568593231961131,
|
8555 |
+
"learning_rate": 3.8114583324772526e-06,
|
8556 |
+
"loss": 0.0,
|
8557 |
+
"step": 1220
|
8558 |
+
},
|
8559 |
+
{
|
8560 |
+
"epoch": 8.768402154398563,
|
8561 |
+
"grad_norm": 0.0028091880958527327,
|
8562 |
+
"learning_rate": 3.7680829769121406e-06,
|
8563 |
+
"loss": 0.0,
|
8564 |
+
"step": 1221
|
8565 |
+
},
|
8566 |
+
{
|
8567 |
+
"epoch": 8.775583482944345,
|
8568 |
+
"grad_norm": 0.004496587440371513,
|
8569 |
+
"learning_rate": 3.724946181058242e-06,
|
8570 |
+
"loss": 0.0001,
|
8571 |
+
"step": 1222
|
8572 |
+
},
|
8573 |
+
{
|
8574 |
+
"epoch": 8.782764811490125,
|
8575 |
+
"grad_norm": 0.001554902526549995,
|
8576 |
+
"learning_rate": 3.682048167504276e-06,
|
8577 |
+
"loss": 0.0,
|
8578 |
+
"step": 1223
|
8579 |
+
},
|
8580 |
+
{
|
8581 |
+
"epoch": 8.789946140035907,
|
8582 |
+
"grad_norm": 0.001350821927189827,
|
8583 |
+
"learning_rate": 3.6393891576067596e-06,
|
8584 |
+
"loss": 0.0,
|
8585 |
+
"step": 1224
|
8586 |
+
},
|
8587 |
+
{
|
8588 |
+
"epoch": 8.797127468581687,
|
8589 |
+
"grad_norm": 0.005168233532458544,
|
8590 |
+
"learning_rate": 3.596969371488995e-06,
|
8591 |
+
"loss": 0.0001,
|
8592 |
+
"step": 1225
|
8593 |
+
},
|
8594 |
+
{
|
8595 |
+
"epoch": 8.804308797127469,
|
8596 |
+
"grad_norm": 0.0025330630596727133,
|
8597 |
+
"learning_rate": 3.554789028039851e-06,
|
8598 |
+
"loss": 0.0,
|
8599 |
+
"step": 1226
|
8600 |
+
},
|
8601 |
+
{
|
8602 |
+
"epoch": 8.811490125673249,
|
8603 |
+
"grad_norm": 0.004454289097338915,
|
8604 |
+
"learning_rate": 3.512848344912628e-06,
|
8605 |
+
"loss": 0.0001,
|
8606 |
+
"step": 1227
|
8607 |
+
},
|
8608 |
+
{
|
8609 |
+
"epoch": 8.81867145421903,
|
8610 |
+
"grad_norm": 0.002888551214709878,
|
8611 |
+
"learning_rate": 3.4711475385240055e-06,
|
8612 |
+
"loss": 0.0,
|
8613 |
+
"step": 1228
|
8614 |
+
},
|
8615 |
+
{
|
8616 |
+
"epoch": 8.825852782764812,
|
8617 |
+
"grad_norm": 0.003765455214306712,
|
8618 |
+
"learning_rate": 3.4296868240528803e-06,
|
8619 |
+
"loss": 0.0001,
|
8620 |
+
"step": 1229
|
8621 |
+
},
|
8622 |
+
{
|
8623 |
+
"epoch": 8.833034111310592,
|
8624 |
+
"grad_norm": 0.010624225251376629,
|
8625 |
+
"learning_rate": 3.388466415439234e-06,
|
8626 |
+
"loss": 0.0001,
|
8627 |
+
"step": 1230
|
8628 |
+
},
|
8629 |
+
{
|
8630 |
+
"epoch": 8.840215439856374,
|
8631 |
+
"grad_norm": 0.002121233381330967,
|
8632 |
+
"learning_rate": 3.347486525383059e-06,
|
8633 |
+
"loss": 0.0,
|
8634 |
+
"step": 1231
|
8635 |
+
},
|
8636 |
+
{
|
8637 |
+
"epoch": 8.847396768402154,
|
8638 |
+
"grad_norm": 0.01041069533675909,
|
8639 |
+
"learning_rate": 3.3067473653432823e-06,
|
8640 |
+
"loss": 0.0001,
|
8641 |
+
"step": 1232
|
8642 |
+
},
|
8643 |
+
{
|
8644 |
+
"epoch": 8.854578096947936,
|
8645 |
+
"grad_norm": 0.001205962966196239,
|
8646 |
+
"learning_rate": 3.266249145536615e-06,
|
8647 |
+
"loss": 0.0,
|
8648 |
+
"step": 1233
|
8649 |
+
},
|
8650 |
+
{
|
8651 |
+
"epoch": 8.861759425493716,
|
8652 |
+
"grad_norm": 0.0004980087978765368,
|
8653 |
+
"learning_rate": 3.2259920749365236e-06,
|
8654 |
+
"loss": 0.0,
|
8655 |
+
"step": 1234
|
8656 |
+
},
|
8657 |
+
{
|
8658 |
+
"epoch": 8.868940754039498,
|
8659 |
+
"grad_norm": 0.001728088827803731,
|
8660 |
+
"learning_rate": 3.1859763612721248e-06,
|
8661 |
+
"loss": 0.0,
|
8662 |
+
"step": 1235
|
8663 |
+
},
|
8664 |
+
{
|
8665 |
+
"epoch": 8.876122082585278,
|
8666 |
+
"grad_norm": 0.002661314560100436,
|
8667 |
+
"learning_rate": 3.146202211027116e-06,
|
8668 |
+
"loss": 0.0,
|
8669 |
+
"step": 1236
|
8670 |
+
},
|
8671 |
+
{
|
8672 |
+
"epoch": 8.88330341113106,
|
8673 |
+
"grad_norm": 0.003290001070126891,
|
8674 |
+
"learning_rate": 3.1066698294386965e-06,
|
8675 |
+
"loss": 0.0001,
|
8676 |
+
"step": 1237
|
8677 |
+
},
|
8678 |
+
{
|
8679 |
+
"epoch": 8.89048473967684,
|
8680 |
+
"grad_norm": 0.0020858109928667545,
|
8681 |
+
"learning_rate": 3.0673794204965467e-06,
|
8682 |
+
"loss": 0.0,
|
8683 |
+
"step": 1238
|
8684 |
+
},
|
8685 |
+
{
|
8686 |
+
"epoch": 8.897666068222621,
|
8687 |
+
"grad_norm": 0.007211592979729176,
|
8688 |
+
"learning_rate": 3.0283311869417407e-06,
|
8689 |
+
"loss": 0.0001,
|
8690 |
+
"step": 1239
|
8691 |
+
},
|
8692 |
+
{
|
8693 |
+
"epoch": 8.904847396768401,
|
8694 |
+
"grad_norm": 0.006036645267158747,
|
8695 |
+
"learning_rate": 2.989525330265719e-06,
|
8696 |
+
"loss": 0.0001,
|
8697 |
+
"step": 1240
|
8698 |
+
},
|
8699 |
+
{
|
8700 |
+
"epoch": 8.912028725314183,
|
8701 |
+
"grad_norm": 0.0010690572671592236,
|
8702 |
+
"learning_rate": 2.950962050709233e-06,
|
8703 |
+
"loss": 0.0,
|
8704 |
+
"step": 1241
|
8705 |
+
},
|
8706 |
+
{
|
8707 |
+
"epoch": 8.919210053859963,
|
8708 |
