Hugging Face's logo

chbh7051
/
vit-driver-drowsiness-detection

Image Classification
Transformers
PyTorch
TensorBoard
vit
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
2
chbh7051 commited on
Commit
f201c82
•
1 Parent(s): bd9d84f

🍻 cheers

Browse files
Files changed (5) hide show
  1. all_results.json +11 -5
  2. eval_results.json +6 -5
  3. runs/Apr30_13-04-08_e481a8e10e39/events.out.tfevents.1682866807.e481a8e10e39.23.6 +3 -0
  4. train_results.json +8 -0
  5. trainer_state.json +4132 -0
all_results.json CHANGED
@@ -1,7 +1,13 @@
1
  {
2
- "eval_accuracy": 0.47794062382562946,
3
- "eval_loss": 0.7614386677742004,
4
- "eval_runtime": 230.1051,
5
- "eval_samples_per_second": 57.821,
6
- "eval_steps_per_second": 7.231
 
 
 
 
 
 
7
  }
 
1
  {
2
+ "epoch": 2.0,
3
+ "eval_accuracy": 0.995640736565201,
4
+ "eval_loss": 0.01245423499494791,
5
+ "eval_runtime": 226.1285,
6
+ "eval_samples_per_second": 58.838,
7
+ "eval_steps_per_second": 7.359,
8
+ "total_flos": 8.247627438660256e+18,
9
+ "train_loss": 0.07274016925322493,
10
+ "train_runtime": 6476.0215,
11
+ "train_samples_per_second": 16.435,
12
+ "train_steps_per_second": 1.027
13
  }
eval_results.json CHANGED
@@ -1,7 +1,8 @@
1
  {
2
- "eval_accuracy": 0.47794062382562946,
3
- "eval_loss": 0.7614386677742004,
4
- "eval_runtime": 230.1051,
5
- "eval_samples_per_second": 57.821,
6
- "eval_steps_per_second": 7.231
 
7
  }
 
1
  {
2
+ "epoch": 2.0,
3
+ "eval_accuracy": 0.995640736565201,
4
+ "eval_loss": 0.01245423499494791,
5
+ "eval_runtime": 226.1285,
6
+ "eval_samples_per_second": 58.838,
7
+ "eval_steps_per_second": 7.359
8
  }
runs/Apr30_13-04-08_e481a8e10e39/events.out.tfevents.1682866807.e481a8e10e39.23.6 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8266341a70b8454e03c8e9cfa8a9d692d479edb7199f84a3ce13eaf82e5b957a
3
+ size 363
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 2.0,
3
+ "total_flos": 8.247627438660256e+18,
4
+ "train_loss": 0.07274016925322493,
5
+ "train_runtime": 6476.0215,
6
+ "train_samples_per_second": 16.435,
7
+ "train_steps_per_second": 1.027
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,4132 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.01245423499494791,
3
+ "best_model_checkpoint": "./vit-driver-drowsiness-detection/checkpoint-6500",
4
+ "epoch": 2.0,
5
+ "global_step": 6652,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 0.00019972940469031872,
13
+ "loss": 0.8546,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.01,
18
+ "learning_rate": 0.00019942874323511728,
19
+ "loss": 0.7189,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.01,
24
+ "learning_rate": 0.00019912808177991581,
25
+ "loss": 0.5254,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 0.01,
30
+ "learning_rate": 0.00019882742032471438,
31
+ "loss": 0.4943,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.02,
36
+ "learning_rate": 0.00019852675886951294,
37
+ "loss": 0.385,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 0.02,
42
+ "learning_rate": 0.00019822609741431148,
43
+ "loss": 0.537,
44
+ "step": 60
45
+ },
46
+ {
47
+ "epoch": 0.02,
48
+ "learning_rate": 0.00019792543595911007,
49
+ "loss": 0.4754,
50
+ "step": 70
51
+ },
52
+ {
53
+ "epoch": 0.02,
54
+ "learning_rate": 0.00019762477450390863,
55
+ "loss": 0.4489,
56
+ "step": 80
57
+ },
58
+ {
59
+ "epoch": 0.03,
60
+ "learning_rate": 0.00019732411304870717,
61
+ "loss": 0.3409,
62
+ "step": 90
63
+ },
64
+ {
65
+ "epoch": 0.03,
66
+ "learning_rate": 0.00019702345159350573,
67
+ "loss": 0.5113,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 0.03,
72
+ "learning_rate": 0.0001967227901383043,
73
+ "loss": 0.3692,
74
+ "step": 110
75
+ },
76
+ {
77
+ "epoch": 0.04,
78
+ "learning_rate": 0.00019642212868310283,
79
+ "loss": 0.2405,
80
+ "step": 120
81
+ },
82
+ {
83
+ "epoch": 0.04,
84
+ "learning_rate": 0.0001961214672279014,
85
+ "loss": 0.256,
86
+ "step": 130
87
+ },
88
+ {
89
+ "epoch": 0.04,
90
+ "learning_rate": 0.00019582080577269995,
91
+ "loss": 0.1718,
92
+ "step": 140
93
+ },
94
+ {
95
+ "epoch": 0.05,
96
+ "learning_rate": 0.0001955201443174985,
97
+ "loss": 0.5432,
98
+ "step": 150
99
+ },
100
+ {
101
+ "epoch": 0.05,
102
+ "learning_rate": 0.00019521948286229705,
103
+ "loss": 0.2357,
104
+ "step": 160
105
+ },
106
+ {
107
+ "epoch": 0.05,
108
+ "learning_rate": 0.00019491882140709561,
109
+ "loss": 0.2814,
110
+ "step": 170
111
+ },
112
+ {
113
+ "epoch": 0.05,
114
+ "learning_rate": 0.00019461815995189418,
115
+ "loss": 0.3477,
116
+ "step": 180
117
+ },
118
+ {
119
+ "epoch": 0.06,
120
+ "learning_rate": 0.00019431749849669274,
121
+ "loss": 0.3202,
122
+ "step": 190
123
+ },
124
+ {
125
+ "epoch": 0.06,
126
+ "learning_rate": 0.0001940168370414913,
127
+ "loss": 0.3072,
128
+ "step": 200
129
+ },
130
+ {
131
+ "epoch": 0.06,
132
+ "learning_rate": 0.00019371617558628984,
133
+ "loss": 0.1754,
134
+ "step": 210
135
+ },
136
+ {
137
+ "epoch": 0.07,
138
+ "learning_rate": 0.0001934155141310884,
139
+ "loss": 0.1892,
140
+ "step": 220
141
+ },
142
+ {
143
+ "epoch": 0.07,
144
+ "learning_rate": 0.00019311485267588696,
145
+ "loss": 0.3008,
146
+ "step": 230
147
+ },
148
+ {
149
+ "epoch": 0.07,
150
+ "learning_rate": 0.00019281419122068553,
151
+ "loss": 0.2058,
152
+ "step": 240
153
+ },
154
+ {
155
+ "epoch": 0.08,
156
+ "learning_rate": 0.00019251352976548406,
157
+ "loss": 0.1859,
158
+ "step": 250
159
+ },
160
+ {
161
+ "epoch": 0.08,
162
+ "learning_rate": 0.00019221286831028263,
163
+ "loss": 0.3031,
164
+ "step": 260
165
+ },
166
+ {
167
+ "epoch": 0.08,
168
+ "learning_rate": 0.0001919122068550812,
169
+ "loss": 0.2275,
170
+ "step": 270
171
+ },
172
+ {
173
+ "epoch": 0.08,
174
+ "learning_rate": 0.00019161154539987975,
175
+ "loss": 0.2927,
176
+ "step": 280
177
+ },
178
+ {
179
+ "epoch": 0.09,
180
+ "learning_rate": 0.00019131088394467831,
181
+ "loss": 0.1919,
182
+ "step": 290
183
+ },
184
+ {
185
+ "epoch": 0.09,
186
+ "learning_rate": 0.00019101022248947688,
187
+ "loss": 0.2789,
188
+ "step": 300
189
+ },
190
+ {
191
+ "epoch": 0.09,
192
+ "learning_rate": 0.0001907095610342754,
193
+ "loss": 0.2592,
194
+ "step": 310
195
+ },
196
+ {
197
+ "epoch": 0.1,
198
+ "learning_rate": 0.00019040889957907398,
199
+ "loss": 0.1957,
200
+ "step": 320
201
+ },
202
+ {
203
+ "epoch": 0.1,
204
+ "learning_rate": 0.00019010823812387254,
205
+ "loss": 0.3101,
206
+ "step": 330
207
+ },
208
+ {
209
+ "epoch": 0.1,
210
+ "learning_rate": 0.00018980757666867107,
211
+ "loss": 0.1067,
212
+ "step": 340
213
+ },
214
+ {
215
+ "epoch": 0.11,
216
+ "learning_rate": 0.00018950691521346964,
217
+ "loss": 0.2988,
218
+ "step": 350
219
+ },
220
+ {
221
+ "epoch": 0.11,
222
+ "learning_rate": 0.0001892062537582682,
223
+ "loss": 0.1753,
224
+ "step": 360
225
+ },
226
+ {
227
+ "epoch": 0.11,
228
+ "learning_rate": 0.00018890559230306674,
229
+ "loss": 0.1293,
230
+ "step": 370
231
+ },
232
+ {
233
+ "epoch": 0.11,
234
+ "learning_rate": 0.0001886049308478653,
235
+ "loss": 0.2626,
236
+ "step": 380
237
+ },
238
+ {
239
+ "epoch": 0.12,
240
+ "learning_rate": 0.0001883042693926639,
241
+ "loss": 0.2491,
242
+ "step": 390
243
+ },
244
+ {
245
+ "epoch": 0.12,
246
+ "learning_rate": 0.00018800360793746242,
247
+ "loss": 0.2234,
248
+ "step": 400
249
+ },
250
+ {
251
+ "epoch": 0.12,
252
+ "learning_rate": 0.000187702946482261,
253
+ "loss": 0.1489,
254
+ "step": 410
255
+ },
256
+ {
257
+ "epoch": 0.13,
258
+ "learning_rate": 0.00018740228502705955,
259
+ "loss": 0.153,
260
+ "step": 420
261
+ },
262
+ {
263
+ "epoch": 0.13,
264
+ "learning_rate": 0.00018710162357185809,
265
+ "loss": 0.1347,
266
+ "step": 430
267
+ },
268
+ {
269
+ "epoch": 0.13,
270
+ "learning_rate": 0.00018680096211665665,
271
+ "loss": 0.1264,
272
+ "step": 440
273
+ },
274
+ {
275
+ "epoch": 0.14,
276
+ "learning_rate": 0.0001865003006614552,
277
+ "loss": 0.1457,
278
+ "step": 450
279
+ },
280
+ {
281
+ "epoch": 0.14,
282
+ "learning_rate": 0.00018619963920625377,
283
+ "loss": 0.1749,
284
+ "step": 460
285
+ },
286
+ {
287
+ "epoch": 0.14,
288
+ "learning_rate": 0.0001858989777510523,
289
+ "loss": 0.1398,
290
+ "step": 470
291
+ },
292
+ {
293
+ "epoch": 0.14,
294
+ "learning_rate": 0.00018559831629585087,
295
+ "loss": 0.1184,
296
+ "step": 480
297
+ },
298
+ {
299
+ "epoch": 0.15,
300
+ "learning_rate": 0.00018529765484064944,
301
+ "loss": 0.0958,
302
+ "step": 490
303
+ },
304
+ {
305
+ "epoch": 0.15,
306
+ "learning_rate": 0.000184996993385448,
307
+ "loss": 0.2436,
308
+ "step": 500
309
+ },
310
+ {
311
+ "epoch": 0.15,
312
+ "eval_accuracy": 0.9050732807215333,
313
+ "eval_loss": 0.23238418996334076,
314
+ "eval_runtime": 229.7918,
315
+ "eval_samples_per_second": 57.9,
316
+ "eval_steps_per_second": 7.241,
317
+ "step": 500
318
+ },
319
+ {
320
+ "epoch": 0.15,
321
+ "learning_rate": 0.00018469633193024656,
322
+ "loss": 0.1342,
323
+ "step": 510
324
+ },
325
+ {
326
+ "epoch": 0.16,
327
+ "learning_rate": 0.00018439567047504513,
328
+ "loss": 0.1673,
329
+ "step": 520
330
+ },
331
+ {
332
+ "epoch": 0.16,
333
+ "learning_rate": 0.00018409500901984366,
334
+ "loss": 0.2167,
335
+ "step": 530
336
+ },
337
+ {
338
+ "epoch": 0.16,
339
+ "learning_rate": 0.00018379434756464222,
340
+ "loss": 0.1364,
341
+ "step": 540
342
+ },
343
+ {
344
+ "epoch": 0.17,
345
+ "learning_rate": 0.0001834936861094408,
346
+ "loss": 0.3077,
347
+ "step": 550
348
+ },
349
+ {
350
+ "epoch": 0.17,
351
+ "learning_rate": 0.00018319302465423932,
352
+ "loss": 0.1875,
353
+ "step": 560
354
+ },
355
+ {
356
+ "epoch": 0.17,
357
+ "learning_rate": 0.00018289236319903789,
358
+ "loss": 0.1344,
359
+ "step": 570
360
+ },
361
+ {
362
+ "epoch": 0.17,
363
+ "learning_rate": 0.00018259170174383645,
364
+ "loss": 0.1299,
365
+ "step": 580
366
+ },
367
+ {
368
+ "epoch": 0.18,
369
+ "learning_rate": 0.00018229104028863498,
370
+ "loss": 0.0836,
371
+ "step": 590
372
+ },
373
+ {
374
+ "epoch": 0.18,
375
+ "learning_rate": 0.00018199037883343357,
376
+ "loss": 0.2805,
377
+ "step": 600
378
+ },
379
+ {
380
+ "epoch": 0.18,
381
+ "learning_rate": 0.00018168971737823214,
382
+ "loss": 0.1024,
383
+ "step": 610
384
+ },
385
+ {
386
+ "epoch": 0.19,
387
+ "learning_rate": 0.00018138905592303067,
388
+ "loss": 0.1117,
389
+ "step": 620
390
+ },
391
+ {
392
+ "epoch": 0.19,
393
+ "learning_rate": 0.00018108839446782924,
394
+ "loss": 0.1694,
395
+ "step": 630
396
+ },
397
+ {
398
+ "epoch": 0.19,
399
+ "learning_rate": 0.0001807877330126278,
400
+ "loss": 0.2399,
401
+ "step": 640
402
+ },
403
+ {
404
+ "epoch": 0.2,
405
+ "learning_rate": 0.00018048707155742633,
406
+ "loss": 0.1797,
407
+ "step": 650
408
+ },
409
+ {
410
+ "epoch": 0.2,
411
+ "learning_rate": 0.0001801864101022249,
412
+ "loss": 0.0956,
413
+ "step": 660
414
+ },
415
+ {
416
+ "epoch": 0.2,
417
+ "learning_rate": 0.00017988574864702346,
418
+ "loss": 0.2777,
419
+ "step": 670
420
+ },
421
+ {
422
+ "epoch": 0.2,
423
+ "learning_rate": 0.00017958508719182202,
424
+ "loss": 0.1504,
425
+ "step": 680
426
+ },
427
+ {
428
+ "epoch": 0.21,
429
+ "learning_rate": 0.00017928442573662056,
430
+ "loss": 0.0971,
431
+ "step": 690
432
+ },
433
+ {
434
+ "epoch": 0.21,
435
+ "learning_rate": 0.00017898376428141912,
436
+ "loss": 0.145,
437
+ "step": 700
438
+ },
439
+ {
440
+ "epoch": 0.21,
441
+ "learning_rate": 0.00017868310282621768,
442
+ "loss": 0.0974,
443
+ "step": 710
444
+ },
445
+ {
446
+ "epoch": 0.22,
447
+ "learning_rate": 0.00017838244137101625,
448
+ "loss": 0.1218,
449
+ "step": 720
450
+ },
451
+ {
452
+ "epoch": 0.22,
453