+
"grad_norm": 0.00833749957382679,
|
8709 |
+
"learning_rate": 2.9126415472613134e-06,
|
8710 |
+
"loss": 0.0001,
|
8711 |
+
"step": 1242
|
8712 |
+
},
|
8713 |
+
{
|
8714 |
+
"epoch": 8.926391382405745,
|
8715 |
+
"grad_norm": 0.0006990973488427699,
|
8716 |
+
"learning_rate": 2.8745640176582765e-06,
|
8717 |
+
"loss": 0.0,
|
8718 |
+
"step": 1243
|
8719 |
+
},
|
8720 |
+
{
|
8721 |
+
"epoch": 8.933572710951527,
|
8722 |
+
"grad_norm": 0.0011197554413229227,
|
8723 |
+
"learning_rate": 2.8367296583826587e-06,
|
8724 |
+
"loss": 0.0,
|
8725 |
+
"step": 1244
|
8726 |
+
},
|
8727 |
+
{
|
8728 |
+
"epoch": 8.940754039497307,
|
8729 |
+
"grad_norm": 0.0009179064654745162,
|
8730 |
+
"learning_rate": 2.799138664662221e-06,
|
8731 |
+
"loss": 0.0,
|
8732 |
+
"step": 1245
|
8733 |
+
},
|
8734 |
+
{
|
8735 |
+
"epoch": 8.947935368043089,
|
8736 |
+
"grad_norm": 0.002500398550182581,
|
8737 |
+
"learning_rate": 2.7617912304689354e-06,
|
8738 |
+
"loss": 0.0,
|
8739 |
+
"step": 1246
|
8740 |
+
},
|
8741 |
+
{
|
8742 |
+
"epoch": 8.955116696588869,
|
8743 |
+
"grad_norm": 0.002693925518542528,
|
8744 |
+
"learning_rate": 2.724687548518029e-06,
|
8745 |
+
"loss": 0.0,
|
8746 |
+
"step": 1247
|
8747 |
+
},
|
8748 |
+
{
|
8749 |
+
"epoch": 8.96229802513465,
|
8750 |
+
"grad_norm": 0.005518902093172073,
|
8751 |
+
"learning_rate": 2.6878278102668965e-06,
|
8752 |
+
"loss": 0.0001,
|
8753 |
+
"step": 1248
|
8754 |
+
},
|
8755 |
+
{
|
8756 |
+
"epoch": 8.96947935368043,
|
8757 |
+
"grad_norm": 0.0013562332605943084,
|
8758 |
+
"learning_rate": 2.651212205914211e-06,
|
8759 |
+
"loss": 0.0,
|
8760 |
+
"step": 1249
|
8761 |
+
},
|
8762 |
+
{
|
8763 |
+
"epoch": 8.976660682226212,
|
8764 |
+
"grad_norm": 0.0032140628900378942,
|
8765 |
+
"learning_rate": 2.614840924398876e-06,
|
8766 |
+
"loss": 0.0,
|
8767 |
+
"step": 1250
|
8768 |
+
},
|
8769 |
+
{
|
8770 |
+
"epoch": 8.983842010771992,
|
8771 |
+
"grad_norm": 0.000525614945217967,
|
8772 |
+
"learning_rate": 2.5787141533990812e-06,
|
8773 |
+
"loss": 0.0,
|
8774 |
+
"step": 1251
|
8775 |
+
},
|
8776 |
+
{
|
8777 |
+
"epoch": 8.991023339317774,
|
8778 |
+
"grad_norm": 0.0005168743664398789,
|
8779 |
+
"learning_rate": 2.5428320793313144e-06,
|
8780 |
+
"loss": 0.0,
|
8781 |
+
"step": 1252
|
8782 |
+
},
|
8783 |
+
{
|
8784 |
+
"epoch": 8.998204667863554,
|
8785 |
+
"grad_norm": 0.0008490683394484222,
|
8786 |
+
"learning_rate": 2.5071948873494457e-06,
|
8787 |
+
"loss": 0.0,
|
8788 |
+
"step": 1253
|
8789 |
+
},
|
8790 |
+
{
|
8791 |
+
"epoch": 9.005385996409336,
|
8792 |
+
"grad_norm": 0.0053194426000118256,
|
8793 |
+
"learning_rate": 2.4718027613436946e-06,
|
8794 |
+
"loss": 0.0001,
|
8795 |
+
"step": 1254
|
8796 |
+
},
|
8797 |
+
{
|
8798 |
+
"epoch": 9.012567324955116,
|
8799 |
+
"grad_norm": 0.004942440427839756,
|
8800 |
+
"learning_rate": 2.436655883939737e-06,
|
8801 |
+
"loss": 0.0001,
|
8802 |
+
"step": 1255
|
8803 |
+
},
|
8804 |
+
{
|
8805 |
+
"epoch": 9.019748653500898,
|
8806 |
+
"grad_norm": 0.0027895679231733084,
|
8807 |
+
"learning_rate": 2.4017544364977775e-06,
|
8808 |
+
"loss": 0.0,
|
8809 |
+
"step": 1256
|
8810 |
+
},
|
8811 |
+
{
|
8812 |
+
"epoch": 9.02692998204668,
|
8813 |
+
"grad_norm": 0.0030093174427747726,
|
8814 |
+
"learning_rate": 2.3670985991115393e-06,
|
8815 |
+
"loss": 0.0001,
|
8816 |
+
"step": 1257
|
8817 |
+
},
|
8818 |
+
{
|
8819 |
+
"epoch": 9.03411131059246,
|
8820 |
+
"grad_norm": 0.0042898040264844894,
|
8821 |
+
"learning_rate": 2.3326885506074314e-06,
|
8822 |
+
"loss": 0.0001,
|
8823 |
+
"step": 1258
|
8824 |
+
},
|
8825 |
+
{
|
8826 |
+
"epoch": 9.041292639138241,
|
8827 |
+
"grad_norm": 0.004731064196676016,
|
8828 |
+
"learning_rate": 2.2985244685435283e-06,
|
8829 |
+
"loss": 0.0,
|
8830 |
+
"step": 1259
|
8831 |
+
},
|
8832 |
+
{
|
8833 |
+
"epoch": 9.048473967684021,
|
8834 |
+
"grad_norm": 0.001769304508343339,
|
8835 |
+
"learning_rate": 2.2646065292087404e-06,
|
8836 |
+
"loss": 0.0,
|
8837 |
+
"step": 1260
|
8838 |
+
},
|
8839 |
+
{
|
8840 |
+
"epoch": 9.055655296229803,
|
8841 |
+
"grad_norm": 0.003293538000434637,
|
8842 |
+
"learning_rate": 2.2309349076218456e-06,
|
8843 |
+
"loss": 0.0001,
|
8844 |
+
"step": 1261
|
8845 |
+
},
|
8846 |
+
{
|
8847 |
+
"epoch": 9.062836624775583,
|
8848 |
+
"grad_norm": 0.0017885896377265453,
|
8849 |
+
"learning_rate": 2.1975097775306097e-06,
|
8850 |
+
"loss": 0.0,
|
8851 |
+
"step": 1262
|
8852 |
+
},
|
8853 |
+
{
|
8854 |
+
"epoch": 9.070017953321365,
|
8855 |
+
"grad_norm": 0.02006862498819828,
|
8856 |
+
"learning_rate": 2.164331311410889e-06,
|
8857 |
+
"loss": 0.0001,
|
8858 |
+
"step": 1263
|
8859 |
+
},
|
8860 |
+
{
|
8861 |
+
"epoch": 9.077199281867145,
|
8862 |
+
"grad_norm": 0.0009690352599136531,
|
8863 |
+
"learning_rate": 2.131399680465729e-06,
|
8864 |
+
"loss": 0.0,
|
8865 |
+
"step": 1264
|
8866 |
+
},
|
8867 |
+
{
|
8868 |
+
"epoch": 9.084380610412927,
|
8869 |
+
"grad_norm": 0.0009640211355872452,
|
8870 |
+
"learning_rate": 2.098715054624506e-06,
|
8871 |
+
"loss": 0.0,
|
8872 |
+
"step": 1265
|
8873 |
+
},
|
8874 |
+
{
|
8875 |
+
"epoch": 9.091561938958707,
|
8876 |
+
"grad_norm": 0.0009525945642963052,
|
8877 |
+
"learning_rate": 2.0662776025420103e-06,
|
8878 |
+
"loss": 0.0,
|
8879 |
+
"step": 1266
|
8880 |
+
},
|
8881 |
+
{
|
8882 |
+
"epoch": 9.098743267504489,
|
8883 |
+
"grad_norm": 0.005768498405814171,
|
8884 |
+
"learning_rate": 2.0340874915976306e-06,
|
8885 |
+
"loss": 0.0001,
|
8886 |
+
"step": 1267
|
8887 |
+
},
|
8888 |
+
{
|
8889 |
+
"epoch": 9.105924596050269,
|
8890 |
+
"grad_norm": 0.003453444456681609,
|
8891 |
+
"learning_rate": 2.002144887894425e-06,
|
8892 |
+
"loss": 0.0001,
|
8893 |
+
"step": 1268
|
8894 |
+
},
|
8895 |
+
{
|
8896 |
+
"epoch": 9.11310592459605,
|
8897 |
+
"grad_norm": 0.0007042995421215892,
|
8898 |
+
"learning_rate": 1.97044995625833e-06,
|
8899 |
+
"loss": 0.0,
|
8900 |
+
"step": 1269
|
8901 |
+
},
|
8902 |
+
{
|
8903 |
+
"epoch": 9.12028725314183,
|
8904 |
+
"grad_norm": 0.0015764246927574277,
|
8905 |
+
"learning_rate": 1.939002860237249e-06,
|
8906 |
+
"loss": 0.0,
|
8907 |
+
"step": 1270
|
8908 |
+
},
|
8909 |
+
{
|
8910 |
+
"epoch": 9.127468581687612,
|
8911 |
+
"grad_norm": 0.002348914509639144,
|
8912 |
+
"learning_rate": 1.907803762100263e-06,
|
8913 |
+
"loss": 0.0,
|
8914 |
+
"step": 1271
|
8915 |
+
},
|
8916 |
+
{
|
8917 |
+
"epoch": 9.134649910233394,
|
8918 |
+
"grad_norm": 0.0004164179554209113,
|
8919 |
+
"learning_rate": 1.87685282283675e-06,
|
8920 |
+
"loss": 0.0,
|
8921 |
+
"step": 1272
|
8922 |
+
},
|
8923 |
+
{
|
8924 |
+
"epoch": 9.141831238779174,
|
8925 |
+
"grad_norm": 0.01101768109947443,
|
8926 |
+
"learning_rate": 1.8461502021555722e-06,
|
8927 |
+
"loss": 0.0001,
|
8928 |
+
"step": 1273
|
8929 |
+
},
|
8930 |
+
{
|
8931 |
+
"epoch": 9.149012567324956,
|
8932 |
+
"grad_norm": 0.001437205821275711,
|
8933 |
+
"learning_rate": 1.8156960584842675e-06,
|
8934 |
+
"loss": 0.0,
|
8935 |
+
"step": 1274
|
8936 |
+
},
|
8937 |
+
{
|
8938 |
+
"epoch": 9.156193895870736,
|
8939 |
+
"grad_norm": 0.0004828003002330661,
|
8940 |
+
"learning_rate": 1.7854905489681994e-06,
|
8941 |
+
"loss": 0.0,
|
8942 |
+
"step": 1275
|
8943 |
+
},
|
8944 |
+
{
|
8945 |
+
"epoch": 9.163375224416518,
|
8946 |
+
"grad_norm": 0.0033631001133471727,
|
8947 |
+
"learning_rate": 1.75553382946978e-06,
|
8948 |
+
"loss": 0.0001,
|
8949 |
+
"step": 1276
|
8950 |
+
},
|
8951 |
+
{
|
8952 |
+
"epoch": 9.170556552962298,
|
8953 |
+
"grad_norm": 0.0023919851519167423,
|
8954 |
+
"learning_rate": 1.7258260545676198e-06,
|
8955 |
+
"loss": 0.0,
|
8956 |
+
"step": 1277
|
8957 |
+
},
|
8958 |
+
{
|
8959 |
+
"epoch": 9.17773788150808,
|
8960 |
+
"grad_norm": 0.00046832981752231717,
|
8961 |
+
"learning_rate": 1.696367377555802e-06,
|
8962 |
+
"loss": 0.0,
|
8963 |
+
"step": 1278
|
8964 |
+
},
|
8965 |
+
{
|
8966 |
+
"epoch": 9.18491921005386,
|
8967 |
+
"grad_norm": 0.010521079413592815,
|
8968 |
+
"learning_rate": 1.6671579504429991e-06,
|
8969 |
+
"loss": 0.0001,
|
8970 |
+
"step": 1279
|
8971 |
+
},
|
8972 |
+
{
|
8973 |
+
"epoch": 9.192100538599641,
|
8974 |
+
"grad_norm": 0.0006425015162676573,
|
8975 |
+
"learning_rate": 1.6381979239517843e-06,
|
8976 |
+
"loss": 0.0,
|
8977 |
+
"step": 1280
|
8978 |
+
},
|
8979 |
+
{
|
8980 |
+
"epoch": 9.199281867145421,
|
8981 |
+
"grad_norm": 0.007516259793192148,
|
8982 |
+
"learning_rate": 1.6094874475177712e-06,
|
8983 |
+
"loss": 0.0001,
|
8984 |
+
"step": 1281
|
8985 |
+
},
|
8986 |
+
{
|
8987 |
+
"epoch": 9.206463195691203,
|
8988 |
+
"grad_norm": 0.005536982323974371,
|
8989 |
+
"learning_rate": 1.5810266692888931e-06,
|
8990 |
+
"loss": 0.0,
|
8991 |
+
"step": 1282
|
8992 |
+
},
|
8993 |
+
{
|
8994 |
+
"epoch": 9.213644524236983,
|
8995 |
+
"grad_norm": 0.008903410285711288,
|
8996 |
+
"learning_rate": 1.552815736124641e-06,
|
8997 |
+
"loss": 0.0001,
|
8998 |
+
"step": 1283
|
8999 |
+
},
|
9000 |
+
{
|
9001 |
+
"epoch": 9.220825852782765,
|
9002 |
+
"grad_norm": 0.003421616740524769,
|
9003 |
+
"learning_rate": 1.5248547935952484e-06,
|
9004 |
+
"loss": 0.0001,
|
9005 |
+
"step": 1284
|
9006 |
+
},
|
9007 |
+
{
|
9008 |
+
"epoch": 9.228007181328545,
|
9009 |
+
"grad_norm": 0.00239805830642581,
|
9010 |
+
"learning_rate": 1.4971439859810198e-06,
|
9011 |
+
"loss": 0.0,
|
9012 |
+
"step": 1285
|
9013 |
+
},
|
9014 |
+
{
|