+ "learning_rate": 0.0001780817799158148,
454
+ "loss": 0.0609,
455
+ "step": 730
456
+ },
457
+ {
458
+ "epoch": 0.22,
459
+ "learning_rate": 0.00017778111846061337,
460
+ "loss": 0.1454,
461
+ "step": 740
462
+ },
463
+ {
464
+ "epoch": 0.23,
465
+ "learning_rate": 0.0001774804570054119,
466
+ "loss": 0.1273,
467
+ "step": 750
468
+ },
469
+ {
470
+ "epoch": 0.23,
471
+ "learning_rate": 0.00017717979555021047,
472
+ "loss": 0.2277,
473
+ "step": 760
474
+ },
475
+ {
476
+ "epoch": 0.23,
477
+ "learning_rate": 0.00017687913409500903,
478
+ "loss": 0.1633,
479
+ "step": 770
480
+ },
481
+ {
482
+ "epoch": 0.23,
483
+ "learning_rate": 0.00017657847263980757,
484
+ "loss": 0.1898,
485
+ "step": 780
486
+ },
487
+ {
488
+ "epoch": 0.24,
489
+ "learning_rate": 0.00017627781118460613,
490
+ "loss": 0.0684,
491
+ "step": 790
492
+ },
493
+ {
494
+ "epoch": 0.24,
495
+ "learning_rate": 0.0001759771497294047,
496
+ "loss": 0.0871,
497
+ "step": 800
498
+ },
499
+ {
500
+ "epoch": 0.24,
501
+ "learning_rate": 0.00017567648827420326,
502
+ "loss": 0.211,
503
+ "step": 810
504
+ },
505
+ {
506
+ "epoch": 0.25,
507
+ "learning_rate": 0.00017537582681900182,
508
+ "loss": 0.15,
509
+ "step": 820
510
+ },
511
+ {
512
+ "epoch": 0.25,
513
+ "learning_rate": 0.00017507516536380038,
514
+ "loss": 0.1125,
515
+ "step": 830
516
+ },
517
+ {
518
+ "epoch": 0.25,
519
+ "learning_rate": 0.00017477450390859892,
520
+ "loss": 0.1748,
521
+ "step": 840
522
+ },
523
+ {
524
+ "epoch": 0.26,
525
+ "learning_rate": 0.00017447384245339748,
526
+ "loss": 0.1253,
527
+ "step": 850
528
+ },
529
+ {
530
+ "epoch": 0.26,
531
+ "learning_rate": 0.00017417318099819605,
532
+ "loss": 0.1589,
533
+ "step": 860
534
+ },
535
+ {
536
+ "epoch": 0.26,
537
+ "learning_rate": 0.00017387251954299458,
538
+ "loss": 0.1286,
539
+ "step": 870
540
+ },
541
+ {
542
+ "epoch": 0.26,
543
+ "learning_rate": 0.00017357185808779314,
544
+ "loss": 0.1169,
545
+ "step": 880
546
+ },
547
+ {
548
+ "epoch": 0.27,
549
+ "learning_rate": 0.0001732711966325917,
550
+ "loss": 0.2119,
551
+ "step": 890
552
+ },
553
+ {
554
+ "epoch": 0.27,
555
+ "learning_rate": 0.00017297053517739027,
556
+ "loss": 0.1145,
557
+ "step": 900
558
+ },
559
+ {
560
+ "epoch": 0.27,
561
+ "learning_rate": 0.0001726698737221888,
562
+ "loss": 0.0758,
563
+ "step": 910
564
+ },
565
+ {
566
+ "epoch": 0.28,
567
+ "learning_rate": 0.0001723692122669874,
568
+ "loss": 0.0998,
569
+ "step": 920
570
+ },
571
+ {
572
+ "epoch": 0.28,
573
+ "learning_rate": 0.00017206855081178593,
574
+ "loss": 0.1862,
575
+ "step": 930
576
+ },
577
+ {
578
+ "epoch": 0.28,
579
+ "learning_rate": 0.0001717678893565845,
580
+ "loss": 0.1858,
581
+ "step": 940
582
+ },
583
+ {
584
+ "epoch": 0.29,
585
+ "learning_rate": 0.00017146722790138306,
586
+ "loss": 0.1629,
587
+ "step": 950
588
+ },
589
+ {
590
+ "epoch": 0.29,
591
+ "learning_rate": 0.00017116656644618162,
592
+ "loss": 0.1312,
593
+ "step": 960
594
+ },
595
+ {
596
+ "epoch": 0.29,
597
+ "learning_rate": 0.00017086590499098016,
598
+ "loss": 0.099,
599
+ "step": 970
600
+ },
601
+ {
602
+ "epoch": 0.29,
603
+ "learning_rate": 0.00017056524353577872,
604
+ "loss": 0.2063,
605
+ "step": 980
606
+ },
607
+ {
608
+ "epoch": 0.3,
609
+ "learning_rate": 0.00017026458208057728,
610
+ "loss": 0.0966,
611
+ "step": 990
612
+ },
613
+ {
614
+ "epoch": 0.3,
615
+ "learning_rate": 0.00016996392062537582,
616
+ "loss": 0.0963,
617
+ "step": 1000
618
+ },
619
+ {
620
+ "epoch": 0.3,
621
+ "eval_accuracy": 0.9683577602405111,
622
+ "eval_loss": 0.10860169678926468,
623
+ "eval_runtime": 227.4819,
624
+ "eval_samples_per_second": 58.488,
625
+ "eval_steps_per_second": 7.315,
626
+ "step": 1000
627
+ },
628
+ {
629
+ "epoch": 0.3,
630
+ "learning_rate": 0.00016966325917017438,
631
+ "loss": 0.0842,
632
+ "step": 1010
633
+ },
634
+ {
635
+ "epoch": 0.31,
636
+ "learning_rate": 0.00016936259771497297,
637
+ "loss": 0.0238,
638
+ "step": 1020
639
+ },
640
+ {
641
+ "epoch": 0.31,
642
+ "learning_rate": 0.0001690619362597715,
643
+ "loss": 0.1495,
644
+ "step": 1030
645
+ },
646
+ {
647
+ "epoch": 0.31,
648
+ "learning_rate": 0.00016876127480457007,
649
+ "loss": 0.1465,
650
+ "step": 1040
651
+ },
652
+ {
653
+ "epoch": 0.32,
654
+ "learning_rate": 0.00016846061334936863,
655
+ "loss": 0.1466,
656
+ "step": 1050
657
+ },
658
+ {
659
+ "epoch": 0.32,
660
+ "learning_rate": 0.00016815995189416717,
661
+ "loss": 0.0678,
662
+ "step": 1060
663
+ },
664
+ {
665
+ "epoch": 0.32,
666
+ "learning_rate": 0.00016785929043896573,
667
+ "loss": 0.1404,
668
+ "step": 1070
669
+ },
670
+ {
671
+ "epoch": 0.32,
672
+ "learning_rate": 0.0001675586289837643,
673
+ "loss": 0.1123,
674
+ "step": 1080
675
+ },
676
+ {
677
+ "epoch": 0.33,
678
+ "learning_rate": 0.00016725796752856283,
679
+ "loss": 0.091,
680
+ "step": 1090
681
+ },
682
+ {
683
+ "epoch": 0.33,
684
+ "learning_rate": 0.0001669573060733614,
685
+ "loss": 0.0828,
686
+ "step": 1100
687
+ },
688
+ {
689
+ "epoch": 0.33,
690
+ "learning_rate": 0.00016665664461815996,
691
+ "loss": 0.0467,
692
+ "step": 1110
693
+ },
694
+ {
695
+ "epoch": 0.34,
696
+ "learning_rate": 0.00016635598316295852,
697
+ "loss": 0.1016,
698
+ "step": 1120
699
+ },
700
+ {
701
+ "epoch": 0.34,
702
+ "learning_rate": 0.00016605532170775708,
703
+ "loss": 0.103,
704
+ "step": 1130
705
+ },
706
+ {
707
+ "epoch": 0.34,
708
+ "learning_rate": 0.00016575466025255564,
709
+ "loss": 0.1531,
710
+ "step": 1140
711
+ },
712
+ {
713
+ "epoch": 0.35,
714
+ "learning_rate": 0.00016545399879735418,
715
+ "loss": 0.0608,
716
+ "step": 1150
717
+ },
718
+ {
719
+ "epoch": 0.35,
720
+ "learning_rate": 0.00016515333734215274,
721
+ "loss": 0.0683,
722
+ "step": 1160
723
+ },
724
+ {
725
+ "epoch": 0.35,
726
+ "learning_rate": 0.0001648526758869513,
727
+ "loss": 0.0613,
728
+ "step": 1170
729
+ },
730
+ {
731
+ "epoch": 0.35,
732
+ "learning_rate": 0.00016455201443174987,
733
+ "loss": 0.1331,
734
+ "step": 1180
735
+ },
736
+ {
737
+ "epoch": 0.36,
738
+ "learning_rate": 0.0001642513529765484,
739
+ "loss": 0.0869,
740
+ "step": 1190
741
+ },
742
+ {
743
+ "epoch": 0.36,
744
+ "learning_rate": 0.00016395069152134697,
745
+ "loss": 0.0784,
746
+ "step": 1200
747
+ },
748
+ {
749
+ "epoch": 0.36,
750
+ "learning_rate": 0.00016365003006614553,
751
+ "loss": 0.1372,
752
+ "step": 1210
753
+ },
754
+ {
755
+ "epoch": 0.37,
756
+ "learning_rate": 0.00016334936861094407,
757
+ "loss": 0.0919,
758
+ "step": 1220
759
+ },
760
+ {
761
+ "epoch": 0.37,
762
+ "learning_rate": 0.00016304870715574266,
763
+ "loss": 0.0834,
764
+ "step": 1230
765
+ },
766
+ {
767
+ "epoch": 0.37,
768
+ "learning_rate": 0.00016274804570054122,
769
+ "loss": 0.1441,
770
+ "step": 1240
771
+ },
772
+ {
773
+ "epoch": 0.38,
774
+ "learning_rate": 0.00016244738424533975,
775
+ "loss": 0.1106,
776
+ "step": 1250
777
+ },
778
+ {
779
+ "epoch": 0.38,
780
+ "learning_rate": 0.00016214672279013832,
781
+ "loss": 0.0765,
782
+ "step": 1260
783
+ },
784
+ {
785
+ "epoch": 0.38,
786
+ "learning_rate": 0.00016184606133493688,
787
+ "loss": 0.095,
788
+ "step": 1270
789
+ },
790
+ {
791
+ "epoch": 0.38,
792
+ "learning_rate": 0.00016154539987973542,
793
+ "loss": 0.0555,
794
+ "step": 1280
795
+ },
796
+ {
797
+ "epoch": 0.39,
798
+ "learning_rate": 0.00016124473842453398,
799
+ "loss": 0.1185,
800
+ "step": 1290
801
+ },
802
+ {
803
+ "epoch": 0.39,
804
+ "learning_rate": 0.00016094407696933254,
805
+ "loss": 0.1238,
806
+ "step": 1300
807
+ },
808
+ {
809
+ "epoch": 0.39,
810
+ "learning_rate": 0.00016064341551413108,
811
+ "loss": 0.1677,
812
+ "step": 1310
813
+ },
814
+ {
815
+ "epoch": 0.4,
816
+ "learning_rate": 0.00016034275405892964,
817
+ "loss": 0.0168,
818
+ "step": 1320
819
+ },
820
+ {
821
+ "epoch": 0.4,
822
+ "learning_rate": 0.0001600420926037282,
823
+ "loss": 0.118,
824
+ "step": 1330
825
+ },
826
+ {
827
+ "epoch": 0.4,
828
+ "learning_rate": 0.00015974143114852677,
829
+ "loss": 0.1389,
830
+ "step": 1340
831
+ },
832
+ {
833
+ "epoch": 0.41,
834
+ "learning_rate": 0.00015944076969332533,
835
+ "loss": 0.1779,
836
+ "step": 1350
837
+ },
838
+ {
839
+ "epoch": 0.41,
840
+ "learning_rate": 0.0001591401082381239,
841
+ "loss": 0.0985,
842
+ "step": 1360
843
+ },
844
+ {
845
+ "epoch": 0.41,
846
+ "learning_rate": 0.00015883944678292243,
847
+ "loss": 0.0973,
848
+ "step": 1370
849
+ },
850
+ {
851
+ "epoch": 0.41,
852
+ "learning_rate": 0.000158538785327721,
853
+ "loss": 0.1245,
854
+ "step": 1380
855
+ },
856
+ {
857
+ "epoch": 0.42,
858
+ "learning_rate": 0.00015823812387251955,
859
+ "loss": 0.0897,
860
+ "step": 1390
861
+ },
862
+ {
863
+ "epoch": 0.42,
864
+ "learning_rate": 0.00015793746241731812,
865
+ "loss": 0.0835,
866
+ "step": 1400
867
+ },
868
+ {
869
+ "epoch": 0.42,
870
+ "learning_rate": 0.00015763680096211665,
871
+ "loss": 0.1386,
872
+ "step": 1410
873
+ },
874
+ {
875
+ "epoch": 0.43,
876
+ "learning_rate": 0.00015733613950691522,
877
+ "loss": 0.0679,
878
+ "step": 1420
879
+ },
880
+ {
881
+ "epoch": 0.43,
882
+ "learning_rate": 0.00015703547805171378,
883
+ "loss": 0.068,
884
+ "step": 1430
885
+ },
886
+ {
887
+ "epoch": 0.43,
888
+ "learning_rate": 0.00015673481659651231,
889
+ "loss": 0.0966,
890
+ "step": 1440
891
+ },
892
+ {
893
+ "epoch": 0.44,
894
+ "learning_rate": 0.0001564341551413109,
895
+ "loss": 0.1881,
896
+ "step": 1450
897
+ },
898
+ {
899
+ "epoch": 0.44,
900
+ "learning_rate": 0.00015613349368610947,
901
+ "loss": 0.1637,
902
+ "step": 1460
903
+ },
904
+ {
905
+ "epoch": 0.44,
906
+ "learning_rate": 0.000155832832230908,
907
+ "loss": 0.0617,
908
+ "step": 1470
909
+ },
910
+ {
911
+ "epoch": 0.44,
912
+ "learning_rate": 0.00015553217077570657,
913
+ "loss": 0.0668,
914
+ "step": 1480
915
+ },
916
+ {
917
+ "epoch": 0.45,
918
+ "learning_rate": 0.00015523150932050513,
919
+ "loss": 0.1743,
920
+ "step": 1490
921
+ },
922
+ {
923
+ "epoch": 0.45,
924
+ "learning_rate": 0.00015493084786530366,
925
+ "loss": 0.1356,
926
+ "step": 1500
927
+ },
928
+ {
929
+ "epoch": 0.45,
930
+ "eval_accuracy": 0.9658774896655393,
931
+ "eval_loss": 0.08633118867874146,
932
+ "eval_runtime": 223.9081,
933
+ "eval_samples_per_second": 59.422,
934
+ "eval_steps_per_second": 7.432,
935
+ "step": 1500
936
+ },
937
+ {
938
+ "epoch": 0.45,
939
+ "learning_rate": 0.00015463018641010223,
940
+ "loss": 0.156,
941
+ "step": 1510
942
+ },
943
+ {
944
+ "epoch": 0.46,
945
+ "learning_rate": 0.0001543295249549008,
946
+ "loss": 0.0486,
947
+ "step": 1520
948
+ },
949
+ {
950
+ "epoch": 0.46,
951
+ "learning_rate": 0.00015402886349969933,
952
+ "loss": 0.0653,
953
+ "step": 1530
954
+ },
955
+ {
956
+ "epoch": 0.46,
957
+ "learning_rate": 0.0001537282020444979,
958
+ "loss": 0.1376,
959
+ "step": 1540
960
+ },
961
+ {
962
+ "epoch": 0.47,
963
+ "learning_rate": 0.00015342754058929648,
964
+ "loss": 0.0178,
965
+ "step": 1550
966
+ },
967
+ {
968
+ "epoch": 0.47,
969
+ "learning_rate": 0.00015312687913409501,
970
+ "loss": 0.0637,
971
+ "step": 1560
972
+ },
973
+ {
974
+ "epoch": 0.47,
975
+ "learning_rate": 0.00015282621767889358,
976
+ "loss": 0.0543,
977
+ "step": 1570
978
+ },
979
+ {
980
+ "epoch": 0.48,
981
+ "learning_rate": 0.00015252555622369214,
982
+ "loss": 0.1036,
983
+ "step": 1580
984
+ },
985
+ {
986
+ "epoch": 0.48,
987
+ "learning_rate": 0.00015222489476849068,
988
+ "loss": 0.155,
989
+ "step": 1590
990
+ },
991
+ {
992
+ "epoch": 0.48,
993
+ "learning_rate": 0.00015192423331328924,
994
+ "loss": 0.0635,
995
+ "step": 1600
996
+ },
997
+ {
998
+ "epoch": 0.48,
999