9015 |
+
"epoch": 9.235188509874327,
|
9016 |
+
"grad_norm": 0.00542791560292244,
|
9017 |
+
"learning_rate": 1.4696834562715312e-06,
|
9018 |
+
"loss": 0.0001,
|
9019 |
+
"step": 1286
|
9020 |
+
},
|
9021 |
+
{
|
9022 |
+
"epoch": 9.242369838420109,
|
9023 |
+
"grad_norm": 0.0025002851616591215,
|
9024 |
+
"learning_rate": 1.4424733461649131e-06,
|
9025 |
+
"loss": 0.0,
|
9026 |
+
"step": 1287
|
9027 |
+
},
|
9028 |
+
{
|
9029 |
+
"epoch": 9.249551166965889,
|
9030 |
+
"grad_norm": 0.0022409127559512854,
|
9031 |
+
"learning_rate": 1.4155137960670972e-06,
|
9032 |
+
"loss": 0.0,
|
9033 |
+
"step": 1288
|
9034 |
+
},
|
9035 |
+
{
|
9036 |
+
"epoch": 9.25673249551167,
|
9037 |
+
"grad_norm": 0.0016906855162233114,
|
9038 |
+
"learning_rate": 1.3888049450911323e-06,
|
9039 |
+
"loss": 0.0,
|
9040 |
+
"step": 1289
|
9041 |
+
},
|
9042 |
+
{
|
9043 |
+
"epoch": 9.26391382405745,
|
9044 |
+
"grad_norm": 0.001018270617350936,
|
9045 |
+
"learning_rate": 1.3623469310564407e-06,
|
9046 |
+
"loss": 0.0,
|
9047 |
+
"step": 1290
|
9048 |
+
},
|
9049 |
+
{
|
9050 |
+
"epoch": 9.271095152603232,
|
9051 |
+
"grad_norm": 0.0011273447889834642,
|
9052 |
+
"learning_rate": 1.3361398904880806e-06,
|
9053 |
+
"loss": 0.0,
|
9054 |
+
"step": 1291
|
9055 |
+
},
|
9056 |
+
{
|
9057 |
+
"epoch": 9.278276481149012,
|
9058 |
+
"grad_norm": 0.003946348559111357,
|
9059 |
+
"learning_rate": 1.3101839586161125e-06,
|
9060 |
+
"loss": 0.0001,
|
9061 |
+
"step": 1292
|
9062 |
+
},
|
9063 |
+
{
|
9064 |
+
"epoch": 9.285457809694794,
|
9065 |
+
"grad_norm": 0.0032709978986531496,
|
9066 |
+
"learning_rate": 1.2844792693748386e-06,
|
9067 |
+
"loss": 0.0,
|
9068 |
+
"step": 1293
|
9069 |
+
},
|
9070 |
+
{
|
9071 |
+
"epoch": 9.292639138240574,
|
9072 |
+
"grad_norm": 0.0048051816411316395,
|
9073 |
+
"learning_rate": 1.259025955402121e-06,
|
9074 |
+
"loss": 0.0,
|
9075 |
+
"step": 1294
|
9076 |
+
},
|
9077 |
+
{
|
9078 |
+
"epoch": 9.299820466786356,
|
9079 |
+
"grad_norm": 0.0077668423764407635,
|
9080 |
+
"learning_rate": 1.233824148038737e-06,
|
9081 |
+
"loss": 0.0001,
|
9082 |
+
"step": 1295
|
9083 |
+
},
|
9084 |
+
{
|
9085 |
+
"epoch": 9.307001795332136,
|
9086 |
+
"grad_norm": 0.0007973044412210584,
|
9087 |
+
"learning_rate": 1.208873977327657e-06,
|
9088 |
+
"loss": 0.0,
|
9089 |
+
"step": 1296
|
9090 |
+
},
|
9091 |
+
{
|
9092 |
+
"epoch": 9.314183123877918,
|
9093 |
+
"grad_norm": 0.006575505714863539,
|
9094 |
+
"learning_rate": 1.1841755720133797e-06,
|
9095 |
+
"loss": 0.0001,
|
9096 |
+
"step": 1297
|
9097 |
+
},
|
9098 |
+
{
|
9099 |
+
"epoch": 9.321364452423698,
|
9100 |
+
"grad_norm": 0.004024865571409464,
|
9101 |
+
"learning_rate": 1.159729059541298e-06,
|
9102 |
+
"loss": 0.0,
|
9103 |
+
"step": 1298
|
9104 |
+
},
|
9105 |
+
{
|
9106 |
+
"epoch": 9.32854578096948,
|
9107 |
+
"grad_norm": 0.0005098568508401513,
|
9108 |
+
"learning_rate": 1.1355345660570172e-06,
|
9109 |
+
"loss": 0.0,
|
9110 |
+
"step": 1299
|
9111 |
+
},
|
9112 |
+
{
|
9113 |
+
"epoch": 9.335727109515261,
|
9114 |
+
"grad_norm": 0.00511940149590373,
|
9115 |
+
"learning_rate": 1.111592216405688e-06,
|
9116 |
+
"loss": 0.0001,
|
9117 |
+
"step": 1300
|
9118 |
+
},
|
9119 |
+
{
|
9120 |
+
"epoch": 9.342908438061041,
|
9121 |
+
"grad_norm": 0.003943296615034342,
|
9122 |
+
"learning_rate": 1.0879021341314077e-06,
|
9123 |
+
"loss": 0.0001,
|
9124 |
+
"step": 1301
|
9125 |
+
},
|
9126 |
+
{
|
9127 |
+
"epoch": 9.350089766606823,
|
9128 |
+
"grad_norm": 0.003307908307760954,
|
9129 |
+
"learning_rate": 1.0644644414765592e-06,
|
9130 |
+
"loss": 0.0001,
|
9131 |
+
"step": 1302
|
9132 |
+
},
|
9133 |
+
{
|
9134 |
+
"epoch": 9.357271095152603,
|
9135 |
+
"grad_norm": 0.0039034478832036257,
|
9136 |
+
"learning_rate": 1.0412792593811504e-06,
|
9137 |
+
"loss": 0.0001,
|
9138 |
+
"step": 1303
|
9139 |
+
},
|
9140 |
+
{
|
9141 |
+
"epoch": 9.364452423698385,
|
9142 |
+
"grad_norm": 0.003994462080299854,
|
9143 |
+
"learning_rate": 1.0183467074822427e-06,
|
9144 |
+
"loss": 0.0001,
|
9145 |
+
"step": 1304
|
9146 |
+
},
|
9147 |
+
{
|
9148 |
+
"epoch": 9.371633752244165,
|
9149 |
+
"grad_norm": 0.011003606021404266,
|
9150 |
+
"learning_rate": 9.956669041133015e-07,
|
9151 |
+
"loss": 0.0001,
|
9152 |
+
"step": 1305
|
9153 |
+
},
|
9154 |
+
{
|
9155 |
+
"epoch": 9.378815080789947,
|
9156 |
+
"grad_norm": 0.006357505451887846,
|
9157 |
+
"learning_rate": 9.732399663035908e-07,
|
9158 |
+
"loss": 0.0001,
|
9159 |
+
"step": 1306
|
9160 |
+
},
|
9161 |
+
{
|
9162 |
+
"epoch": 9.385996409335727,
|
9163 |
+
"grad_norm": 0.002158686053007841,
|
9164 |
+
"learning_rate": 9.510660097775792e-07,
|
9165 |
+
"loss": 0.0,
|
9166 |
+
"step": 1307
|
9167 |
+
},
|
9168 |
+
{
|
9169 |
+
"epoch": 9.393177737881508,
|
9170 |
+
"grad_norm": 0.008415409363806248,
|
9171 |
+
"learning_rate": 9.291451489543246e-07,
|
9172 |
+
"loss": 0.0001,
|
9173 |
+
"step": 1308
|
9174 |
+
},
|
9175 |
+
{
|
9176 |
+
"epoch": 9.400359066427288,
|
9177 |
+
"grad_norm": 0.0016714815283194184,
|
9178 |
+
"learning_rate": 9.074774969469013e-07,
|
9179 |
+
"loss": 0.0,
|
9180 |
+
"step": 1309
|
9181 |
+
},
|
9182 |
+
{
|
9183 |
+
"epoch": 9.40754039497307,
|
9184 |
+
"grad_norm": 0.0017930882750079036,
|
9185 |
+
"learning_rate": 8.860631655618124e-07,
|
9186 |
+
"loss": 0.0,
|
9187 |
+
"step": 1310
|
9188 |
+
},
|
9189 |
+
{
|
9190 |
+
"epoch": 9.41472172351885,
|
9191 |
+
"grad_norm": 0.0028353806119412184,
|
9192 |
+
"learning_rate": 8.649022652984062e-07,
|
9193 |
+
"loss": 0.0,
|
9194 |
+
"step": 1311
|
9195 |
+
},
|
9196 |
+
{
|
9197 |
+
"epoch": 9.421903052064632,
|
9198 |
+
"grad_norm": 0.007203152868896723,
|
9199 |
+
"learning_rate": 8.439949053483054e-07,
|
9200 |
+
"loss": 0.0001,
|
9201 |
+
"step": 1312
|
9202 |
+
},
|
9203 |
+
{
|
9204 |
+
"epoch": 9.429084380610412,
|
9205 |
+
"grad_norm": 0.008097037672996521,
|
9206 |
+
"learning_rate": 8.233411935948565e-07,
|
9207 |
+
"loss": 0.0001,
|
9208 |
+
"step": 1313
|
9209 |
+
},
|
9210 |
+
{
|
9211 |
+
"epoch": 9.436265709156194,
|
9212 |
+
"grad_norm": 0.0008651064126752317,
|
9213 |
+
"learning_rate": 8.029412366125643e-07,
|
9214 |
+
"loss": 0.0,
|
9215 |
+
"step": 1314
|
9216 |
+
},
|
9217 |
+
{
|
9218 |
+
"epoch": 9.443447037701976,
|
9219 |
+
"grad_norm": 0.002910839393734932,
|
9220 |
+
"learning_rate": 7.827951396665312e-07,
|
9221 |
+
"loss": 0.0,
|
9222 |
+
"step": 1315
|
9223 |
+
},
|
9224 |
+
{
|
9225 |
+
"epoch": 9.450628366247756,
|
9226 |
+
"grad_norm": 0.002898834180086851,
|
9227 |
+
"learning_rate": 7.629030067119347e-07,
|
9228 |
+
"loss": 0.0,
|
9229 |
+
"step": 1316
|
9230 |
+
},
|
9231 |
+
{
|
9232 |
+
"epoch": 9.457809694793538,
|
9233 |
+
"grad_norm": 0.0035262927412986755,
|
9234 |
+
"learning_rate": 7.432649403934899e-07,
|
9235 |
+
"loss": 0.0001,
|
9236 |
+
"step": 1317
|
9237 |
+
},
|
9238 |
+
{
|
9239 |
+
"epoch": 9.464991023339318,
|
9240 |
+
"grad_norm": 0.0023212425876408815,
|
9241 |
+
"learning_rate": 7.238810420448883e-07,
|
9242 |
+
"loss": 0.0,
|
9243 |
+
"step": 1318
|
9244 |
+
},
|
9245 |
+
{
|
9246 |
+
"epoch": 9.4721723518851,
|
9247 |
+
"grad_norm": 0.006389750633388758,
|
9248 |
+
"learning_rate": 7.047514116883203e-07,
|
9249 |
+
"loss": 0.0001,
|
9250 |
+
"step": 1319
|
9251 |
+
},
|
9252 |
+
{
|
9253 |
+
"epoch": 9.47935368043088,
|
9254 |
+
"grad_norm": 0.00800410844385624,
|
9255 |
+
"learning_rate": 6.858761480339259e-07,
|
9256 |
+
"loss": 0.0001,
|
9257 |
+
"step": 1320
|
9258 |
+
},
|
9259 |
+
{
|
9260 |
+
"epoch": 9.486535008976661,
|
9261 |
+
"grad_norm": 0.000574652396608144,
|
9262 |
+
"learning_rate": 6.672553484792954e-07,
|
9263 |
+
"loss": 0.0,
|
9264 |
+
"step": 1321
|
9265 |
+
},
|
9266 |
+
{
|
9267 |
+
"epoch": 9.493716337522441,
|
9268 |
+
"grad_norm": 0.004203511867672205,
|
9269 |
+
"learning_rate": 6.488891091089744e-07,
|
9270 |
+
"loss": 0.0,
|
9271 |
+
"step": 1322
|
9272 |
+
},
|
9273 |
+
{
|
9274 |
+
"epoch": 9.500897666068223,
|
9275 |
+
"grad_norm": 0.0021888138726353645,
|
9276 |
+
"learning_rate": 6.30777524693954e-07,
|
9277 |
+
"loss": 0.0,
|
9278 |
+
"step": 1323
|
9279 |
+
},
|
9280 |
+
{
|
9281 |
+
"epoch": 9.508078994614003,
|
9282 |
+
"grad_norm": 0.0033636968582868576,
|
9283 |
+
"learning_rate": 6.129206886911931e-07,
|
9284 |
+
"loss": 0.0,
|
9285 |
+
"step": 1324
|
9286 |
+
},
|
9287 |
+
{
|
9288 |
+
"epoch": 9.515260323159785,
|
9289 |
+
"grad_norm": 0.0030494180973619223,
|
9290 |
+
"learning_rate": 5.953186932431298e-07,
|
9291 |
+
"loss": 0.0,
|
9292 |
+
"step": 1325
|
9293 |
+
},
|
9294 |
+
{
|
9295 |
+
"epoch": 9.522441651705565,
|
9296 |
+
"grad_norm": 0.004318075720220804,
|
9297 |
+
"learning_rate": 5.779716291772208e-07,
|
9298 |
+
"loss": 0.0001,
|
9299 |
+
"step": 1326
|
9300 |
+
},
|
9301 |
+
{
|
9302 |
+
"epoch": 9.529622980251347,
|
9303 |
+
"grad_norm": 0.003387854201719165,
|
9304 |
+
"learning_rate": 5.608795860054416e-07,
|
9305 |
+
"loss": 0.0,
|
9306 |
+
"step": 1327
|
9307 |
+
},
|
9308 |
+
{
|
9309 |
+
"epoch": 9.536804308797127,
|
9310 |
+
"grad_norm": 0.005705864634364843,
|
9311 |
+
"learning_rate": 5.440426519238429e-07,
|
9312 |
+
"loss": 0.0,
|
9313 |
+
"step": 1328
|
9314 |
+
},
|
9315 |
+
{
|
9316 |
+
"epoch": 9.543985637342908,
|
9317 |
+
"grad_norm": 0.0007057313923723996,
|
9318 |
+
"learning_rate": 5.274609138121223e-07,
|
9319 |
+
"loss": 0.0,
|
9320 |
+
"step": 1329