+ "learning_rate": 0.0001516235718580878,
1000
+ "loss": 0.0657,
1001
+ "step": 1610
1002
+ },
1003
+ {
1004
+ "epoch": 0.49,
1005
+ "learning_rate": 0.00015132291040288636,
1006
+ "loss": 0.1205,
1007
+ "step": 1620
1008
+ },
1009
+ {
1010
+ "epoch": 0.49,
1011
+ "learning_rate": 0.0001510222489476849,
1012
+ "loss": 0.1736,
1013
+ "step": 1630
1014
+ },
1015
+ {
1016
+ "epoch": 0.49,
1017
+ "learning_rate": 0.00015072158749248346,
1018
+ "loss": 0.0463,
1019
+ "step": 1640
1020
+ },
1021
+ {
1022
+ "epoch": 0.5,
1023
+ "learning_rate": 0.00015042092603728203,
1024
+ "loss": 0.0882,
1025
+ "step": 1650
1026
+ },
1027
+ {
1028
+ "epoch": 0.5,
1029
+ "learning_rate": 0.0001501202645820806,
1030
+ "loss": 0.0655,
1031
+ "step": 1660
1032
+ },
1033
+ {
1034
+ "epoch": 0.5,
1035
+ "learning_rate": 0.00014981960312687915,
1036
+ "loss": 0.1401,
1037
+ "step": 1670
1038
+ },
1039
+ {
1040
+ "epoch": 0.51,
1041
+ "learning_rate": 0.00014951894167167771,
1042
+ "loss": 0.033,
1043
+ "step": 1680
1044
+ },
1045
+ {
1046
+ "epoch": 0.51,
1047
+ "learning_rate": 0.00014921828021647625,
1048
+ "loss": 0.11,
1049
+ "step": 1690
1050
+ },
1051
+ {
1052
+ "epoch": 0.51,
1053
+ "learning_rate": 0.0001489176187612748,
1054
+ "loss": 0.1682,
1055
+ "step": 1700
1056
+ },
1057
+ {
1058
+ "epoch": 0.51,
1059
+ "learning_rate": 0.00014861695730607338,
1060
+ "loss": 0.1222,
1061
+ "step": 1710
1062
+ },
1063
+ {
1064
+ "epoch": 0.52,
1065
+ "learning_rate": 0.0001483162958508719,
1066
+ "loss": 0.0585,
1067
+ "step": 1720
1068
+ },
1069
+ {
1070
+ "epoch": 0.52,
1071
+ "learning_rate": 0.00014801563439567047,
1072
+ "loss": 0.0721,
1073
+ "step": 1730
1074
+ },
1075
+ {
1076
+ "epoch": 0.52,
1077
+ "learning_rate": 0.00014771497294046904,
1078
+ "loss": 0.1137,
1079
+ "step": 1740
1080
+ },
1081
+ {
1082
+ "epoch": 0.53,
1083
+ "learning_rate": 0.00014741431148526757,
1084
+ "loss": 0.0819,
1085
+ "step": 1750
1086
+ },
1087
+ {
1088
+ "epoch": 0.53,
1089
+ "learning_rate": 0.00014711365003006616,
1090
+ "loss": 0.0821,
1091
+ "step": 1760
1092
+ },
1093
+ {
1094
+ "epoch": 0.53,
1095
+ "learning_rate": 0.00014681298857486473,
1096
+ "loss": 0.1678,
1097
+ "step": 1770
1098
+ },
1099
+ {
1100
+ "epoch": 0.54,
1101
+ "learning_rate": 0.00014651232711966326,
1102
+ "loss": 0.0353,
1103
+ "step": 1780
1104
+ },
1105
+ {
1106
+ "epoch": 0.54,
1107
+ "learning_rate": 0.00014621166566446182,
1108
+ "loss": 0.0858,
1109
+ "step": 1790
1110
+ },
1111
+ {
1112
+ "epoch": 0.54,
1113
+ "learning_rate": 0.0001459110042092604,
1114
+ "loss": 0.113,
1115
+ "step": 1800
1116
+ },
1117
+ {
1118
+ "epoch": 0.54,
1119
+ "learning_rate": 0.00014561034275405892,
1120
+ "loss": 0.103,
1121
+ "step": 1810
1122
+ },
1123
+ {
1124
+ "epoch": 0.55,
1125
+ "learning_rate": 0.0001453096812988575,
1126
+ "loss": 0.0603,
1127
+ "step": 1820
1128
+ },
1129
+ {
1130
+ "epoch": 0.55,
1131
+ "learning_rate": 0.00014500901984365605,
1132
+ "loss": 0.1572,
1133
+ "step": 1830
1134
+ },
1135
+ {
1136
+ "epoch": 0.55,
1137
+ "learning_rate": 0.0001447083583884546,
1138
+ "loss": 0.0367,
1139
+ "step": 1840
1140
+ },
1141
+ {
1142
+ "epoch": 0.56,
1143
+ "learning_rate": 0.00014440769693325315,
1144
+ "loss": 0.0319,
1145
+ "step": 1850
1146
+ },
1147
+ {
1148
+ "epoch": 0.56,
1149
+ "learning_rate": 0.0001441070354780517,
1150
+ "loss": 0.0905,
1151
+ "step": 1860
1152
+ },
1153
+ {
1154
+ "epoch": 0.56,
1155
+ "learning_rate": 0.0001438063740228503,
1156
+ "loss": 0.0402,
1157
+ "step": 1870
1158
+ },
1159
+ {
1160
+ "epoch": 0.57,
1161
+ "learning_rate": 0.00014350571256764884,
1162
+ "loss": 0.0476,
1163
+ "step": 1880
1164
+ },
1165
+ {
1166
+ "epoch": 0.57,
1167
+ "learning_rate": 0.0001432050511124474,
1168
+ "loss": 0.066,
1169
+ "step": 1890
1170
+ },
1171
+ {
1172
+ "epoch": 0.57,
1173
+ "learning_rate": 0.00014290438965724596,
1174
+ "loss": 0.0531,
1175
+ "step": 1900
1176
+ },
1177
+ {
1178
+ "epoch": 0.57,
1179
+ "learning_rate": 0.0001426037282020445,
1180
+ "loss": 0.0641,
1181
+ "step": 1910
1182
+ },
1183
+ {
1184
+ "epoch": 0.58,
1185
+ "learning_rate": 0.00014230306674684306,
1186
+ "loss": 0.0147,
1187
+ "step": 1920
1188
+ },
1189
+ {
1190
+ "epoch": 0.58,
1191
+ "learning_rate": 0.00014200240529164162,
1192
+ "loss": 0.07,
1193
+ "step": 1930
1194
+ },
1195
+ {
1196
+ "epoch": 0.58,
1197
+ "learning_rate": 0.00014170174383644016,
1198
+ "loss": 0.0866,
1199
+ "step": 1940
1200
+ },
1201
+ {
1202
+ "epoch": 0.59,
1203
+ "learning_rate": 0.00014140108238123872,
1204
+ "loss": 0.0586,
1205
+ "step": 1950
1206
+ },
1207
+ {
1208
+ "epoch": 0.59,
1209
+ "learning_rate": 0.00014110042092603729,
1210
+ "loss": 0.1149,
1211
+ "step": 1960
1212
+ },
1213
+ {
1214
+ "epoch": 0.59,
1215
+ "learning_rate": 0.00014079975947083582,
1216
+ "loss": 0.0778,
1217
+ "step": 1970
1218
+ },
1219
+ {
1220
+ "epoch": 0.6,
1221
+ "learning_rate": 0.0001404990980156344,
1222
+ "loss": 0.1018,
1223
+ "step": 1980
1224
+ },
1225
+ {
1226
+ "epoch": 0.6,
1227
+ "learning_rate": 0.00014019843656043297,
1228
+ "loss": 0.0708,
1229
+ "step": 1990
1230
+ },
1231
+ {
1232
+ "epoch": 0.6,
1233
+ "learning_rate": 0.0001398977751052315,
1234
+ "loss": 0.1092,
1235
+ "step": 2000
1236
+ },
1237
+ {
1238
+ "epoch": 0.6,
1239
+ "eval_accuracy": 0.9786546411123638,
1240
+ "eval_loss": 0.056445054709911346,
1241
+ "eval_runtime": 224.4822,
1242
+ "eval_samples_per_second": 59.27,
1243
+ "eval_steps_per_second": 7.413,
1244
+ "step": 2000
1245
+ },
1246
+ {
1247
+ "epoch": 0.6,
1248
+ "learning_rate": 0.00013959711365003007,
1249
+ "loss": 0.027,
1250
+ "step": 2010
1251
+ },
1252
+ {
1253
+ "epoch": 0.61,
1254
+ "learning_rate": 0.00013929645219482864,
1255
+ "loss": 0.0617,
1256
+ "step": 2020
1257
+ },
1258
+ {
1259
+ "epoch": 0.61,
1260
+ "learning_rate": 0.0001389957907396272,
1261
+ "loss": 0.0603,
1262
+ "step": 2030
1263
+ },
1264
+ {
1265
+ "epoch": 0.61,
1266
+ "learning_rate": 0.00013869512928442573,
1267
+ "loss": 0.0865,
1268
+ "step": 2040
1269
+ },
1270
+ {
1271
+ "epoch": 0.62,
1272
+ "learning_rate": 0.0001383944678292243,
1273
+ "loss": 0.0379,
1274
+ "step": 2050
1275
+ },
1276
+ {
1277
+ "epoch": 0.62,
1278
+ "learning_rate": 0.00013809380637402286,
1279
+ "loss": 0.0408,
1280
+ "step": 2060
1281
+ },
1282
+ {
1283
+ "epoch": 0.62,
1284
+ "learning_rate": 0.0001377931449188214,
1285
+ "loss": 0.099,
1286
+ "step": 2070
1287
+ },
1288
+ {
1289
+ "epoch": 0.63,
1290
+ "learning_rate": 0.00013749248346361999,
1291
+ "loss": 0.0564,
1292
+ "step": 2080
1293
+ },
1294
+ {
1295
+ "epoch": 0.63,
1296
+ "learning_rate": 0.00013719182200841855,
1297
+ "loss": 0.0518,
1298
+ "step": 2090
1299
+ },
1300
+ {
1301
+ "epoch": 0.63,
1302
+ "learning_rate": 0.00013689116055321708,
1303
+ "loss": 0.0526,
1304
+ "step": 2100
1305
+ },
1306
+ {
1307
+ "epoch": 0.63,
1308
+ "learning_rate": 0.00013659049909801565,
1309
+ "loss": 0.0592,
1310
+ "step": 2110
1311
+ },
1312
+ {
1313
+ "epoch": 0.64,
1314
+ "learning_rate": 0.0001362898376428142,
1315
+ "loss": 0.2238,
1316
+ "step": 2120
1317
+ },
1318
+ {
1319
+ "epoch": 0.64,
1320
+ "learning_rate": 0.00013598917618761275,
1321
+ "loss": 0.0427,
1322
+ "step": 2130
1323
+ },
1324
+ {
1325
+ "epoch": 0.64,
1326
+ "learning_rate": 0.0001356885147324113,
1327
+ "loss": 0.0417,
1328
+ "step": 2140
1329
+ },
1330
+ {
1331
+ "epoch": 0.65,
1332
+ "learning_rate": 0.00013538785327720987,
1333
+ "loss": 0.0126,
1334
+ "step": 2150
1335
+ },
1336
+ {
1337
+ "epoch": 0.65,
1338
+ "learning_rate": 0.0001350871918220084,
1339
+ "loss": 0.2194,
1340
+ "step": 2160
1341
+ },
1342
+ {
1343
+ "epoch": 0.65,
1344
+ "learning_rate": 0.00013478653036680697,
1345
+ "loss": 0.0718,
1346
+ "step": 2170
1347
+ },
1348
+ {
1349
+ "epoch": 0.66,
1350
+ "learning_rate": 0.00013448586891160553,
1351
+ "loss": 0.0855,
1352
+ "step": 2180
1353
+ },
1354
+ {
1355
+ "epoch": 0.66,
1356
+ "learning_rate": 0.0001341852074564041,
1357
+ "loss": 0.0684,
1358
+ "step": 2190
1359
+ },
1360
+ {
1361
+ "epoch": 0.66,
1362
+ "learning_rate": 0.00013388454600120266,
1363
+ "loss": 0.0941,
1364
+ "step": 2200
1365
+ },
1366
+ {
1367
+ "epoch": 0.66,
1368
+ "learning_rate": 0.00013358388454600122,
1369
+ "loss": 0.0329,
1370
+ "step": 2210
1371
+ },
1372
+ {
1373
+ "epoch": 0.67,
1374
+ "learning_rate": 0.00013328322309079976,
1375
+ "loss": 0.0832,
1376
+ "step": 2220
1377
+ },
1378
+ {
1379
+ "epoch": 0.67,
1380
+ "learning_rate": 0.00013298256163559832,
1381
+ "loss": 0.0602,
1382
+ "step": 2230
1383
+ },
1384
+ {
1385
+ "epoch": 0.67,
1386
+ "learning_rate": 0.00013268190018039688,
1387
+ "loss": 0.0535,
1388
+ "step": 2240
1389
+ },
1390
+ {
1391
+ "epoch": 0.68,
1392
+ "learning_rate": 0.00013238123872519545,
1393
+ "loss": 0.0427,
1394
+ "step": 2250
1395
+ },
1396
+ {
1397
+ "epoch": 0.68,
1398
+ "learning_rate": 0.00013208057726999398,
1399
+ "loss": 0.0318,
1400
+ "step": 2260
1401
+ },
1402
+ {
1403
+ "epoch": 0.68,
1404
+ "learning_rate": 0.00013177991581479255,
1405
+ "loss": 0.1045,
1406
+ "step": 2270
1407
+ },
1408
+ {
1409
+ "epoch": 0.69,
1410
+ "learning_rate": 0.0001314792543595911,
1411
+ "loss": 0.0226,
1412
+ "step": 2280
1413
+ },
1414
+ {
1415
+ "epoch": 0.69,
1416
+ "learning_rate": 0.00013117859290438967,
1417
+ "loss": 0.0407,
1418
+ "step": 2290
1419
+ },
1420
+ {
1421
+ "epoch": 0.69,
1422
+ "learning_rate": 0.00013087793144918823,
1423
+ "loss": 0.019,
1424
+ "step": 2300
1425
+ },
1426
+ {
1427
+ "epoch": 0.69,
1428
+ "learning_rate": 0.0001305772699939868,
1429
+ "loss": 0.0072,
1430
+ "step": 2310
1431
+ },
1432
+ {
1433
+ "epoch": 0.7,
1434
+ "learning_rate": 0.00013027660853878533,
1435
+ "loss": 0.157,
1436
+ "step": 2320
1437
+ },
1438
+ {
1439
+ "epoch": 0.7,
1440
+ "learning_rate": 0.0001299759470835839,
1441
+ "loss": 0.0205,
1442
+ "step": 2330
1443
+ },
1444
+ {
1445
+ "epoch": 0.7,
1446
+ "learning_rate": 0.00012967528562838246,
1447
+ "loss": 0.0613,
1448
+ "step": 2340
1449
+ },
1450
+ {
1451
+ "epoch": 0.71,
1452
+ "learning_rate": 0.000129374624173181,
1453
+ "loss": 0.0833,
1454
+ "step": 2350
1455
+ },
1456
+ {
1457
+ "epoch": 0.71,
1458
+ "learning_rate": 0.00012907396271797956,
1459
+ "loss": 0.0542,
1460
+ "step": 2360
1461
+ },
1462
+ {
1463
+ "epoch": 0.71,
1464
+ "learning_rate": 0.00012877330126277812,
1465
+ "loss": 0.0548,
1466
+ "step": 2370
1467
+ },
1468
+ {
1469
+ "epoch": 0.72,
1470
+ "learning_rate": 0.00012847263980757666,
1471
+ "loss": 0.0838,
1472
+ "step": 2380
1473
+ },
1474
+ {
1475
+ "epoch": 0.72,
1476
+ "learning_rate": 0.00012817197835237522,
1477
+ "loss": 0.0457,
1478
+ "step": 2390
1479
+ },
1480
+ {
1481
+ "epoch": 0.72,
1482
+ "learning_rate": 0.0001278713168971738,
1483
+ "loss": 0.0445,
1484
+ "step": 2400
1485
+ },
1486
+ {
1487
+ "epoch": 0.72,
1488
+ "learning_rate": 0.00012757065544197234,
1489
+ "loss": 0.0267,
1490
+ "step": 2410
1491
+ },
1492
+ {
1493
+ "epoch": 0.73,
1494
+ "learning_rate": 0.0001272699939867709,
1495
+ "loss": 0.0546,
1496
+ "step": 2420
1497
+ },
1498
+ {
1499
+ "epoch": 0.73,
1500
+ "learning_rate": 0.00012696933253156947,
1501
+ "loss": 0.0086,
1502
+ "step": 2430
1503
+ },
1504
+ {
1505
+ "epoch": 0.73,
1506
+ "learning_rate": 0.00012669873722188818,
1507
+ "loss": 0.03,
1508
+ "step": 2440
1509
+ },
1510
+ {
1511
+ "epoch": 0.74,
1512
+ "learning_rate": 0.0001263980757666867,
1513
+ "loss": 0.0315,
1514
+ "step": 2450
1515
+ },
1516
+ {
1517
+ "epoch": 0.74,
1518
+ "learning_rate": 0.00012609741431148528,
1519
+ "loss": 0.1645,
1520
+ "step": 2460
1521
+ },
1522
+ {
1523
+ "epoch": 0.74,