|
9321 |
+
},
|
9322 |
+
{
|
9323 |
+
"epoch": 9.55116696588869,
|
9324 |
+
"grad_norm": 0.001151918200775981,
|
9325 |
+
"learning_rate": 5.111344572331145e-07,
|
9326 |
+
"loss": 0.0,
|
9327 |
+
"step": 1330
|
9328 |
+
},
|
9329 |
+
{
|
9330 |
+
"epoch": 9.55834829443447,
|
9331 |
+
"grad_norm": 0.0010176684008911252,
|
9332 |
+
"learning_rate": 4.950633664324189e-07,
|
9333 |
+
"loss": 0.0,
|
9334 |
+
"step": 1331
|
9335 |
+
},
|
9336 |
+
{
|
9337 |
+
"epoch": 9.565529622980252,
|
9338 |
+
"grad_norm": 0.003837242489680648,
|
9339 |
+
"learning_rate": 4.792477243379001e-07,
|
9340 |
+
"loss": 0.0001,
|
9341 |
+
"step": 1332
|
9342 |
+
},
|
9343 |
+
{
|
9344 |
+
"epoch": 9.572710951526032,
|
9345 |
+
"grad_norm": 0.0014953743666410446,
|
9346 |
+
"learning_rate": 4.636876125593048e-07,
|
9347 |
+
"loss": 0.0,
|
9348 |
+
"step": 1333
|
9349 |
+
},
|
9350 |
+
{
|
9351 |
+
"epoch": 9.579892280071814,
|
9352 |
+
"grad_norm": 0.0008955992525443435,
|
9353 |
+
"learning_rate": 4.4838311138782897e-07,
|
9354 |
+
"loss": 0.0,
|
9355 |
+
"step": 1334
|
9356 |
+
},
|
9357 |
+
{
|
9358 |
+
"epoch": 9.587073608617594,
|
9359 |
+
"grad_norm": 0.0019382266327738762,
|
9360 |
+
"learning_rate": 4.333342997957013e-07,
|
9361 |
+
"loss": 0.0,
|
9362 |
+
"step": 1335
|
9363 |
+
},
|
9364 |
+
{
|
9365 |
+
"epoch": 9.594254937163376,
|
9366 |
+
"grad_norm": 0.0018870545318350196,
|
9367 |
+
"learning_rate": 4.1854125543576705e-07,
|
9368 |
+
"loss": 0.0,
|
9369 |
+
"step": 1336
|
9370 |
+
},
|
9371 |
+
{
|
9372 |
+
"epoch": 9.601436265709156,
|
9373 |
+
"grad_norm": 0.004259036388248205,
|
9374 |
+
"learning_rate": 4.04004054641105e-07,
|
9375 |
+
"loss": 0.0001,
|
9376 |
+
"step": 1337
|
9377 |
+
},
|
9378 |
+
{
|
9379 |
+
"epoch": 9.608617594254937,
|
9380 |
+
"grad_norm": 0.0013580111553892493,
|
9381 |
+
"learning_rate": 3.897227724246222e-07,
|
9382 |
+
"loss": 0.0,
|
9383 |
+
"step": 1338
|
9384 |
+
},
|
9385 |
+
{
|
9386 |
+
"epoch": 9.615798922800717,
|
9387 |
+
"grad_norm": 0.0010445443913340569,
|
9388 |
+
"learning_rate": 3.7569748247866523e-07,
|
9389 |
+
"loss": 0.0,
|
9390 |
+
"step": 1339
|
9391 |
+
},
|
9392 |
+
{
|
9393 |
+
"epoch": 9.6229802513465,
|
9394 |
+
"grad_norm": 0.006067071110010147,
|
9395 |
+
"learning_rate": 3.6192825717464294e-07,
|
9396 |
+
"loss": 0.0001,
|
9397 |
+
"step": 1340
|
9398 |
+
},
|
9399 |
+
{
|
9400 |
+
"epoch": 9.63016157989228,
|
9401 |
+
"grad_norm": 0.0007093562744557858,
|
9402 |
+
"learning_rate": 3.4841516756266015e-07,
|
9403 |
+
"loss": 0.0,
|
9404 |
+
"step": 1341
|
9405 |
+
},
|
9406 |
+
{
|
9407 |
+
"epoch": 9.637342908438061,
|
9408 |
+
"grad_norm": 0.0034481927286833525,
|
9409 |
+
"learning_rate": 3.351582833711453e-07,
|
9410 |
+
"loss": 0.0,
|
9411 |
+
"step": 1342
|
9412 |
+
},
|
9413 |
+
{
|
9414 |
+
"epoch": 9.644524236983841,
|
9415 |
+
"grad_norm": 0.0062695154920220375,
|
9416 |
+
"learning_rate": 3.2215767300647906e-07,
|
9417 |
+
"loss": 0.0001,
|
9418 |
+
"step": 1343
|
9419 |
+
},
|
9420 |
+
{
|
9421 |
+
"epoch": 9.651705565529623,
|
9422 |
+
"grad_norm": 0.0011026825523003936,
|
9423 |
+
"learning_rate": 3.0941340355266637e-07,
|
9424 |
+
"loss": 0.0,
|
9425 |
+
"step": 1344
|
9426 |
+
},
|
9427 |
+
{
|
9428 |
+
"epoch": 9.658886894075405,
|
9429 |
+
"grad_norm": 0.002446260768920183,
|
9430 |
+
"learning_rate": 2.969255407709648e-07,
|
9431 |
+
"loss": 0.0,
|
9432 |
+
"step": 1345
|
9433 |
+
},
|
9434 |
+
{
|
9435 |
+
"epoch": 9.666068222621185,
|
9436 |
+
"grad_norm": 0.0011774407466873527,
|
9437 |
+
"learning_rate": 2.8469414909956804e-07,
|
9438 |
+
"loss": 0.0,
|
9439 |
+
"step": 1346
|
9440 |
+
},
|
9441 |
+
{
|
9442 |
+
"epoch": 9.673249551166966,
|
9443 |
+
"grad_norm": 0.008814037777483463,
|
9444 |
+
"learning_rate": 2.727192916532506e-07,
|
9445 |
+
"loss": 0.0001,
|
9446 |
+
"step": 1347
|
9447 |
+
},
|
9448 |
+
{
|
9449 |
+
"epoch": 9.680430879712747,
|
9450 |
+
"grad_norm": 0.0010899716289713979,
|
9451 |
+
"learning_rate": 2.610010302230625e-07,
|
9452 |
+
"loss": 0.0,
|
9453 |
+
"step": 1348
|
9454 |
+
},
|
9455 |
+
{
|
9456 |
+
"epoch": 9.687612208258528,
|
9457 |
+
"grad_norm": 0.0008330538985319436,
|
9458 |
+
"learning_rate": 2.495394252759964e-07,
|
9459 |
+
"loss": 0.0,
|
9460 |
+
"step": 1349
|
9461 |
+
},
|
9462 |
+
{
|
9463 |
+
"epoch": 9.694793536804308,
|
9464 |
+
"grad_norm": 0.001670084660872817,
|
9465 |
+
"learning_rate": 2.383345359546818e-07,
|
9466 |
+
"loss": 0.0,
|
9467 |
+
"step": 1350
|
9468 |
+
},
|
9469 |
+
{
|
9470 |
+
"epoch": 9.70197486535009,
|
9471 |
+
"grad_norm": 0.003384950337931514,
|
9472 |
+
"learning_rate": 2.2738642007708033e-07,
|
9473 |
+
"loss": 0.0,
|
9474 |