1524
+ "learning_rate": 0.00012579675285628384,
1525
+ "loss": 0.1208,
1526
+ "step": 2470
1527
+ },
1528
+ {
1529
+ "epoch": 0.75,
1530
+ "learning_rate": 0.00012549609140108237,
1531
+ "loss": 0.0735,
1532
+ "step": 2480
1533
+ },
1534
+ {
1535
+ "epoch": 0.75,
1536
+ "learning_rate": 0.00012519542994588094,
1537
+ "loss": 0.0807,
1538
+ "step": 2490
1539
+ },
1540
+ {
1541
+ "epoch": 0.75,
1542
+ "learning_rate": 0.0001248947684906795,
1543
+ "loss": 0.0952,
1544
+ "step": 2500
1545
+ },
1546
+ {
1547
+ "epoch": 0.75,
1548
+ "eval_accuracy": 0.9848177376925967,
1549
+ "eval_loss": 0.04439860209822655,
1550
+ "eval_runtime": 224.8259,
1551
+ "eval_samples_per_second": 59.179,
1552
+ "eval_steps_per_second": 7.401,
1553
+ "step": 2500
1554
+ },
1555
+ {
1556
+ "epoch": 0.75,
1557
+ "learning_rate": 0.00012459410703547804,
1558
+ "loss": 0.1217,
1559
+ "step": 2510
1560
+ },
1561
+ {
1562
+ "epoch": 0.76,
1563
+ "learning_rate": 0.00012429344558027663,
1564
+ "loss": 0.0596,
1565
+ "step": 2520
1566
+ },
1567
+ {
1568
+ "epoch": 0.76,
1569
+ "learning_rate": 0.0001239927841250752,
1570
+ "loss": 0.0748,
1571
+ "step": 2530
1572
+ },
1573
+ {
1574
+ "epoch": 0.76,
1575
+ "learning_rate": 0.00012369212266987372,
1576
+ "loss": 0.094,
1577
+ "step": 2540
1578
+ },
1579
+ {
1580
+ "epoch": 0.77,
1581
+ "learning_rate": 0.0001233914612146723,
1582
+ "loss": 0.0311,
1583
+ "step": 2550
1584
+ },
1585
+ {
1586
+ "epoch": 0.77,
1587
+ "learning_rate": 0.00012309079975947085,
1588
+ "loss": 0.1084,
1589
+ "step": 2560
1590
+ },
1591
+ {
1592
+ "epoch": 0.77,
1593
+ "learning_rate": 0.00012279013830426939,
1594
+ "loss": 0.1171,
1595
+ "step": 2570
1596
+ },
1597
+ {
1598
+ "epoch": 0.78,
1599
+ "learning_rate": 0.00012248947684906795,
1600
+ "loss": 0.0299,
1601
+ "step": 2580
1602
+ },
1603
+ {
1604
+ "epoch": 0.78,
1605
+ "learning_rate": 0.0001221888153938665,
1606
+ "loss": 0.0503,
1607
+ "step": 2590
1608
+ },
1609
+ {
1610
+ "epoch": 0.78,
1611
+ "learning_rate": 0.00012188815393866506,
1612
+ "loss": 0.0285,
1613
+ "step": 2600
1614
+ },
1615
+ {
1616
+ "epoch": 0.78,
1617
+ "learning_rate": 0.00012158749248346361,
1618
+ "loss": 0.0198,
1619
+ "step": 2610
1620
+ },
1621
+ {
1622
+ "epoch": 0.79,
1623
+ "learning_rate": 0.0001212868310282622,
1624
+ "loss": 0.0203,
1625
+ "step": 2620
1626
+ },
1627
+ {
1628
+ "epoch": 0.79,
1629
+ "learning_rate": 0.00012098616957306075,
1630
+ "loss": 0.0611,
1631
+ "step": 2630
1632
+ },
1633
+ {
1634
+ "epoch": 0.79,
1635
+ "learning_rate": 0.0001206855081178593,
1636
+ "loss": 0.046,
1637
+ "step": 2640
1638
+ },
1639
+ {
1640
+ "epoch": 0.8,
1641
+ "learning_rate": 0.00012038484666265786,
1642
+ "loss": 0.0691,
1643
+ "step": 2650
1644
+ },
1645
+ {
1646
+ "epoch": 0.8,
1647
+ "learning_rate": 0.00012008418520745641,
1648
+ "loss": 0.0845,
1649
+ "step": 2660
1650
+ },
1651
+ {
1652
+ "epoch": 0.8,
1653
+ "learning_rate": 0.00011978352375225496,
1654
+ "loss": 0.0176,
1655
+ "step": 2670
1656
+ },
1657
+ {
1658
+ "epoch": 0.81,
1659
+ "learning_rate": 0.00011948286229705352,
1660
+ "loss": 0.0469,
1661
+ "step": 2680
1662
+ },
1663
+ {
1664
+ "epoch": 0.81,
1665
+ "learning_rate": 0.00011918220084185207,
1666
+ "loss": 0.0545,
1667
+ "step": 2690
1668
+ },
1669
+ {
1670
+ "epoch": 0.81,
1671
+ "learning_rate": 0.00011888153938665064,
1672
+ "loss": 0.1596,
1673
+ "step": 2700
1674
+ },
1675
+ {
1676
+ "epoch": 0.81,
1677
+ "learning_rate": 0.00011858087793144918,
1678
+ "loss": 0.0716,
1679
+ "step": 2710
1680
+ },
1681
+ {
1682
+ "epoch": 0.82,
1683
+ "learning_rate": 0.00011828021647624773,
1684
+ "loss": 0.0324,
1685
+ "step": 2720
1686
+ },
1687
+ {
1688
+ "epoch": 0.82,
1689
+ "learning_rate": 0.00011797955502104632,
1690
+ "loss": 0.0719,
1691
+ "step": 2730
1692
+ },
1693
+ {
1694
+ "epoch": 0.82,
1695
+ "learning_rate": 0.00011767889356584487,
1696
+ "loss": 0.0181,
1697
+ "step": 2740
1698
+ },
1699
+ {
1700
+ "epoch": 0.83,
1701
+ "learning_rate": 0.00011737823211064342,
1702
+ "loss": 0.0793,
1703
+ "step": 2750
1704
+ },
1705
+ {
1706
+ "epoch": 0.83,
1707
+ "learning_rate": 0.00011707757065544199,
1708
+ "loss": 0.0834,
1709
+ "step": 2760
1710
+ },
1711
+ {
1712
+ "epoch": 0.83,
1713
+ "learning_rate": 0.00011677690920024053,
1714
+ "loss": 0.0376,
1715
+ "step": 2770
1716
+ },
1717
+ {
1718
+ "epoch": 0.84,
1719
+ "learning_rate": 0.0001164762477450391,
1720
+ "loss": 0.0378,
1721
+ "step": 2780
1722
+ },
1723
+ {
1724
+ "epoch": 0.84,
1725
+ "learning_rate": 0.00011617558628983765,
1726
+ "loss": 0.0378,
1727
+ "step": 2790
1728
+ },
1729
+ {
1730
+ "epoch": 0.84,
1731
+ "learning_rate": 0.0001158749248346362,
1732
+ "loss": 0.0708,
1733
+ "step": 2800
1734
+ },
1735
+ {
1736
+ "epoch": 0.84,
1737
+ "learning_rate": 0.00011557426337943476,
1738
+ "loss": 0.1401,
1739
+ "step": 2810
1740
+ },
1741
+ {
1742
+ "epoch": 0.85,
1743
+ "learning_rate": 0.00011527360192423331,
1744
+ "loss": 0.0351,
1745
+ "step": 2820
1746
+ },
1747
+ {
1748
+ "epoch": 0.85,
1749
+ "learning_rate": 0.00011497294046903186,
1750
+ "loss": 0.0362,
1751
+ "step": 2830
1752
+ },
1753
+ {
1754
+ "epoch": 0.85,
1755
+ "learning_rate": 0.00011467227901383045,
1756
+ "loss": 0.0617,
1757
+ "step": 2840
1758
+ },
1759
+ {
1760
+ "epoch": 0.86,
1761
+ "learning_rate": 0.000114371617558629,
1762
+ "loss": 0.0712,
1763
+ "step": 2850
1764
+ },
1765
+ {
1766
+ "epoch": 0.86,
1767
+ "learning_rate": 0.00011407095610342755,
1768
+ "loss": 0.0372,
1769
+ "step": 2860
1770
+ },
1771
+ {
1772
+ "epoch": 0.86,
1773
+ "learning_rate": 0.00011377029464822611,
1774
+ "loss": 0.0871,
1775
+ "step": 2870
1776
+ },
1777
+ {
1778
+ "epoch": 0.87,
1779
+ "learning_rate": 0.00011346963319302466,
1780
+ "loss": 0.0474,
1781
+ "step": 2880
1782
+ },
1783
+ {
1784
+ "epoch": 0.87,
1785
+ "learning_rate": 0.00011316897173782322,
1786
+ "loss": 0.0371,
1787
+ "step": 2890
1788
+ },
1789
+ {
1790
+ "epoch": 0.87,
1791
+ "learning_rate": 0.00011286831028262177,
1792
+ "loss": 0.0877,
1793
+ "step": 2900
1794
+ },
1795
+ {
1796
+ "epoch": 0.87,
1797
+ "learning_rate": 0.00011256764882742032,
1798
+ "loss": 0.0529,
1799
+ "step": 2910
1800
+ },
1801
+ {
1802
+ "epoch": 0.88,
1803
+ "learning_rate": 0.00011226698737221888,
1804
+ "loss": 0.0433,
1805
+ "step": 2920
1806
+ },
1807
+ {
1808
+ "epoch": 0.88,
1809
+ "learning_rate": 0.00011196632591701743,
1810
+ "loss": 0.1294,
1811
+ "step": 2930
1812
+ },
1813
+ {
1814
+ "epoch": 0.88,
1815
+ "learning_rate": 0.00011166566446181601,
1816
+ "loss": 0.066,
1817
+ "step": 2940
1818
+ },
1819
+ {
1820
+ "epoch": 0.89,
1821
+ "learning_rate": 0.00011136500300661457,
1822
+ "loss": 0.0566,
1823
+ "step": 2950
1824
+ },
1825
+ {
1826
+ "epoch": 0.89,
1827
+ "learning_rate": 0.00011106434155141312,
1828
+ "loss": 0.0622,
1829
+ "step": 2960
1830
+ },
1831
+ {
1832
+ "epoch": 0.89,
1833
+ "learning_rate": 0.00011076368009621167,
1834
+ "loss": 0.0667,
1835
+ "step": 2970
1836
+ },
1837
+ {
1838
+ "epoch": 0.9,
1839
+ "learning_rate": 0.00011046301864101023,
1840
+ "loss": 0.0818,
1841
+ "step": 2980
1842
+ },
1843
+ {
1844
+ "epoch": 0.9,
1845
+ "learning_rate": 0.00011016235718580878,
1846
+ "loss": 0.049,
1847
+ "step": 2990
1848
+ },
1849
+ {
1850
+ "epoch": 0.9,
1851
+ "learning_rate": 0.00010986169573060735,
1852
+ "loss": 0.0594,
1853
+ "step": 3000
1854
+ },
1855
+ {
1856
+ "epoch": 0.9,
1857
+ "eval_accuracy": 0.9827884254039835,
1858
+ "eval_loss": 0.04768187925219536,
1859
+ "eval_runtime": 228.4877,
1860
+ "eval_samples_per_second": 58.231,
1861
+ "eval_steps_per_second": 7.283,
1862
+ "step": 3000
1863
+ },
1864
+ {
1865
+ "epoch": 0.9,
1866
+ "learning_rate": 0.0001095610342754059,
1867
+ "loss": 0.0586,
1868
+ "step": 3010
1869
+ },
1870
+ {
1871
+ "epoch": 0.91,
1872
+ "learning_rate": 0.00010926037282020444,
1873
+ "loss": 0.1068,
1874
+ "step": 3020
1875
+ },
1876
+ {
1877
+ "epoch": 0.91,
1878
+ "learning_rate": 0.00010895971136500301,
1879
+ "loss": 0.0125,
1880
+ "step": 3030
1881
+ },
1882
+ {
1883
+ "epoch": 0.91,
1884
+ "learning_rate": 0.00010865904990980156,
1885
+ "loss": 0.0833,
1886
+ "step": 3040
1887
+ },
1888
+ {
1889
+ "epoch": 0.92,
1890
+ "learning_rate": 0.00010835838845460013,
1891
+ "loss": 0.0705,
1892
+ "step": 3050
1893
+ },
1894
+ {
1895
+ "epoch": 0.92,
1896
+ "learning_rate": 0.0001080577269993987,
1897
+ "loss": 0.0339,
1898
+ "step": 3060
1899
+ },
1900
+ {
1901
+ "epoch": 0.92,
1902
+ "learning_rate": 0.00010775706554419725,
1903
+ "loss": 0.0893,
1904
+ "step": 3070
1905
+ },
1906
+ {
1907
+ "epoch": 0.93,
1908
+ "learning_rate": 0.0001074564040889958,
1909
+ "loss": 0.0679,
1910
+ "step": 3080
1911
+ },
1912
+ {
1913
+ "epoch": 0.93,
1914
+ "learning_rate": 0.00010715574263379436,
1915
+ "loss": 0.0196,
1916
+ "step": 3090
1917
+ },
1918
+ {
1919
+ "epoch": 0.93,
1920
+ "learning_rate": 0.0001068550811785929,
1921
+ "loss": 0.093,
1922
+ "step": 3100
1923
+ },
1924
+ {
1925
+ "epoch": 0.94,
1926
+ "learning_rate": 0.00010655441972339147,
1927
+ "loss": 0.0787,
1928
+ "step": 3110
1929
+ },
1930
+ {
1931
+ "epoch": 0.94,
1932
+ "learning_rate": 0.00010625375826819002,
1933
+ "loss": 0.0646,
1934
+ "step": 3120
1935
+ },
1936
+ {
1937
+ "epoch": 0.94,
1938
+ "learning_rate": 0.00010595309681298857,
1939
+ "loss": 0.0542,
1940
+ "step": 3130
1941
+ },
1942
+ {
1943
+ "epoch": 0.94,
1944
+ "learning_rate": 0.00010565243535778713,
1945
+ "loss": 0.0393,
1946
+ "step": 3140
1947
+ },
1948
+ {
1949
+ "epoch": 0.95,
1950
+ "learning_rate": 0.00010535177390258571,
1951
+ "loss": 0.0271,
1952
+ "step": 3150
1953
+ },
1954
+ {
1955
+ "epoch": 0.95,
1956
+ "learning_rate": 0.00010505111244738426,
1957
+ "loss": 0.0467,
1958
+ "step": 3160
1959
+ },
1960
+ {
1961
+ "epoch": 0.95,
1962
+ "learning_rate": 0.00010475045099218282,
1963
+ "loss": 0.1462,
1964
+ "step": 3170
1965
+ },
1966
+ {
1967
+ "epoch": 0.96,
1968
+ "learning_rate": 0.00010444978953698137,
1969
+ "loss": 0.0479,
1970
+ "step": 3180
1971
+ },
1972
+ {
1973
+ "epoch": 0.96,
1974
+ "learning_rate": 0.00010414912808177992,
1975
+ "loss": 0.032,
1976
+ "step": 3190
1977
+ },
1978
+ {
1979
+ "epoch": 0.96,
1980
+ "learning_rate": 0.00010384846662657848,
1981
+ "loss": 0.0969,
1982
+ "step": 3200
1983
+ },
1984
+ {
1985
+ "epoch": 0.97,
1986
+ "learning_rate": 0.00010354780517137703,
1987
+ "loss": 0.0063,
1988
+ "step": 3210
1989
+ },
1990
+ {
1991
+ "epoch": 0.97,
1992
+ "learning_rate": 0.0001032471437161756,
1993
+ "loss": 0.098,
1994
+ "step": 3220
1995
+ },
1996
+ {
1997
+ "epoch": 0.97,
1998
+ "learning_rate": 0.00010294648226097414,
1999
+ "loss": 0.0553,
2000
+ "step": 3230
2001
+ },
2002
+ {
2003
+ "epoch": 0.97,
2004
+ "learning_rate": 0.00010264582080577269,
2005
+ "loss": 0.0217,
2006
+ "step": 3240
2007
+ },
2008
+ {
2009
+ "epoch": 0.98,
2010
+ "learning_rate": 0.00010234515935057125,
2011
+ "loss": 0.0032,
2012
+ "step": 3250
2013
+ },
2014
+ {
2015
+ "epoch": 0.98,
2016
+ "learning_rate": 0.00010204449789536983,
2017
+ "loss": 0.1247,
2018
+ "step": 3260
2019
+ },
2020
+ {
2021
+ "epoch": 0.98,
2022
+ "learning_rate": 0.00010174383644016838,
2023
+ "loss": 0.0922,
2024
+ "step": 3270
2025
+ },
2026
+ {
2027
+ "epoch": 0.99,
2028
+ "learning_rate": 0.00010144317498496694,
2029
+ "loss": 0.0458,
2030
+ "step": 3280
2031
+ },
2032
+ {
2033
+ "epoch": 0.99,
2034
+ "learning_rate": 0.00010114251352976549,
2035
+ "loss": 0.0206,
2036
+ "step": 3290
2037
+ },
2038
+ {
2039
+ "epoch": 0.99,
2040
+ "learning_rate": 0.00010084185207456404,
2041
+ "loss": 0.0151,
2042
+ "step": 3300