+
"step": 1351
|
9475 |
+
},
|
9476 |
+
{
|
9477 |
+
"epoch": 9.70915619389587,
|
9478 |
+
"grad_norm": 0.004845671821385622,
|
9479 |
+
"learning_rate": 2.166951341361856e-07,
|
9480 |
+
"loss": 0.0001,
|
9481 |
+
"step": 1352
|
9482 |
+
},
|
9483 |
+
{
|
9484 |
+
"epoch": 9.716337522441652,
|
9485 |
+
"grad_norm": 0.0023180951829999685,
|
9486 |
+
"learning_rate": 2.0626073329972905e-07,
|
9487 |
+
"loss": 0.0,
|
9488 |
+
"step": 1353
|
9489 |
+
},
|
9490 |
+
{
|
9491 |
+
"epoch": 9.723518850987432,
|
9492 |
+
"grad_norm": 0.0038995614740997553,
|
9493 |
+
"learning_rate": 1.9608327140990256e-07,
|
9494 |
+
"loss": 0.0,
|
9495 |
+
"step": 1354
|
9496 |
+
},
|
9497 |
+
{
|
9498 |
+
"epoch": 9.730700179533214,
|
9499 |
+
"grad_norm": 0.0008373655728064477,
|
9500 |
+
"learning_rate": 1.861628009830696e-07,
|
9501 |
+
"loss": 0.0,
|
9502 |
+
"step": 1355
|
9503 |
+
},
|
9504 |
+
{
|
9505 |
+
"epoch": 9.737881508078996,
|
9506 |
+
"grad_norm": 0.003047862323001027,
|
9507 |
+
"learning_rate": 1.7649937320951548e-07,
|
9508 |
+
"loss": 0.0001,
|
9509 |
+
"step": 1356
|
9510 |
+
},
|
9511 |
+
{
|
9512 |
+
"epoch": 9.745062836624776,
|
9513 |
+
"grad_norm": 0.004954088944941759,
|
9514 |
+
"learning_rate": 1.6709303795314767e-07,
|
9515 |
+
"loss": 0.0,
|
9516 |
+
"step": 1357
|
9517 |
+
},
|
9518 |
+
{
|
9519 |
+
"epoch": 9.752244165170557,
|
9520 |
+
"grad_norm": 0.004162759054452181,
|
9521 |
+
"learning_rate": 1.579438437512737e-07,
|
9522 |
+
"loss": 0.0,
|
9523 |
+
"step": 1358
|
9524 |
+
},
|
9525 |
+
{
|
9526 |
+
"epoch": 9.759425493716337,
|
9527 |
+
"grad_norm": 0.012499436736106873,
|
9528 |
+
"learning_rate": 1.4905183781434573e-07,
|
9529 |
+
"loss": 0.0002,
|
9530 |
+
"step": 1359
|
9531 |
+
},
|
9532 |
+
{
|
9533 |
+
"epoch": 9.76660682226212,
|
9534 |
+
"grad_norm": 0.0014654555125162005,
|
9535 |
+
"learning_rate": 1.4041706602567207e-07,
|
9536 |
+
"loss": 0.0,
|
9537 |
+
"step": 1360
|
9538 |
+
},
|
9539 |
+
{
|
9540 |
+
"epoch": 9.7737881508079,
|
9541 |
+
"grad_norm": 0.004668499808758497,
|
9542 |
+
"learning_rate": 1.32039572941256e-07,
|
9543 |
+
"loss": 0.0001,
|
9544 |
+
"step": 1361
|
9545 |
+
},
|
9546 |
+
{
|
9547 |
+
"epoch": 9.780969479353681,
|
9548 |
+
"grad_norm": 0.0022428377997130156,
|
9549 |
+
"learning_rate": 1.239194017895018e-07,
|
9550 |
+
"loss": 0.0,
|
9551 |
+
"step": 1362
|
9552 |
+
},
|
9553 |
+
{
|
9554 |
+
"epoch": 9.788150807899461,
|
9555 |
+
"grad_norm": 0.003603378077968955,
|
9556 |
+
"learning_rate": 1.1605659447102568e-07,
|
9557 |
+
"loss": 0.0001,
|
9558 |
+
"step": 1363
|
9559 |
+
},
|
9560 |
+
{
|
9561 |
+
"epoch": 9.795332136445243,
|
9562 |
+
"grad_norm": 0.002531717298552394,
|
9563 |
+
"learning_rate": 1.0845119155842853e-07,
|
9564 |
+
"loss": 0.0,
|
9565 |
+
"step": 1364
|
9566 |
+
},
|
9567 |
+
{
|
9568 |
+
"epoch": 9.802513464991023,
|
9569 |
+
"grad_norm": 0.0031889344099909067,
|
9570 |
+
"learning_rate": 1.0110323229608476e-07,
|
9571 |
+
"loss": 0.0001,
|
9572 |
+
"step": 1365
|
9573 |
+
},
|
9574 |
+
{
|
9575 |
+
"epoch": 9.809694793536805,
|
9576 |
+
"grad_norm": 0.0018446670146659017,
|
9577 |
+
"learning_rate": 9.401275459994252e-08,
|
9578 |
+
"loss": 0.0,
|
9579 |
+
"step": 1366
|
9580 |
+
},
|
9581 |
+
{
|
9582 |
+
"epoch": 9.816876122082585,
|
9583 |
+
"grad_norm": 0.0019796814303845167,
|
9584 |
+
"learning_rate": 8.717979505732943e-08,
|
9585 |
+
"loss": 0.0,
|
9586 |
+
"step": 1367
|
9587 |
+
},
|
9588 |
+
{
|
9589 |
+
"epoch": 9.824057450628366,
|
9590 |
+
"grad_norm": 0.0005245081265456975,
|
9591 |
+
"learning_rate": 8.06043889267638e-08,
|
9592 |
+
"loss": 0.0,
|
9593 |
+
"step": 1368
|
9594 |
+
},
|
9595 |
+
{
|
9596 |
+
"epoch": 9.831238779174146,
|
9597 |
+
"grad_norm": 0.0012920144945383072,
|
9598 |
+
"learning_rate": 7.428657013777152e-08,
|
9599 |
+
"loss": 0.0,
|
9600 |
+
"step": 1369
|
9601 |
+
},
|
9602 |
+
{
|
9603 |
+
"epoch": 9.838420107719928,
|
9604 |
+
"grad_norm": 0.001015665358863771,
|
9605 |
+
"learning_rate": 6.82263712907083e-08,
|
9606 |
+
"loss": 0.0,
|
9607 |
+
"step": 1370
|
9608 |
+
},
|
9609 |
+
{
|
9610 |
+
"epoch": 9.84560143626571,
|
9611 |
+
"grad_norm": 0.0012212838046252728,
|
9612 |
+
"learning_rate": 6.242382365659883e-08,
|
9613 |
+
"loss": 0.0,
|
9614 |
+
"step": 1371
|
9615 |
+
},
|
9616 |
+
{
|
9617 |
+
"epoch": 9.85278276481149,
|
9618 |
+
"grad_norm": 0.0013459456386044621,
|
9619 |
+
"learning_rate": 5.6878957176964564e-08,
|
9620 |
+
"loss": 0.0,
|
9621 |
+
"step": 1372
|
9622 |
+
},
|
9623 |
+
{
|
9624 |
+
"epoch": 9.859964093357272,
|
9625 |
+
"grad_norm": 0.009229972958564758,
|
9626 |
+
"learning_rate": 5.159180046368506e-08,
|
9627 |
+
"loss": 0.0001,