2043
+ },
2044
+ {
2045
+ "epoch": 1.0,
2046
+ "learning_rate": 0.0001005411906193626,
2047
+ "loss": 0.1185,
2048
+ "step": 3310
2049
+ },
2050
+ {
2051
+ "epoch": 1.0,
2052
+ "learning_rate": 0.00010024052916416115,
2053
+ "loss": 0.0413,
2054
+ "step": 3320
2055
+ },
2056
+ {
2057
+ "epoch": 1.0,
2058
+ "learning_rate": 9.993986770895972e-05,
2059
+ "loss": 0.0461,
2060
+ "step": 3330
2061
+ },
2062
+ {
2063
+ "epoch": 1.0,
2064
+ "learning_rate": 9.963920625375828e-05,
2065
+ "loss": 0.0053,
2066
+ "step": 3340
2067
+ },
2068
+ {
2069
+ "epoch": 1.01,
2070
+ "learning_rate": 9.933854479855683e-05,
2071
+ "loss": 0.0409,
2072
+ "step": 3350
2073
+ },
2074
+ {
2075
+ "epoch": 1.01,
2076
+ "learning_rate": 9.903788334335539e-05,
2077
+ "loss": 0.0906,
2078
+ "step": 3360
2079
+ },
2080
+ {
2081
+ "epoch": 1.01,
2082
+ "learning_rate": 9.873722188815394e-05,
2083
+ "loss": 0.0093,
2084
+ "step": 3370
2085
+ },
2086
+ {
2087
+ "epoch": 1.02,
2088
+ "learning_rate": 9.843656043295249e-05,
2089
+ "loss": 0.0147,
2090
+ "step": 3380
2091
+ },
2092
+ {
2093
+ "epoch": 1.02,
2094
+ "learning_rate": 9.813589897775107e-05,
2095
+ "loss": 0.1454,
2096
+ "step": 3390
2097
+ },
2098
+ {
2099
+ "epoch": 1.02,
2100
+ "learning_rate": 9.783523752254962e-05,
2101
+ "loss": 0.1407,
2102
+ "step": 3400
2103
+ },
2104
+ {
2105
+ "epoch": 1.03,
2106
+ "learning_rate": 9.753457606734817e-05,
2107
+ "loss": 0.0622,
2108
+ "step": 3410
2109
+ },
2110
+ {
2111
+ "epoch": 1.03,
2112
+ "learning_rate": 9.723391461214673e-05,
2113
+ "loss": 0.0187,
2114
+ "step": 3420
2115
+ },
2116
+ {
2117
+ "epoch": 1.03,
2118
+ "learning_rate": 9.693325315694528e-05,
2119
+ "loss": 0.0404,
2120
+ "step": 3430
2121
+ },
2122
+ {
2123
+ "epoch": 1.03,
2124
+ "learning_rate": 9.663259170174384e-05,
2125
+ "loss": 0.0373,
2126
+ "step": 3440
2127
+ },
2128
+ {
2129
+ "epoch": 1.04,
2130
+ "learning_rate": 9.63319302465424e-05,
2131
+ "loss": 0.0675,
2132
+ "step": 3450
2133
+ },
2134
+ {
2135
+ "epoch": 1.04,
2136
+ "learning_rate": 9.603126879134095e-05,
2137
+ "loss": 0.039,
2138
+ "step": 3460
2139
+ },
2140
+ {
2141
+ "epoch": 1.04,
2142
+ "learning_rate": 9.573060733613952e-05,
2143
+ "loss": 0.0434,
2144
+ "step": 3470
2145
+ },
2146
+ {
2147
+ "epoch": 1.05,
2148
+ "learning_rate": 9.542994588093807e-05,
2149
+ "loss": 0.0238,
2150
+ "step": 3480
2151
+ },
2152
+ {
2153
+ "epoch": 1.05,
2154
+ "learning_rate": 9.512928442573662e-05,
2155
+ "loss": 0.0066,
2156
+ "step": 3490
2157
+ },
2158
+ {
2159
+ "epoch": 1.05,
2160
+ "learning_rate": 9.482862297053519e-05,
2161
+ "loss": 0.1134,
2162
+ "step": 3500
2163
+ },
2164
+ {
2165
+ "epoch": 1.05,
2166
+ "eval_accuracy": 0.9868470499812101,
2167
+ "eval_loss": 0.03534827008843422,
2168
+ "eval_runtime": 226.1058,
2169
+ "eval_samples_per_second": 58.844,
2170
+ "eval_steps_per_second": 7.359,
2171
+ "step": 3500
2172
+ },
2173
+ {
2174
+ "epoch": 1.06,
2175
+ "learning_rate": 9.452796151533374e-05,
2176
+ "loss": 0.0306,
2177
+ "step": 3510
2178
+ },
2179
+ {
2180
+ "epoch": 1.06,
2181
+ "learning_rate": 9.422730006013229e-05,
2182
+ "loss": 0.0366,
2183
+ "step": 3520
2184
+ },
2185
+ {
2186
+ "epoch": 1.06,
2187
+ "learning_rate": 9.392663860493085e-05,
2188
+ "loss": 0.0669,
2189
+ "step": 3530
2190
+ },
2191
+ {
2192
+ "epoch": 1.06,
2193
+ "learning_rate": 9.36259771497294e-05,
2194
+ "loss": 0.011,
2195
+ "step": 3540
2196
+ },
2197
+ {
2198
+ "epoch": 1.07,
2199
+ "learning_rate": 9.332531569452797e-05,
2200
+ "loss": 0.0477,
2201
+ "step": 3550
2202
+ },
2203
+ {
2204
+ "epoch": 1.07,
2205
+ "learning_rate": 9.302465423932653e-05,
2206
+ "loss": 0.0708,
2207
+ "step": 3560
2208
+ },
2209
+ {
2210
+ "epoch": 1.07,
2211
+ "learning_rate": 9.272399278412508e-05,
2212
+ "loss": 0.0248,
2213
+ "step": 3570
2214
+ },
2215
+ {
2216
+ "epoch": 1.08,
2217
+ "learning_rate": 9.242333132892364e-05,
2218
+ "loss": 0.0409,
2219
+ "step": 3580
2220
+ },
2221
+ {
2222
+ "epoch": 1.08,
2223
+ "learning_rate": 9.212266987372219e-05,
2224
+ "loss": 0.0399,
2225
+ "step": 3590
2226
+ },
2227
+ {
2228
+ "epoch": 1.08,
2229
+ "learning_rate": 9.182200841852075e-05,
2230
+ "loss": 0.0477,
2231
+ "step": 3600
2232
+ },
2233
+ {
2234
+ "epoch": 1.09,
2235
+ "learning_rate": 9.152134696331932e-05,
2236
+ "loss": 0.0016,
2237
+ "step": 3610
2238
+ },
2239
+ {
2240
+ "epoch": 1.09,
2241
+ "learning_rate": 9.122068550811786e-05,
2242
+ "loss": 0.0049,
2243
+ "step": 3620
2244
+ },
2245
+ {
2246
+ "epoch": 1.09,
2247
+ "learning_rate": 9.092002405291641e-05,
2248
+ "loss": 0.0252,
2249
+ "step": 3630
2250
+ },
2251
+ {
2252
+ "epoch": 1.09,
2253
+ "learning_rate": 9.061936259771498e-05,
2254
+ "loss": 0.106,
2255
+ "step": 3640
2256
+ },
2257
+ {
2258
+ "epoch": 1.1,
2259
+ "learning_rate": 9.031870114251353e-05,
2260
+ "loss": 0.0392,
2261
+ "step": 3650
2262
+ },
2263
+ {
2264
+ "epoch": 1.1,
2265
+ "learning_rate": 9.001803968731209e-05,
2266
+ "loss": 0.0101,
2267
+ "step": 3660
2268
+ },
2269
+ {
2270
+ "epoch": 1.1,
2271
+ "learning_rate": 8.971737823211065e-05,
2272
+ "loss": 0.0384,
2273
+ "step": 3670
2274
+ },
2275
+ {
2276
+ "epoch": 1.11,
2277
+ "learning_rate": 8.94167167769092e-05,
2278
+ "loss": 0.0802,
2279
+ "step": 3680
2280
+ },
2281
+ {
2282
+ "epoch": 1.11,
2283
+ "learning_rate": 8.911605532170776e-05,
2284
+ "loss": 0.0287,
2285
+ "step": 3690
2286
+ },
2287
+ {
2288
+ "epoch": 1.11,
2289
+ "learning_rate": 8.881539386650631e-05,
2290
+ "loss": 0.0426,
2291
+ "step": 3700
2292
+ },
2293
+ {
2294
+ "epoch": 1.12,
2295
+ "learning_rate": 8.851473241130488e-05,
2296
+ "loss": 0.015,
2297
+ "step": 3710
2298
+ },
2299
+ {
2300
+ "epoch": 1.12,
2301
+ "learning_rate": 8.821407095610344e-05,
2302
+ "loss": 0.0099,
2303
+ "step": 3720
2304
+ },
2305
+ {
2306
+ "epoch": 1.12,
2307
+ "learning_rate": 8.791340950090199e-05,
2308
+ "loss": 0.0741,
2309
+ "step": 3730
2310
+ },
2311
+ {
2312
+ "epoch": 1.12,
2313
+ "learning_rate": 8.761274804570054e-05,
2314
+ "loss": 0.0255,
2315
+ "step": 3740
2316
+ },
2317
+ {
2318
+ "epoch": 1.13,
2319
+ "learning_rate": 8.73120865904991e-05,
2320
+ "loss": 0.0229,
2321
+ "step": 3750
2322
+ },
2323
+ {
2324
+ "epoch": 1.13,
2325
+ "learning_rate": 8.701142513529766e-05,
2326
+ "loss": 0.0338,
2327
+ "step": 3760
2328
+ },
2329
+ {
2330
+ "epoch": 1.13,
2331
+ "learning_rate": 8.671076368009621e-05,
2332
+ "loss": 0.0198,
2333
+ "step": 3770
2334
+ },
2335
+ {
2336
+ "epoch": 1.14,
2337
+ "learning_rate": 8.641010222489478e-05,
2338
+ "loss": 0.0248,
2339
+ "step": 3780
2340
+ },
2341
+ {
2342
+ "epoch": 1.14,
2343
+ "learning_rate": 8.610944076969333e-05,
2344
+ "loss": 0.1184,
2345
+ "step": 3790
2346
+ },
2347
+ {
2348
+ "epoch": 1.14,
2349
+ "learning_rate": 8.580877931449189e-05,
2350
+ "loss": 0.0966,
2351
+ "step": 3800
2352
+ },
2353
+ {
2354
+ "epoch": 1.15,
2355
+ "learning_rate": 8.550811785929045e-05,
2356
+ "loss": 0.0148,
2357
+ "step": 3810
2358
+ },
2359
+ {
2360
+ "epoch": 1.15,
2361
+ "learning_rate": 8.5207456404089e-05,
2362
+ "loss": 0.0428,
2363
+ "step": 3820
2364
+ },
2365
+ {
2366
+ "epoch": 1.15,
2367
+ "learning_rate": 8.490679494888756e-05,
2368
+ "loss": 0.0767,
2369
+ "step": 3830
2370
+ },
2371
+ {
2372
+ "epoch": 1.15,
2373
+ "learning_rate": 8.460613349368611e-05,
2374
+ "loss": 0.0297,
2375
+ "step": 3840
2376
+ },
2377
+ {
2378
+ "epoch": 1.16,
2379
+ "learning_rate": 8.430547203848466e-05,
2380
+ "loss": 0.0047,
2381
+ "step": 3850
2382
+ },
2383
+ {
2384
+ "epoch": 1.16,
2385
+ "learning_rate": 8.400481058328322e-05,
2386
+ "loss": 0.007,
2387
+ "step": 3860
2388
+ },
2389
+ {
2390
+ "epoch": 1.16,
2391
+ "learning_rate": 8.370414912808179e-05,
2392
+ "loss": 0.0556,
2393
+ "step": 3870
2394
+ },
2395
+ {
2396
+ "epoch": 1.17,
2397
+ "learning_rate": 8.340348767288034e-05,
2398
+ "loss": 0.0145,
2399
+ "step": 3880
2400
+ },
2401
+ {
2402
+ "epoch": 1.17,
2403
+ "learning_rate": 8.31028262176789e-05,
2404
+ "loss": 0.0099,
2405
+ "step": 3890
2406
+ },
2407
+ {
2408
+ "epoch": 1.17,
2409
+ "learning_rate": 8.280216476247745e-05,
2410
+ "loss": 0.0049,
2411
+ "step": 3900
2412
+ },
2413
+ {
2414
+ "epoch": 1.18,
2415
+ "learning_rate": 8.250150330727601e-05,
2416
+ "loss": 0.0156,
2417
+ "step": 3910
2418
+ },
2419
+ {
2420
+ "epoch": 1.18,
2421
+ "learning_rate": 8.220084185207458e-05,
2422
+ "loss": 0.0362,
2423
+ "step": 3920
2424
+ },
2425
+ {
2426
+ "epoch": 1.18,
2427
+ "learning_rate": 8.190018039687312e-05,
2428
+ "loss": 0.0311,
2429
+ "step": 3930
2430
+ },
2431
+ {
2432
+ "epoch": 1.18,
2433
+ "learning_rate": 8.159951894167169e-05,
2434
+ "loss": 0.0385,
2435
+ "step": 3940
2436
+ },
2437
+ {
2438
+ "epoch": 1.19,
2439
+ "learning_rate": 8.129885748647024e-05,
2440
+ "loss": 0.006,
2441
+ "step": 3950
2442
+ },
2443
+ {
2444
+ "epoch": 1.19,
2445
+ "learning_rate": 8.099819603126879e-05,
2446
+ "loss": 0.0018,
2447
+ "step": 3960
2448
+ },
2449
+ {
2450
+ "epoch": 1.19,
2451
+ "learning_rate": 8.069753457606736e-05,
2452
+ "loss": 0.072,
2453
+ "step": 3970
2454
+ },
2455
+ {
2456
+ "epoch": 1.2,
2457
+ "learning_rate": 8.039687312086591e-05,
2458
+ "loss": 0.0156,
2459
+ "step": 3980
2460
+ },
2461
+ {
2462
+ "epoch": 1.2,
2463
+ "learning_rate": 8.009621166566446e-05,
2464
+ "loss": 0.0235,
2465
+ "step": 3990
2466
+ },
2467
+ {
2468
+ "epoch": 1.2,
2469
+ "learning_rate": 7.979555021046302e-05,
2470
+ "loss": 0.0285,
2471
+ "step": 4000
2472
+ },
2473
+ {
2474
+ "epoch": 1.2,
2475
+ "eval_accuracy": 0.9897031191281473,
2476
+ "eval_loss": 0.03260853514075279,
2477
+ "eval_runtime": 226.7177,
2478
+ "eval_samples_per_second": 58.685,
2479
+ "eval_steps_per_second": 7.34,
2480
+ "step": 4000
2481
+ },
2482
+ {
2483
+ "epoch": 1.21,
2484
+ "learning_rate": 7.949488875526157e-05,
2485
+ "loss": 0.0308,
2486
+ "step": 4010
2487
+ },
2488
+ {
2489
+ "epoch": 1.21,
2490
+ "learning_rate": 7.919422730006014e-05,
2491
+ "loss": 0.0135,
2492
+ "step": 4020
2493
+ },
2494
+ {
2495
+ "epoch": 1.21,
2496
+ "learning_rate": 7.88935658448587e-05,
2497
+ "loss": 0.0066,
2498
+ "step": 4030
2499
+ },
2500
+ {
2501
+ "epoch": 1.21,
2502
+ "learning_rate": 7.859290438965725e-05,
2503
+ "loss": 0.1052,
2504
+ "step": 4040
2505
+ },
2506
+ {
2507
+ "epoch": 1.22,
2508
+ "learning_rate": 7.829224293445581e-05,
2509
+ "loss": 0.0038,
2510
+ "step": 4050
2511
+ },
2512
+ {
2513
+ "epoch": 1.22,
2514
+ "learning_rate": 7.799158147925436e-05,
2515
+ "loss": 0.0271,
2516
+ "step": 4060
2517
+ },
2518
+ {
2519
+ "epoch": 1.22,
2520
+ "learning_rate": 7.769092002405291e-05,
2521
+ "loss": 0.0145,
2522
+ "step": 4070
2523
+ },
2524
+ {
2525
+ "epoch": 1.23,
2526
+ "learning_rate": 7.739025856885149e-05,
2527
+ "loss": 0.0388,
2528
+ "step": 4080
2529
+ },
2530
+ {
2531
+ "epoch": 1.23,
2532
+ "learning_rate": 7.708959711365004e-05,
2533
+ "loss": 0.0543,
2534
+ "step": 4090
2535
+ },
2536
+ {
2537
+ "epoch": 1.23,
2538
+ "learning_rate": 7.678893565844858e-05,
2539
+ "loss": 0.023,
2540
+ "step": 4100
2541
+ },
2542
+ {
2543
+ "epoch": 1.24,
2544
+ "learning_rate": 7.648827420324715e-05,
2545
+ "loss": 0.0641,
2546
+ "step": 4110
2547
+ },
2548
+ {
2549
+ "epoch": 1.24,
2550
+ "learning_rate": 7.61876127480457e-05,
2551
+ "loss": 0.0448,
2552
+ "step": 4120
2553
+ },
2554
+ {
2555
+ "epoch": 1.24,
2556
+ "learning_rate": 7.588695129284426e-05,
2557
+ "loss": 0.0759,
2558
+ "step": 4130
2559
+ },
2560
+ {
2561
+ "epoch": 1.24,
2562
+ "learning_rate": 7.558628983764282e-05,
2563
+ "loss": 0.0183,
2564
+ "step": 4140
2565
+ },
2566
+ {
2567
+ "epoch": 1.25,
2568
+ "learning_rate": 7.528562838244137e-05,
2569
+ "loss": 0.0698,
2570
+ "step": 4150
2571
+ },
2572
+ {
2573
+ "epoch": 1.25,
2574
+ "learning_rate": 7.498496692723994e-05,
2575
+ "loss": 0.0269,
2576
+ "step": 4160
2577
+ },
2578
+ {
2579
+ "epoch": 1.25,
2580
+ "learning_rate": 7.468430547203848e-05,
2581
+ "loss": 0.0331,
2582
+ "step": 4170
2583
+ },
2584
+ {
2585
+ "epoch": 1.26,
2586
+ "learning_rate": 7.438364401683703e-05,
2587
+ "loss": 0.0064,
2588
+ "step": 4180
2589
+ },
2590
+ {
2591
+ "epoch": 1.26,
2592
+ "learning_rate": 7.408298256163561e-05,
2593
+ "loss": 0.0828,
2594
+ "step": 4190
2595
+ },
2596
+ {
2597
+ "epoch": 1.26,
2598
+ "learning_rate": 7.378232110643416e-05,
2599
+ "loss": 0.0281,
2600
+ "step": 4200
2601
+ },
2602
+ {
2603
+ "epoch": 1.27,
2604
+ "learning_rate": 7.348165965123271e-05,
2605
+ "loss": 0.0164,
2606
+ "step": 4210
2607
+ },
2608
+ {
2609
+ "epoch": 1.27,
2610
+ "learning_rate": 7.318099819603127e-05,
2611
+ "loss": 0.0737,
2612
+ "step": 4220
2613
+ },
2614
+ {
2615
+ "epoch": 1.27,
2616
+ "learning_rate": 7.288033674082982e-05,
2617
+ "loss": 0.0104,
2618
+ "step": 4230
2619
+ },
2620
+ {
2621
+ "epoch": 1.27,
2622
+ "learning_rate": 7.257967528562838e-05,
2623
+ "loss": 0.0316,
2624
+ "step": 4240
2625
+ },
2626
+ {
2627
+ "epoch": 1.28,
2628
+ "learning_rate": 7.227901383042695e-05,
2629
+ "loss": 0.0746,
2630
+ "step": 4250
2631
+ },
2632
+ {
2633
+ "epoch": 1.28,
2634
+ "learning_rate": 7.19783523752255e-05,
2635
+ "loss": 0.0268,
2636
+ "step": 4260
2637
+ },
2638
+ {
2639
+ "epoch": 1.28,
2640
+ "learning_rate": 7.167769092002406e-05,
2641
+ "loss": 0.0365,
2642
+ "step": 4270
2643
+ },
2644
+ {
2645
+ "epoch": 1.29,
2646
+ "learning_rate": 7.137702946482261e-05,
2647
+ "loss": 0.0206,
2648
+ "step": 4280
2649
+ },
2650
+ {
2651
+ "epoch": 1.29,
2652
+ "learning_rate": 7.107636800962117e-05,
2653
+ "loss": 0.0058,
2654
+ "step": 4290
2655
+ },
2656
+ {
2657
+ "epoch": 1.29,
2658
+ "learning_rate": 7.077570655441973e-05,
2659
+ "loss": 0.0437,
2660
+ "step": 4300
2661
+ },
2662
+ {
2663
+ "epoch": 1.3,
2664
+ "learning_rate": 7.047504509921828e-05,
2665
+ "loss": 0.028,
2666
+ "step": 4310
2667
+ },
2668
+ {
2669
+ "epoch": 1.3,
2670
+ "learning_rate": 7.017438364401683e-05,
2671
+ "loss": 0.0416,
2672
+ "step": 4320
2673
+ },
2674
+ {
2675
+ "epoch": 1.3,
2676
+ "learning_rate": 6.98737221888154e-05,
2677
+ "loss": 0.0031,
2678
+ "step": 4330
2679
+ },
2680
+ {
2681
+ "epoch": 1.3,
2682
+ "learning_rate": 6.957306073361396e-05,
2683
+ "loss": 0.0247,
2684
+ "step": 4340
2685
+ },
2686
+ {
2687
+ "epoch": 1.31,
2688
+ "learning_rate": 6.927239927841251e-05,
2689
+ "loss": 0.0469,
2690
+ "step": 4350
2691
+ },
2692
+ {
2693
+ "epoch": 1.31,
2694
+ "learning_rate": 6.897173782321107e-05,
2695
+ "loss": 0.0241,
2696
+ "step": 4360
2697
+ },
2698
+ {
2699
+ "epoch": 1.31,
2700
+ "learning_rate": 6.867107636800962e-05,
2701
+ "loss": 0.0156,
2702
+ "step": 4370
2703
+ },
2704
+ {
2705
+ "epoch": 1.32,
2706
+ "learning_rate": 6.837041491280818e-05,
2707
+ "loss": 0.0247,
2708
+ "step": 4380
2709
+ },
2710
+ {
2711
+ "epoch": 1.32,
2712
+ "learning_rate": 6.806975345760673e-05,
2713
+ "loss": 0.0068,
2714
+ "step": 4390
2715
+ },
2716
+ {
2717
+ "epoch": 1.32,
2718
+ "learning_rate": 6.77690920024053e-05,
2719
+ "loss": 0.0057,
2720
+ "step": 4400
2721
+ },
2722
+ {
2723
+ "epoch": 1.33,
2724
+ "learning_rate": 6.746843054720386e-05,
2725
+ "loss": 0.0007,
2726
+ "step": 4410
2727
+ },
2728
+ {
2729
+ "epoch": 1.33,
2730
+ "learning_rate": 6.716776909200241e-05,
2731
+ "loss": 0.0187,
2732
+ "step": 4420
2733
+ },
2734
+ {
2735
+ "epoch": 1.33,
2736
+ "learning_rate": 6.686710763680096e-05,
2737
+ "loss": 0.0144,
2738
+ "step": 4430
2739
+ },
2740
+ {
2741
+ "epoch": 1.33,
2742
+ "learning_rate": 6.656644618159952e-05,
2743
+ "loss": 0.1132,
2744
+ "step": 4440
2745
+ },
2746
+ {
2747
+ "epoch": 1.34,
2748
+ "learning_rate": 6.626578472639808e-05,
2749
+ "loss": 0.0126,
2750
+ "step": 4450
2751
+ },
2752
+ {
2753
+ "epoch": 1.34,
2754
+ "learning_rate": 6.596512327119663e-05,
2755
+ "loss": 0.007,
2756
+ "step": 4460
2757
+ },
2758
+ {
2759
+ "epoch": 1.34,
2760
+ "learning_rate": 6.56644618159952e-05,
2761
+ "loss": 0.0182,
2762
+ "step": 4470
2763
+ },
2764
+ {
2765
+ "epoch": 1.35,
2766
+ "learning_rate": 6.536380036079374e-05,
2767
+ "loss": 0.0348,
2768
+ "step": 4480
2769
+ },
2770
+ {
2771
+ "epoch": 1.35,
2772
+ "learning_rate": 6.506313890559231e-05,
2773
+ "loss": 0.0837,
2774
+ "step": 4490
2775
+ },
2776
+ {
2777
+ "epoch": 1.35,
2778
+ "learning_rate": 6.476247745039087e-05,
2779
+ "loss": 0.0442,
2780
+ "step": 4500
2781
+ },
2782
+ {
2783
+ "epoch": 1.35,
2784
+ "eval_accuracy": 0.9909808342728298,
2785
+ "eval_loss": 0.023428408429026604,
2786
+ "eval_runtime": 224.2844,
2787
+ "eval_samples_per_second": 59.322,
2788
+ "eval_steps_per_second": 7.419,
2789
+ "step": 4500
2790
+ },
2791
+ {
2792
+ "epoch": 1.36,
2793
+ "learning_rate": 6.446181599518942e-05,
2794
+ "loss": 0.0075,
2795
+ "step": 4510
2796
+ },
2797
+ {
2798
+ "epoch": 1.36,
2799
+ "learning_rate": 6.416115453998798e-05,
2800
+ "loss": 0.0393,
2801
+ "step": 4520
2802
+ },
2803
+ {
2804
+ "epoch": 1.36,
2805
+ "learning_rate": 6.386049308478653e-05,
2806
+ "loss": 0.0192,
2807
+ "step": 4530
2808
+ },
2809
+ {
2810
+ "epoch": 1.37,
2811
+ "learning_rate": 6.355983162958508e-05,
2812
+ "loss": 0.0212,
2813
+ "step": 4540
2814
+ },
2815
+ {
2816
+ "epoch": 1.37,
2817
+ "learning_rate": 6.325917017438364e-05,
2818
+ "loss": 0.011,
2819
+ "step": 4550
2820
+ },
2821
+ {
2822
+ "epoch": 1.37,
2823
+ "learning_rate": 6.29585087191822e-05,
2824
+ "loss": 0.0055,
2825
+ "step": 4560
2826
+ },
2827
+ {
2828
+ "epoch": 1.37,
2829
+ "learning_rate": 6.265784726398076e-05,
2830
+ "loss": 0.0562,
2831
+ "step": 4570
2832
+ },
2833
+ {
2834
+ "epoch": 1.38,
2835
+ "learning_rate": 6.235718580877932e-05,
2836
+ "loss": 0.0858,
2837
+ "step": 4580
2838
+ },
2839
+ {
2840
+ "epoch": 1.38,
2841
+ "learning_rate": 6.205652435357787e-05,
2842
+ "loss": 0.0636,
2843
+ "step": 4590
2844
+ },
2845
+ {
2846
+ "epoch": 1.38,
2847
+ "learning_rate": 6.175586289837643e-05,
2848
+ "loss": 0.0654,
2849
+ "step": 4600
2850
+ },
2851
+ {
2852
+ "epoch": 1.39,
2853
+ "learning_rate": 6.1455201443175e-05,
2854
+ "loss": 0.0245,
2855
+ "step": 4610
2856
+ },
2857
+ {
2858
+ "epoch": 1.39,
2859
+ "learning_rate": 6.115453998797354e-05,
2860
+ "loss": 0.0175,
2861
+ "step": 4620
2862
+ },
2863
+ {
2864
+ "epoch": 1.39,
2865
+ "learning_rate": 6.08538785327721e-05,
2866
+ "loss": 0.0166,
2867
+ "step": 4630
2868
+ },
2869
+ {
2870
+ "epoch": 1.4,
2871
+ "learning_rate": 6.0553217077570655e-05,
2872
+ "loss": 0.0312,
2873
+ "step": 4640
2874
+ },
2875
+ {
2876
+ "epoch": 1.4,
2877
+ "learning_rate": 6.025255562236921e-05,
2878
+ "loss": 0.0217,
2879
+ "step": 4650
2880
+ },
2881
+ {
2882
+ "epoch": 1.4,
2883
+ "learning_rate": 5.9951894167167774e-05,
2884
+ "loss": 0.0144,
2885
+ "step": 4660
2886
+ },
2887
+ {
2888
+ "epoch": 1.4,
2889
+ "learning_rate": 5.965123271196633e-05,
2890
+ "loss": 0.0087,
2891
+ "step": 4670
2892
+ },
2893
+ {
2894
+ "epoch": 1.41,
2895
+ "learning_rate": 5.9350571256764887e-05,
2896
+ "loss": 0.024,
2897
+ "step": 4680
2898
+ },
2899
+ {
2900
+ "epoch": 1.41,
2901
+ "learning_rate": 5.904990980156344e-05,
2902
+ "loss": 0.061,
2903
+ "step": 4690
2904
+ },
2905
+ {
2906
+ "epoch": 1.41,
2907
+ "learning_rate": 5.8749248346362e-05,
2908
+ "loss": 0.0207,
2909
+ "step": 4700
2910
+ },
2911
+ {
2912
+ "epoch": 1.42,
2913
+ "learning_rate": 5.844858689116055e-05,
2914
+ "loss": 0.0739,
2915
+ "step": 4710
2916
+ },
2917
+ {
2918
+ "epoch": 1.42,
2919
+ "learning_rate": 5.814792543595912e-05,
2920
+ "loss": 0.0075,
2921
+ "step": 4720
2922
+ },
2923
+ {
2924
+ "epoch": 1.42,
2925
+ "learning_rate": 5.7847263980757674e-05,
2926
+ "loss": 0.0685,
2927
+ "step": 4730
2928
+ },
2929
+ {
2930
+ "epoch": 1.43,
2931
+ "learning_rate": 5.754660252555622e-05,
2932
+ "loss": 0.0493,
2933
+ "step": 4740
2934
+ },
2935
+ {
2936
+ "epoch": 1.43,
2937
+ "learning_rate": 5.724594107035478e-05,
2938
+ "loss": 0.0057,
2939
+ "step": 4750
2940
+ },
2941
+ {
2942
+ "epoch": 1.43,
2943
+ "learning_rate": 5.6945279615153335e-05,
2944
+ "loss": 0.098,
2945
+ "step": 4760
2946
+ },
2947
+ {
2948
+ "epoch": 1.43,
2949
+ "learning_rate": 5.66446181599519e-05,
2950
+ "loss": 0.0055,
2951
+ "step": 4770
2952
+ },
2953
+ {
2954
+ "epoch": 1.44,
2955
+ "learning_rate": 5.6343956704750454e-05,
2956
+ "loss": 0.0472,
2957
+ "step": 4780
2958
+ },
2959
+ {
2960
+ "epoch": 1.44,
2961
+ "learning_rate": 5.604329524954901e-05,
2962
+ "loss": 0.004,
2963
+ "step": 4790
2964
+ },
2965
+ {
2966
+ "epoch": 1.44,
2967
+ "learning_rate": 5.5742633794347567e-05,
2968
+ "loss": 0.0552,
2969
+ "step": 4800
2970
+ },
2971
+ {
2972
+ "epoch": 1.45,
2973
+ "learning_rate": 5.544197233914612e-05,
2974
+ "loss": 0.0226,
2975
+ "step": 4810
2976
+ },
2977
+ {
2978
+ "epoch": 1.45,
2979
+ "learning_rate": 5.5141310883944686e-05,
2980
+ "loss": 0.0059,
2981
+ "step": 4820
2982
+ },
2983
+ {
2984
+ "epoch": 1.45,
2985
+ "learning_rate": 5.484064942874324e-05,
2986
+ "loss": 0.0065,
2987
+ "step": 4830
2988
+ },
2989
+ {
2990
+ "epoch": 1.46,
2991
+ "learning_rate": 5.45399879735418e-05,
2992
+ "loss": 0.0042,
2993
+ "step": 4840
2994
+ },
2995
+ {
2996
+ "epoch": 1.46,
2997
+ "learning_rate": 5.423932651834035e-05,
2998
+ "loss": 0.0355,
2999
+ "step": 4850
3000
+ },
3001
+ {
3002
+ "epoch": 1.46,
3003
+ "learning_rate": 5.39386650631389e-05,
3004
+ "loss": 0.0055,
3005
+ "step": 4860
3006
+ },
3007
+ {
3008
+ "epoch": 1.46,
3009
+ "learning_rate": 5.363800360793747e-05,
3010
+ "loss": 0.0255,
3011
+ "step": 4870
3012
+ },
3013
+ {
3014
+ "epoch": 1.47,
3015
+ "learning_rate": 5.333734215273602e-05,
3016
+ "loss": 0.0044,
3017
+ "step": 4880
3018
+ },
3019
+ {
3020
+ "epoch": 1.47,
3021
+ "learning_rate": 5.303668069753458e-05,
3022
+ "loss": 0.0683,
3023
+ "step": 4890
3024
+ },
3025
+ {
3026
+ "epoch": 1.47,
3027
+ "learning_rate": 5.2736019242333134e-05,
3028
+ "loss": 0.0234,
3029
+ "step": 4900
3030
+ },
3031
+ {
3032
+ "epoch": 1.48,
3033
+ "learning_rate": 5.243535778713169e-05,
3034
+ "loss": 0.0229,
3035
+ "step": 4910
3036
+ },
3037
+ {
3038
+ "epoch": 1.48,
3039
+ "learning_rate": 5.2134696331930247e-05,
3040
+ "loss": 0.0118,
3041
+ "step": 4920
3042
+ },
3043
+ {
3044
+ "epoch": 1.48,
3045
+ "learning_rate": 5.183403487672881e-05,
3046
+ "loss": 0.0622,
3047
+ "step": 4930
3048
+ },
3049
+ {
3050
+ "epoch": 1.49,
3051
+ "learning_rate": 5.1533373421527366e-05,
3052
+ "loss": 0.0171,
3053
+ "step": 4940
3054
+ },
3055
+ {
3056
+ "epoch": 1.49,
3057
+ "learning_rate": 5.123271196632592e-05,
3058
+ "loss": 0.0118,
3059
+ "step": 4950
3060
+ },
3061
+ {
3062
+ "epoch": 1.49,
3063
+ "learning_rate": 5.093205051112447e-05,
3064
+ "loss": 0.0483,
3065
+ "step": 4960
3066
+ },
3067
+ {
3068
+ "epoch": 1.49,
3069
+ "learning_rate": 5.063138905592303e-05,
3070
+ "loss": 0.0323,
3071
+ "step": 4970
3072
+ },
3073
+ {
3074
+ "epoch": 1.5,
3075
+ "learning_rate": 5.03307276007216e-05,
3076
+ "loss": 0.0124,
3077
+ "step": 4980
3078
+ },
3079
+ {
3080
+ "epoch": 1.5,
3081
+ "learning_rate": 5.0030066145520146e-05,
3082
+ "loss": 0.025,
3083
+ "step": 4990
3084
+ },
3085
+ {
3086
+ "epoch": 1.5,
3087
+ "learning_rate": 4.97294046903187e-05,
3088
+ "loss": 0.016,
3089
+ "step": 5000
3090
+ },
3091
+ {
3092
+ "epoch": 1.5,
3093
+ "eval_accuracy": 0.9922585494175122,