|
9628 |
+
"step": 1373
|
9629 |
+
},
|
9630 |
+
{
|
9631 |
+
"epoch": 9.867145421903052,
|
9632 |
+
"grad_norm": 0.0007870359695516527,
|
9633 |
+
"learning_rate": 4.65623807988258e-08,
|
9634 |
+
"loss": 0.0,
|
9635 |
+
"step": 1374
|
9636 |
+
},
|
9637 |
+
{
|
9638 |
+
"epoch": 9.874326750448834,
|
9639 |
+
"grad_norm": 0.0021520848385989666,
|
9640 |
+
"learning_rate": 4.1790724134521675e-08,
|
9641 |
+
"loss": 0.0,
|
9642 |
+
"step": 1375
|
9643 |
+
},
|
9644 |
+
{
|
9645 |
+
"epoch": 9.881508078994614,
|
9646 |
+
"grad_norm": 0.0025529288686811924,
|
9647 |
+
"learning_rate": 3.727685509284373e-08,
|
9648 |
+
"loss": 0.0,
|
9649 |
+
"step": 1376
|
9650 |
+
},
|
9651 |
+
{
|
9652 |
+
"epoch": 9.888689407540395,
|
9653 |
+
"grad_norm": 0.004391699563711882,
|
9654 |
+
"learning_rate": 3.30207969656382e-08,
|
9655 |
+
"loss": 0.0001,
|
9656 |
+
"step": 1377
|
9657 |
+
},
|
9658 |
+
{
|
9659 |
+
"epoch": 9.895870736086176,
|
9660 |
+
"grad_norm": 0.01228626910597086,
|
9661 |
+
"learning_rate": 2.9022571714448776e-08,
|
9662 |
+
"loss": 0.0001,
|
9663 |
+
"step": 1378
|
9664 |
+
},
|
9665 |
+
{
|
9666 |
+
"epoch": 9.903052064631957,
|
9667 |
+
"grad_norm": 0.0017042134422808886,
|
9668 |
+
"learning_rate": 2.5282199970372288e-08,
|
9669 |
+
"loss": 0.0,
|
9670 |
+
"step": 1379
|
9671 |
+
},
|
9672 |
+
{
|
9673 |
+
"epoch": 9.910233393177737,
|
9674 |
+
"grad_norm": 0.002146362094208598,
|
9675 |
+
"learning_rate": 2.1799701033986543e-08,
|
9676 |
+
"loss": 0.0,
|
9677 |
+
"step": 1380
|
9678 |
+
},
|
9679 |
+
{
|
9680 |
+
"epoch": 9.91741472172352,
|
9681 |
+
"grad_norm": 0.0068658581003546715,
|
9682 |
+
"learning_rate": 1.8575092875194876e-08,
|
9683 |
+
"loss": 0.0001,
|
9684 |
+
"step": 1381
|
9685 |
+
},
|
9686 |
+
{
|
9687 |
+
"epoch": 9.9245960502693,
|
9688 |
+
"grad_norm": 0.0008882509428076446,
|
9689 |
+
"learning_rate": 1.5608392133198425e-08,
|
9690 |
+
"loss": 0.0,
|
9691 |
+
"step": 1382
|
9692 |
+
},
|
9693 |
+
{
|
9694 |
+
"epoch": 9.931777378815081,
|
9695 |
+
"grad_norm": 0.0020740588661283255,
|
9696 |
+
"learning_rate": 1.2899614116362867e-08,
|
9697 |
+
"loss": 0.0,
|
9698 |
+
"step": 1383
|
9699 |
+
},
|
9700 |
+
{
|
9701 |
+
"epoch": 9.938958707360861,
|
9702 |
+
"grad_norm": 0.002014560392126441,
|
9703 |
+
"learning_rate": 1.0448772802162943e-08,
|
9704 |
+
"loss": 0.0,
|
9705 |
+
"step": 1384
|
9706 |
+
},
|
9707 |
+
{
|
9708 |
+
"epoch": 9.946140035906643,
|
9709 |
+
"grad_norm": 0.004955880343914032,
|
9710 |
+
"learning_rate": 8.25588083709361e-09,
|
9711 |
+
"loss": 0.0,
|
9712 |
+
"step": 1385
|
9713 |
+
},
|
9714 |
+
{
|
9715 |
+
"epoch": 9.953321364452425,
|
9716 |
+
"grad_norm": 0.0034909662790596485,
|
9717 |
+
"learning_rate": 6.320949536625653e-09,
|
9718 |
+
"loss": 0.0001,
|
9719 |
+
"step": 1386
|
9720 |
+
},
|
9721 |
+
{
|
9722 |
+
"epoch": 9.960502692998205,
|
9723 |
+
"grad_norm": 0.0021931135561317205,
|
9724 |
+
"learning_rate": 4.643988885127959e-09,
|
9725 |
+
"loss": 0.0,
|
9726 |
+
"step": 1387
|
9727 |
+
},
|
9728 |
+
{
|
9729 |
+
"epoch": 9.967684021543986,
|
9730 |
+
"grad_norm": 0.00938593689352274,
|
9731 |
+
"learning_rate": 3.2250075358286614e-09,
|
9732 |
+
"loss": 0.0001,
|
9733 |
+
"step": 1388
|
9734 |
+
},
|
9735 |
+
{
|
9736 |
+
"epoch": 9.974865350089766,
|
9737 |
+
"grad_norm": 0.000684427039232105,
|
9738 |
+
"learning_rate": 2.0640128107596302e-09,
|
9739 |
+
"loss": 0.0,
|
9740 |
+
"step": 1389
|
9741 |
+
},
|
9742 |
+
{
|
9743 |
+
"epoch": 9.982046678635548,
|
9744 |
+
"grad_norm": 0.0008046108996495605,
|
9745 |
+
"learning_rate": 1.1610107007398174e-09,
|
9746 |
+
"loss": 0.0,
|
9747 |
+
"step": 1390
|
9748 |
+
},
|
9749 |
+
{
|
9750 |
+
"epoch": 9.989228007181328,
|
9751 |
+
"grad_norm": 0.003727276111021638,
|
9752 |
+
"learning_rate": 5.160058653086442e-10,
|
9753 |
+
"loss": 0.0001,
|
9754 |
+
"step": 1391
|
9755 |
+
},
|
9756 |
+
{
|
9757 |
+
"epoch": 9.99640933572711,
|
9758 |
+
"grad_norm": 0.006966722197830677,
|
9759 |
+
"learning_rate": 1.2900163274265354e-10,
|
9760 |
+
"loss": 0.0001,
|
9761 |
+
"step": 1392
|
9762 |
+
},
|
9763 |
+
{
|
9764 |
+
"epoch": 10.00359066427289,
|
9765 |
+
"grad_norm": 0.00568317761644721,
|
9766 |
+
"learning_rate": 0.0,
|
9767 |
+
"loss": 0.0001,
|
9768 |
+
"step": 1393
|
9769 |
}
|
9770 |
],
|
9771 |
"logging_steps": 1,
|
|
|
9780 |
"should_evaluate": false,
|
9781 |
"should_log": false,
|
9782 |
"should_save": true,
|
9783 |
+
"should_training_stop": true
|
9784 |
},
|
9785 |
"attributes": {}
|
9786 |
}
|
9787 |
},
|
9788 |
+
"total_flos": 5.021446705250304e+16,
|
9789 |
"train_batch_size": 4,
|
9790 |
"trial_name": null,
|
9791 |
"trial_params": null
|