3094
+ "eval_loss": 0.02205909602344036,
3095
+ "eval_runtime": 226.2583,
3096
+ "eval_samples_per_second": 58.804,
3097
+ "eval_steps_per_second": 7.354,
3098
+ "step": 5000
3099
+ },
3100
+ {
3101
+ "epoch": 1.51,
3102
+ "learning_rate": 4.942874323511726e-05,
3103
+ "loss": 0.019,
3104
+ "step": 5010
3105
+ },
3106
+ {
3107
+ "epoch": 1.51,
3108
+ "learning_rate": 4.912808177991582e-05,
3109
+ "loss": 0.0685,
3110
+ "step": 5020
3111
+ },
3112
+ {
3113
+ "epoch": 1.51,
3114
+ "learning_rate": 4.882742032471437e-05,
3115
+ "loss": 0.0548,
3116
+ "step": 5030
3117
+ },
3118
+ {
3119
+ "epoch": 1.52,
3120
+ "learning_rate": 4.852675886951293e-05,
3121
+ "loss": 0.0358,
3122
+ "step": 5040
3123
+ },
3124
+ {
3125
+ "epoch": 1.52,
3126
+ "learning_rate": 4.822609741431149e-05,
3127
+ "loss": 0.0054,
3128
+ "step": 5050
3129
+ },
3130
+ {
3131
+ "epoch": 1.52,
3132
+ "learning_rate": 4.7925435959110046e-05,
3133
+ "loss": 0.0157,
3134
+ "step": 5060
3135
+ },
3136
+ {
3137
+ "epoch": 1.52,
3138
+ "learning_rate": 4.76247745039086e-05,
3139
+ "loss": 0.0317,
3140
+ "step": 5070
3141
+ },
3142
+ {
3143
+ "epoch": 1.53,
3144
+ "learning_rate": 4.732411304870716e-05,
3145
+ "loss": 0.0091,
3146
+ "step": 5080
3147
+ },
3148
+ {
3149
+ "epoch": 1.53,
3150
+ "learning_rate": 4.7023451593505714e-05,
3151
+ "loss": 0.0041,
3152
+ "step": 5090
3153
+ },
3154
+ {
3155
+ "epoch": 1.53,
3156
+ "learning_rate": 4.672279013830427e-05,
3157
+ "loss": 0.001,
3158
+ "step": 5100
3159
+ },
3160
+ {
3161
+ "epoch": 1.54,
3162
+ "learning_rate": 4.6422128683102826e-05,
3163
+ "loss": 0.0246,
3164
+ "step": 5110
3165
+ },
3166
+ {
3167
+ "epoch": 1.54,
3168
+ "learning_rate": 4.612146722790138e-05,
3169
+ "loss": 0.0062,
3170
+ "step": 5120
3171
+ },
3172
+ {
3173
+ "epoch": 1.54,
3174
+ "learning_rate": 4.5820805772699945e-05,
3175
+ "loss": 0.0496,
3176
+ "step": 5130
3177
+ },
3178
+ {
3179
+ "epoch": 1.55,
3180
+ "learning_rate": 4.5520144317498494e-05,
3181
+ "loss": 0.0229,
3182
+ "step": 5140
3183
+ },
3184
+ {
3185
+ "epoch": 1.55,
3186
+ "learning_rate": 4.521948286229706e-05,
3187
+ "loss": 0.0235,
3188
+ "step": 5150
3189
+ },
3190
+ {
3191
+ "epoch": 1.55,
3192
+ "learning_rate": 4.4918821407095613e-05,
3193
+ "loss": 0.0077,
3194
+ "step": 5160
3195
+ },
3196
+ {
3197
+ "epoch": 1.55,
3198
+ "learning_rate": 4.461815995189417e-05,
3199
+ "loss": 0.0125,
3200
+ "step": 5170
3201
+ },
3202
+ {
3203
+ "epoch": 1.56,
3204
+ "learning_rate": 4.4317498496692726e-05,
3205
+ "loss": 0.006,
3206
+ "step": 5180
3207
+ },
3208
+ {
3209
+ "epoch": 1.56,
3210
+ "learning_rate": 4.401683704149128e-05,
3211
+ "loss": 0.0253,
3212
+ "step": 5190
3213
+ },
3214
+ {
3215
+ "epoch": 1.56,
3216
+ "learning_rate": 4.371617558628984e-05,
3217
+ "loss": 0.0435,
3218
+ "step": 5200
3219
+ },
3220
+ {
3221
+ "epoch": 1.57,
3222
+ "learning_rate": 4.3415514131088394e-05,
3223
+ "loss": 0.0053,
3224
+ "step": 5210
3225
+ },
3226
+ {
3227
+ "epoch": 1.57,
3228
+ "learning_rate": 4.311485267588695e-05,
3229
+ "loss": 0.0028,
3230
+ "step": 5220
3231
+ },
3232
+ {
3233
+ "epoch": 1.57,
3234
+ "learning_rate": 4.281419122068551e-05,
3235
+ "loss": 0.0132,
3236
+ "step": 5230
3237
+ },
3238
+ {
3239
+ "epoch": 1.58,
3240
+ "learning_rate": 4.251352976548407e-05,
3241
+ "loss": 0.02,
3242
+ "step": 5240
3243
+ },
3244
+ {
3245
+ "epoch": 1.58,
3246
+ "learning_rate": 4.221286831028262e-05,
3247
+ "loss": 0.0114,
3248
+ "step": 5250
3249
+ },
3250
+ {
3251
+ "epoch": 1.58,
3252
+ "learning_rate": 4.191220685508118e-05,
3253
+ "loss": 0.0315,
3254
+ "step": 5260
3255
+ },
3256
+ {
3257
+ "epoch": 1.58,
3258
+ "learning_rate": 4.161154539987974e-05,
3259
+ "loss": 0.0188,
3260
+ "step": 5270
3261
+ },
3262
+ {
3263
+ "epoch": 1.59,
3264
+ "learning_rate": 4.13108839446783e-05,
3265
+ "loss": 0.016,
3266
+ "step": 5280
3267
+ },
3268
+ {
3269
+ "epoch": 1.59,
3270
+ "learning_rate": 4.101022248947685e-05,
3271
+ "loss": 0.0719,
3272
+ "step": 5290
3273
+ },
3274
+ {
3275
+ "epoch": 1.59,
3276
+ "learning_rate": 4.0709561034275406e-05,
3277
+ "loss": 0.0157,
3278
+ "step": 5300
3279
+ },
3280
+ {
3281
+ "epoch": 1.6,
3282
+ "learning_rate": 4.040889957907397e-05,
3283
+ "loss": 0.0004,
3284
+ "step": 5310
3285
+ },
3286
+ {
3287
+ "epoch": 1.6,
3288
+ "learning_rate": 4.0108238123872525e-05,
3289
+ "loss": 0.022,
3290
+ "step": 5320
3291
+ },
3292
+ {
3293
+ "epoch": 1.6,
3294
+ "learning_rate": 3.9807576668671074e-05,
3295
+ "loss": 0.0267,
3296
+ "step": 5330
3297
+ },
3298
+ {
3299
+ "epoch": 1.61,
3300
+ "learning_rate": 3.950691521346964e-05,
3301
+ "loss": 0.0343,
3302
+ "step": 5340
3303
+ },
3304
+ {
3305
+ "epoch": 1.61,
3306
+ "learning_rate": 3.920625375826819e-05,
3307
+ "loss": 0.0118,
3308
+ "step": 5350
3309
+ },
3310
+ {
3311
+ "epoch": 1.61,
3312
+ "learning_rate": 3.890559230306675e-05,
3313
+ "loss": 0.0226,
3314
+ "step": 5360
3315
+ },
3316
+ {
3317
+ "epoch": 1.61,
3318
+ "learning_rate": 3.8604930847865305e-05,
3319
+ "loss": 0.0244,
3320
+ "step": 5370
3321
+ },
3322
+ {
3323
+ "epoch": 1.62,
3324
+ "learning_rate": 3.830426939266386e-05,
3325
+ "loss": 0.0394,
3326
+ "step": 5380
3327
+ },
3328
+ {
3329
+ "epoch": 1.62,
3330
+ "learning_rate": 3.8003607937462424e-05,
3331
+ "loss": 0.0154,
3332
+ "step": 5390
3333
+ },
3334
+ {
3335
+ "epoch": 1.62,
3336
+ "learning_rate": 3.7702946482260974e-05,
3337
+ "loss": 0.0653,
3338
+ "step": 5400
3339
+ },
3340
+ {
3341
+ "epoch": 1.63,
3342
+ "learning_rate": 3.740228502705953e-05,
3343
+ "loss": 0.0153,
3344
+ "step": 5410
3345
+ },
3346
+ {
3347
+ "epoch": 1.63,
3348
+ "learning_rate": 3.710162357185809e-05,
3349
+ "loss": 0.0089,
3350
+ "step": 5420
3351
+ },
3352
+ {
3353
+ "epoch": 1.63,
3354
+ "learning_rate": 3.680096211665665e-05,
3355
+ "loss": 0.0137,
3356
+ "step": 5430
3357
+ },
3358
+ {
3359
+ "epoch": 1.64,
3360
+ "learning_rate": 3.6500300661455205e-05,
3361
+ "loss": 0.0069,
3362
+ "step": 5440
3363
+ },
3364
+ {
3365
+ "epoch": 1.64,
3366
+ "learning_rate": 3.619963920625376e-05,
3367
+ "loss": 0.0007,
3368
+ "step": 5450
3369
+ },
3370
+ {
3371
+ "epoch": 1.64,
3372
+ "learning_rate": 3.589897775105232e-05,
3373
+ "loss": 0.0034,
3374
+ "step": 5460
3375
+ },
3376
+ {
3377
+ "epoch": 1.64,
3378
+ "learning_rate": 3.559831629585087e-05,
3379
+ "loss": 0.0034,
3380
+ "step": 5470
3381
+ },
3382
+ {
3383
+ "epoch": 1.65,
3384
+ "learning_rate": 3.529765484064943e-05,
3385
+ "loss": 0.0474,
3386
+ "step": 5480
3387
+ },
3388
+ {
3389
+ "epoch": 1.65,
3390
+ "learning_rate": 3.4996993385447985e-05,
3391
+ "loss": 0.007,
3392
+ "step": 5490
3393
+ },
3394
+ {
3395
+ "epoch": 1.65,
3396
+ "learning_rate": 3.469633193024655e-05,
3397
+ "loss": 0.0376,
3398
+ "step": 5500
3399
+ },
3400
+ {
3401
+ "epoch": 1.65,
3402
+ "eval_accuracy": 0.9950394588500564,
3403
+ "eval_loss": 0.017912130802869797,
3404
+ "eval_runtime": 224.7117,
3405
+ "eval_samples_per_second": 59.209,
3406
+ "eval_steps_per_second": 7.405,
3407
+ "step": 5500
3408
+ },
3409
+ {
3410
+ "epoch": 1.66,
3411
+ "learning_rate": 3.43956704750451e-05,
3412
+ "loss": 0.0104,
3413
+ "step": 5510
3414
+ },
3415
+ {
3416
+ "epoch": 1.66,
3417
+ "learning_rate": 3.409500901984366e-05,
3418
+ "loss": 0.0081,
3419
+ "step": 5520
3420
+ },
3421
+ {
3422
+ "epoch": 1.66,
3423
+ "learning_rate": 3.3794347564642216e-05,
3424
+ "loss": 0.0007,
3425
+ "step": 5530
3426
+ },
3427
+ {
3428
+ "epoch": 1.67,
3429
+ "learning_rate": 3.349368610944077e-05,
3430
+ "loss": 0.0049,
3431
+ "step": 5540
3432
+ },
3433
+ {
3434
+ "epoch": 1.67,
3435
+ "learning_rate": 3.319302465423933e-05,
3436
+ "loss": 0.0257,
3437
+ "step": 5550
3438
+ },
3439
+ {
3440
+ "epoch": 1.67,
3441
+ "learning_rate": 3.2892363199037885e-05,
3442
+ "loss": 0.0142,
3443
+ "step": 5560
3444
+ },
3445
+ {
3446
+ "epoch": 1.67,
3447
+ "learning_rate": 3.259170174383644e-05,
3448
+ "loss": 0.0159,
3449
+ "step": 5570
3450
+ },
3451
+ {
3452
+ "epoch": 1.68,
3453
+ "learning_rate": 3.2291040288635e-05,
3454
+ "loss": 0.1096,
3455
+ "step": 5580
3456
+ },
3457
+ {
3458
+ "epoch": 1.68,
3459
+ "learning_rate": 3.199037883343355e-05,
3460
+ "loss": 0.0051,
3461
+ "step": 5590
3462
+ },
3463
+ {
3464
+ "epoch": 1.68,
3465
+ "learning_rate": 3.1689717378232116e-05,
3466
+ "loss": 0.0038,
3467
+ "step": 5600
3468
+ },
3469
+ {
3470
+ "epoch": 1.69,
3471
+ "learning_rate": 3.138905592303067e-05,
3472
+ "loss": 0.0428,
3473
+ "step": 5610
3474
+ },
3475
+ {
3476
+ "epoch": 1.69,
3477
+ "learning_rate": 3.108839446782922e-05,
3478
+ "loss": 0.0005,
3479
+ "step": 5620
3480
+ },
3481
+ {
3482
+ "epoch": 1.69,
3483
+ "learning_rate": 3.0787733012627784e-05,
3484
+ "loss": 0.0226,
3485
+ "step": 5630
3486
+ },
3487
+ {
3488
+ "epoch": 1.7,
3489
+ "learning_rate": 3.048707155742634e-05,
3490
+ "loss": 0.0415,
3491
+ "step": 5640
3492
+ },
3493
+ {
3494
+ "epoch": 1.7,
3495
+ "learning_rate": 3.0186410102224893e-05,
3496
+ "loss": 0.0227,
3497
+ "step": 5650
3498
+ },
3499
+ {
3500
+ "epoch": 1.7,
3501
+ "learning_rate": 2.9885748647023453e-05,
3502
+ "loss": 0.0346,
3503
+ "step": 5660
3504
+ },
3505
+ {
3506
+ "epoch": 1.7,
3507
+ "learning_rate": 2.958508719182201e-05,
3508
+ "loss": 0.0366,
3509
+ "step": 5670
3510
+ },
3511
+ {
3512
+ "epoch": 1.71,
3513
+ "learning_rate": 2.9284425736620568e-05,
3514
+ "loss": 0.0077,
3515
+ "step": 5680
3516
+ },
3517
+ {
3518
+ "epoch": 1.71,
3519
+ "learning_rate": 2.8983764281419124e-05,
3520
+ "loss": 0.02,
3521
+ "step": 5690
3522
+ },
3523
+ {
3524
+ "epoch": 1.71,
3525
+ "learning_rate": 2.8683102826217677e-05,
3526
+ "loss": 0.0016,
3527
+ "step": 5700
3528
+ },
3529
+ {
3530
+ "epoch": 1.72,
3531
+ "learning_rate": 2.838244137101624e-05,
3532
+ "loss": 0.0061,
3533
+ "step": 5710
3534
+ },
3535
+ {
3536
+ "epoch": 1.72,
3537
+ "learning_rate": 2.8081779915814793e-05,
3538
+ "loss": 0.008,
3539
+ "step": 5720
3540
+ },
3541
+ {
3542
+ "epoch": 1.72,
3543
+ "learning_rate": 2.7781118460613352e-05,
3544
+ "loss": 0.0003,
3545
+ "step": 5730
3546
+ },
3547
+ {
3548
+ "epoch": 1.73,
3549
+ "learning_rate": 2.7480457005411908e-05,
3550
+ "loss": 0.0074,
3551
+ "step": 5740
3552
+ },
3553
+ {
3554
+ "epoch": 1.73,
3555
+ "learning_rate": 2.7179795550210464e-05,
3556
+ "loss": 0.0419,
3557
+ "step": 5750
3558
+ },
3559
+ {
3560
+ "epoch": 1.73,
3561
+ "learning_rate": 2.6879134095009024e-05,
3562
+ "loss": 0.0249,
3563
+ "step": 5760
3564
+ },
3565
+ {
3566
+ "epoch": 1.73,
3567
+ "learning_rate": 2.6578472639807577e-05,
3568
+ "loss": 0.0086,
3569
+ "step": 5770
3570
+ },
3571
+ {
3572
+ "epoch": 1.74,
3573
+ "learning_rate": 2.6277811184606133e-05,
3574
+ "loss": 0.0325,
3575
+ "step": 5780
3576
+ },
3577
+ {
3578
+ "epoch": 1.74,
3579
+ "learning_rate": 2.5977149729404692e-05,
3580
+ "loss": 0.0105,
3581
+ "step": 5790
3582
+ },
3583
+ {
3584
+ "epoch": 1.74,
3585
+ "learning_rate": 2.5676488274203248e-05,
3586
+ "loss": 0.0061,
3587
+ "step": 5800
3588
+ },
3589
+ {
3590
+ "epoch": 1.75,
3591
+ "learning_rate": 2.5405892964521948e-05,
3592
+ "loss": 0.0387,
3593
+ "step": 5810
3594
+ },
3595
+ {
3596
+ "epoch": 1.75,
3597
+ "learning_rate": 2.5105231509320504e-05,
3598
+ "loss": 0.0297,
3599
+ "step": 5820
3600
+ },
3601
+ {
3602
+ "epoch": 1.75,
3603
+ "learning_rate": 2.4804570054119064e-05,
3604
+ "loss": 0.0209,
3605
+ "step": 5830
3606
+ },
3607
+ {
3608
+ "epoch": 1.76,
3609
+ "learning_rate": 2.450390859891762e-05,
3610
+ "loss": 0.0062,
3611
+ "step": 5840
3612
+ },
3613
+ {
3614
+ "epoch": 1.76,
3615
+ "learning_rate": 2.4203247143716176e-05,
3616
+ "loss": 0.0006,
3617
+ "step": 5850
3618
+ },
3619
+ {
3620
+ "epoch": 1.76,
3621
+ "learning_rate": 2.3902585688514732e-05,
3622
+ "loss": 0.0093,
3623
+ "step": 5860
3624
+ },
3625
+ {
3626
+ "epoch": 1.76,
3627
+ "learning_rate": 2.360192423331329e-05,
3628
+ "loss": 0.0055,
3629
+ "step": 5870
3630
+ },
3631
+ {
3632
+ "epoch": 1.77,
3633
+ "learning_rate": 2.3301262778111844e-05,
3634
+ "loss": 0.0248,
3635
+ "step": 5880
3636
+ },
3637
+ {
3638
+ "epoch": 1.77,
3639
+ "learning_rate": 2.3000601322910404e-05,
3640
+ "loss": 0.0646,
3641
+ "step": 5890
3642
+ },
3643
+ {
3644
+ "epoch": 1.77,
3645
+ "learning_rate": 2.269993986770896e-05,
3646
+ "loss": 0.0052,
3647
+ "step": 5900
3648
+ },
3649
+ {
3650
+ "epoch": 1.78,
3651
+ "learning_rate": 2.239927841250752e-05,
3652
+ "loss": 0.0421,
3653
+ "step": 5910
3654
+ },
3655
+ {
3656
+ "epoch": 1.78,
3657
+ "learning_rate": 2.2098616957306072e-05,
3658
+ "loss": 0.0104,
3659
+ "step": 5920
3660
+ },
3661
+ {
3662
+ "epoch": 1.78,
3663
+ "learning_rate": 2.179795550210463e-05,
3664
+ "loss": 0.009,
3665
+ "step": 5930
3666
+ },
3667
+ {
3668
+ "epoch": 1.79,
3669
+ "learning_rate": 2.1497294046903188e-05,
3670
+ "loss": 0.0137,
3671
+ "step": 5940
3672
+ },
3673
+ {
3674
+ "epoch": 1.79,
3675
+ "learning_rate": 2.1196632591701747e-05,
3676
+ "loss": 0.0026,
3677
+ "step": 5950
3678
+ },
3679
+ {
3680
+ "epoch": 1.79,
3681
+ "learning_rate": 2.0895971136500303e-05,
3682
+ "loss": 0.004,
3683
+ "step": 5960
3684
+ },
3685
+ {
3686
+ "epoch": 1.79,
3687
+ "learning_rate": 2.059530968129886e-05,
3688
+ "loss": 0.0098,
3689
+ "step": 5970
3690
+ },
3691
+ {
3692
+ "epoch": 1.8,
3693
+ "learning_rate": 2.0294648226097415e-05,
3694
+ "loss": 0.0028,
3695
+ "step": 5980
3696
+ },
3697
+ {
3698
+ "epoch": 1.8,
3699
+ "learning_rate": 1.999398677089597e-05,
3700
+ "loss": 0.0476,
3701
+ "step": 5990
3702
+ },
3703
+ {
3704
+ "epoch": 1.8,
3705
+ "learning_rate": 1.969332531569453e-05,
3706
+ "loss": 0.0051,
3707
+ "step": 6000
3708
+ },
3709
+ {
3710
+ "epoch": 1.8,
3711
+ "eval_accuracy": 0.9932356257046223,
3712
+ "eval_loss": 0.019992699846625328,
3713
+ "eval_runtime": 228.6869,
3714
+ "eval_samples_per_second": 58.18,
3715
+ "eval_steps_per_second": 7.276,
3716
+ "step": 6000
3717
+ },
3718
+ {
3719
+ "epoch": 1.81,
3720
+ "learning_rate": 1.9392663860493084e-05,
3721
+ "loss": 0.0059,
3722
+ "step": 6010
3723
+ },
3724
+ {
3725
+ "epoch": 1.81,
3726
+ "learning_rate": 1.9092002405291643e-05,
3727
+ "loss": 0.0248,
3728
+ "step": 6020
3729
+ },
3730
+ {
3731
+ "epoch": 1.81,
3732
+ "learning_rate": 1.87913409500902e-05,
3733
+ "loss": 0.0229,
3734
+ "step": 6030
3735
+ },
3736
+ {
3737
+ "epoch": 1.82,
3738
+ "learning_rate": 1.849067949488876e-05,
3739
+ "loss": 0.0323,
3740
+ "step": 6040
3741
+ },
3742
+ {
3743
+ "epoch": 1.82,
3744
+ "learning_rate": 1.819001803968731e-05,
3745
+ "loss": 0.0015,
3746
+ "step": 6050
3747
+ },
3748
+ {
3749
+ "epoch": 1.82,
3750
+ "learning_rate": 1.788935658448587e-05,
3751
+ "loss": 0.0197,
3752
+ "step": 6060
3753
+ },
3754
+ {
3755
+ "epoch": 1.83,
3756
+ "learning_rate": 1.7588695129284427e-05,
3757
+ "loss": 0.0016,
3758
+ "step": 6070
3759
+ },
3760
+ {
3761
+ "epoch": 1.83,
3762
+ "learning_rate": 1.7288033674082983e-05,
3763
+ "loss": 0.0104,
3764
+ "step": 6080
3765
+ },
3766
+ {
3767
+ "epoch": 1.83,
3768
+ "learning_rate": 1.698737221888154e-05,
3769
+ "loss": 0.009,
3770
+ "step": 6090
3771
+ },
3772
+ {
3773
+ "epoch": 1.83,
3774
+ "learning_rate": 1.6686710763680095e-05,
3775
+ "loss": 0.0144,
3776
+ "step": 6100
3777
+ },
3778
+ {
3779
+ "epoch": 1.84,
3780
+ "learning_rate": 1.6386049308478655e-05,
3781
+ "loss": 0.0251,
3782
+ "step": 6110
3783
+ },
3784
+ {
3785
+ "epoch": 1.84,
3786
+ "learning_rate": 1.608538785327721e-05,
3787
+ "loss": 0.0175,
3788
+ "step": 6120
3789
+ },
3790
+ {
3791
+ "epoch": 1.84,
3792
+ "learning_rate": 1.5784726398075767e-05,
3793
+ "loss": 0.0103,
3794
+ "step": 6130
3795
+ },
3796
+ {
3797
+ "epoch": 1.85,
3798
+ "learning_rate": 1.5484064942874323e-05,
3799
+ "loss": 0.0133,
3800
+ "step": 6140
3801
+ },
3802
+ {
3803
+ "epoch": 1.85,
3804
+ "learning_rate": 1.5183403487672881e-05,
3805
+ "loss": 0.0029,
3806
+ "step": 6150
3807
+ },
3808
+ {
3809
+ "epoch": 1.85,
3810
+ "learning_rate": 1.4882742032471439e-05,
3811
+ "loss": 0.0297,
3812
+ "step": 6160
3813
+ },
3814
+ {
3815
+ "epoch": 1.86,
3816
+ "learning_rate": 1.4582080577269993e-05,
3817
+ "loss": 0.0011,
3818
+ "step": 6170
3819
+ },
3820
+ {
3821
+ "epoch": 1.86,
3822
+ "learning_rate": 1.4281419122068551e-05,
3823
+ "loss": 0.0004,
3824
+ "step": 6180
3825
+ },
3826
+ {
3827
+ "epoch": 1.86,
3828
+ "learning_rate": 1.3980757666867109e-05,
3829
+ "loss": 0.0117,
3830
+ "step": 6190
3831
+ },
3832
+ {
3833
+ "epoch": 1.86,
3834
+ "learning_rate": 1.3680096211665667e-05,
3835
+ "loss": 0.0312,
3836
+ "step": 6200
3837
+ },
3838
+ {
3839
+ "epoch": 1.87,
3840
+ "learning_rate": 1.3379434756464221e-05,
3841
+ "loss": 0.0014,
3842
+ "step": 6210
3843
+ },
3844
+ {
3845
+ "epoch": 1.87,
3846
+ "learning_rate": 1.3078773301262779e-05,
3847
+ "loss": 0.0024,
3848
+ "step": 6220
3849
+ },
3850
+ {
3851
+ "epoch": 1.87,
3852
+ "learning_rate": 1.2778111846061335e-05,
3853
+ "loss": 0.0011,
3854
+ "step": 6230
3855
+ },
3856
+ {
3857
+ "epoch": 1.88,
3858
+ "learning_rate": 1.2477450390859891e-05,
3859
+ "loss": 0.0239,
3860
+ "step": 6240
3861
+ },
3862
+ {
3863
+ "epoch": 1.88,
3864
+ "learning_rate": 1.2176788935658449e-05,
3865
+ "loss": 0.0023,
3866
+ "step": 6250
3867
+ },
3868
+ {
3869
+ "epoch": 1.88,
3870
+ "learning_rate": 1.1876127480457005e-05,
3871
+ "loss": 0.0332,
3872
+ "step": 6260
3873
+ },
3874
+ {
3875
+ "epoch": 1.89,
3876
+ "learning_rate": 1.1575466025255563e-05,
3877
+ "loss": 0.0015,
3878
+ "step": 6270
3879
+ },
3880
+ {
3881
+ "epoch": 1.89,
3882
+ "learning_rate": 1.1274804570054119e-05,
3883
+ "loss": 0.0008,
3884
+ "step": 6280
3885
+ },
3886
+ {
3887
+ "epoch": 1.89,
3888
+ "learning_rate": 1.0974143114852677e-05,
3889
+ "loss": 0.0364,
3890
+ "step": 6290
3891
+ },
3892
+ {
3893
+ "epoch": 1.89,
3894
+ "learning_rate": 1.0673481659651233e-05,
3895
+ "loss": 0.001,
3896
+ "step": 6300
3897
+ },
3898
+ {
3899
+ "epoch": 1.9,
3900
+ "learning_rate": 1.037282020444979e-05,
3901
+ "loss": 0.0007,
3902
+ "step": 6310
3903
+ },
3904
+ {
3905
+ "epoch": 1.9,
3906
+ "learning_rate": 1.0072158749248347e-05,
3907
+ "loss": 0.011,
3908
+ "step": 6320
3909
+ },
3910
+ {
3911
+ "epoch": 1.9,
3912
+ "learning_rate": 9.771497294046905e-06,
3913
+ "loss": 0.0007,
3914
+ "step": 6330
3915
+ },
3916
+ {
3917
+ "epoch": 1.91,
3918
+ "learning_rate": 9.47083583884546e-06,
3919
+ "loss": 0.0009,
3920
+ "step": 6340
3921
+ },
3922
+ {
3923
+ "epoch": 1.91,
3924
+ "learning_rate": 9.170174383644017e-06,
3925
+ "loss": 0.0748,
3926
+ "step": 6350
3927
+ },
3928
+ {
3929
+ "epoch": 1.91,
3930
+ "learning_rate": 8.869512928442573e-06,
3931
+ "loss": 0.0243,
3932
+ "step": 6360
3933
+ },
3934
+ {
3935
+ "epoch": 1.92,
3936
+ "learning_rate": 8.56885147324113e-06,
3937
+ "loss": 0.0041,
3938
+ "step": 6370
3939
+ },
3940
+ {
3941
+ "epoch": 1.92,
3942
+ "learning_rate": 8.268190018039687e-06,
3943
+ "loss": 0.0024,
3944
+ "step": 6380
3945
+ },
3946
+ {
3947
+ "epoch": 1.92,
3948
+ "learning_rate": 7.967528562838245e-06,
3949
+ "loss": 0.0326,
3950
+ "step": 6390
3951
+ },
3952
+ {
3953
+ "epoch": 1.92,
3954
+ "learning_rate": 7.6668671076368e-06,
3955
+ "loss": 0.0018,
3956
+ "step": 6400
3957
+ },
3958
+ {
3959
+ "epoch": 1.93,
3960
+ "learning_rate": 7.3662056524353585e-06,
3961
+ "loss": 0.0029,
3962
+ "step": 6410
3963
+ },
3964
+ {
3965
+ "epoch": 1.93,
3966
+ "learning_rate": 7.065544197233915e-06,
3967
+ "loss": 0.0155,
3968
+ "step": 6420
3969
+ },
3970
+ {
3971
+ "epoch": 1.93,
3972
+ "learning_rate": 6.7648827420324715e-06,
3973
+ "loss": 0.0039,
3974
+ "step": 6430
3975
+ },
3976
+ {
3977
+ "epoch": 1.94,
3978
+ "learning_rate": 6.464221286831029e-06,
3979
+ "loss": 0.0137,
3980
+ "step": 6440
3981
+ },
3982
+ {
3983
+ "epoch": 1.94,
3984
+ "learning_rate": 6.1635598316295854e-06,
3985
+ "loss": 0.0062,
3986
+ "step": 6450
3987
+ },
3988
+ {
3989
+ "epoch": 1.94,
3990
+ "learning_rate": 5.862898376428142e-06,
3991
+ "loss": 0.0116,
3992
+ "step": 6460
3993
+ },
3994
+ {
3995
+ "epoch": 1.95,
3996
+ "learning_rate": 5.562236921226699e-06,
3997
+ "loss": 0.0005,
3998
+ "step": 6470
3999
+ },
4000
+ {
4001
+ "epoch": 1.95,
4002
+ "learning_rate": 5.2615754660252554e-06,
4003
+ "loss": 0.0017,
4004
+ "step": 6480
4005
+ },
4006
+ {
4007
+ "epoch": 1.95,
4008
+ "learning_rate": 4.960914010823812e-06,
4009
+ "loss": 0.0015,
4010
+ "step": 6490
4011
+ },
4012
+ {
4013
+ "epoch": 1.95,
4014
+ "learning_rate": 4.660252555622369e-06,
4015
+ "loss": 0.0047,
4016
+ "step": 6500
4017
+ },
4018
+ {
4019
+ "epoch": 1.95,
4020
+ "eval_accuracy": 0.995640736565201,
4021
+ "eval_loss": 0.01245423499494791,
4022
+ "eval_runtime": 226.114,
4023
+ "eval_samples_per_second": 58.842,
4024
+ "eval_steps_per_second": 7.359,
4025
+ "step": 6500
4026
+ },
4027
+ {
4028
+ "epoch": 1.96,
4029
+ "learning_rate": 4.359591100420926e-06,
4030
+ "loss": 0.0157,
4031
+ "step": 6510
4032
+ },
4033
+ {
4034
+ "epoch": 1.96,
4035
+ "learning_rate": 4.058929645219483e-06,
4036
+ "loss": 0.016,
4037
+ "step": 6520
4038
+ },
4039
+ {
4040
+ "epoch": 1.96,
4041
+ "learning_rate": 3.7582681900180398e-06,
4042
+ "loss": 0.0105,
4043
+ "step": 6530
4044
+ },
4045
+ {
4046
+ "epoch": 1.97,
4047
+ "learning_rate": 3.4576067348165963e-06,
4048
+ "loss": 0.0051,
4049
+ "step": 6540
4050
+ },
4051
+ {
4052
+ "epoch": 1.97,
4053
+ "learning_rate": 3.1569452796151533e-06,
4054
+ "loss": 0.0019,
4055
+ "step": 6550
4056
+ },
4057
+ {
4058
+ "epoch": 1.97,
4059
+ "learning_rate": 2.8562838244137106e-06,
4060
+ "loss": 0.0093,
4061
+ "step": 6560
4062
+ },
4063
+ {
4064
+ "epoch": 1.98,
4065
+ "learning_rate": 2.555622369212267e-06,
4066
+ "loss": 0.0079,
4067
+ "step": 6570
4068
+ },
4069
+ {
4070
+ "epoch": 1.98,
4071
+ "learning_rate": 2.254960914010824e-06,
4072
+ "loss": 0.0032,
4073
+ "step": 6580
4074
+ },
4075
+ {
4076
+ "epoch": 1.98,
4077
+ "learning_rate": 1.9542994588093807e-06,
4078
+ "loss": 0.002,
4079
+ "step": 6590
4080
+ },
4081
+ {
4082
+ "epoch": 1.98,
4083
+ "learning_rate": 1.6536380036079376e-06,
4084
+ "loss": 0.018,
4085
+ "step": 6600
4086
+ },
4087
+ {
4088
+ "epoch": 1.99,
4089
+ "learning_rate": 1.3529765484064943e-06,
4090
+ "loss": 0.0184,
4091
+ "step": 6610
4092
+ },
4093
+ {
4094
+ "epoch": 1.99,
4095
+ "learning_rate": 1.0523150932050513e-06,
4096
+ "loss": 0.002,
4097
+ "step": 6620
4098
+ },
4099
+ {
4100
+ "epoch": 1.99,
4101
+ "learning_rate": 7.51653638003608e-07,
4102
+ "loss": 0.0055,
4103
+ "step": 6630
4104
+ },
4105
+ {
4106
+ "epoch": 2.0,
4107
+ "learning_rate": 4.509921828021648e-07,
4108
+ "loss": 0.0086,
4109
+ "step": 6640
4110
+ },
4111
+ {
4112
+ "epoch": 2.0,
4113
+ "learning_rate": 1.503307276007216e-07,
4114
+ "loss": 0.0198,
4115
+ "step": 6650
4116
+ },
4117
+ {
4118
+ "epoch": 2.0,
4119
+ "step": 6652,
4120
+ "total_flos": 8.247627438660256e+18,
4121
+ "train_loss": 0.07274016925322493,
4122
+ "train_runtime": 6476.0215,
4123
+ "train_samples_per_second": 16.435,
4124
+ "train_steps_per_second": 1.027
4125
+ }
4126
+ ],
4127
+ "max_steps": 6652,
4128
+ "num_train_epochs": 2,
4129
+ "total_flos": 8.247627438660256e+18,
4130
+ "trial_name": null,
4131
+ "trial_params": null
4132
+ }