End of training
Browse files- all_results.json +13 -0
- eval_results.json +8 -0
- runs/May04_11-38-48_f32ee6fae861/events.out.tfevents.1683207126.f32ee6fae861.5465.4 +3 -0
- train_results.json +8 -0
- trainer_state.json +1969 -0
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.994535519125683,
|
| 4 |
+
"eval_loss": 0.035519640892744064,
|
| 5 |
+
"eval_runtime": 30.2809,
|
| 6 |
+
"eval_samples_per_second": 30.217,
|
| 7 |
+
"eval_steps_per_second": 3.798,
|
| 8 |
+
"total_flos": 4.089918706717778e+18,
|
| 9 |
+
"train_loss": 0.08860804434029423,
|
| 10 |
+
"train_runtime": 6732.3279,
|
| 11 |
+
"train_samples_per_second": 24.44,
|
| 12 |
+
"train_steps_per_second": 0.437
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.994535519125683,
|
| 4 |
+
"eval_loss": 0.035519640892744064,
|
| 5 |
+
"eval_runtime": 30.2809,
|
| 6 |
+
"eval_samples_per_second": 30.217,
|
| 7 |
+
"eval_steps_per_second": 3.798
|
| 8 |
+
}
|
runs/May04_11-38-48_f32ee6fae861/events.out.tfevents.1683207126.f32ee6fae861.5465.4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:19f5217e07e78e373066a61ab2b92b0a9e9b5008f0036b073ae06758fce2f691
|
| 3 |
+
size 411
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"total_flos": 4.089918706717778e+18,
|
| 4 |
+
"train_loss": 0.08860804434029423,
|
| 5 |
+
"train_runtime": 6732.3279,
|
| 6 |
+
"train_samples_per_second": 24.44,
|
| 7 |
+
"train_steps_per_second": 0.437
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1969 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.994535519125683,
|
| 3 |
+
"best_model_checkpoint": "swin-tiny-patch4-window7-224-shortSleeveCleanedData/checkpoint-2352",
|
| 4 |
+
"epoch": 20.0,
|
| 5 |
+
"global_step": 2940,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.07,
|
| 12 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 13 |
+
"loss": 1.098,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.14,
|
| 18 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 19 |
+
"loss": 0.9229,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.2,
|
| 24 |
+
"learning_rate": 5e-05,
|
| 25 |
+
"loss": 0.6598,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.27,
|
| 30 |
+
"learning_rate": 4.982817869415808e-05,
|
| 31 |
+
"loss": 0.3916,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.34,
|
| 36 |
+
"learning_rate": 4.965635738831615e-05,
|
| 37 |
+
"loss": 0.2737,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.41,
|
| 42 |
+
"learning_rate": 4.948453608247423e-05,
|
| 43 |
+
"loss": 0.2705,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.48,
|
| 48 |
+
"learning_rate": 4.931271477663231e-05,
|
| 49 |
+
"loss": 0.245,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.54,
|
| 54 |
+
"learning_rate": 4.9140893470790375e-05,
|
| 55 |
+
"loss": 0.1672,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.61,
|
| 60 |
+
"learning_rate": 4.8969072164948454e-05,
|
| 61 |
+
"loss": 0.1976,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.68,
|
| 66 |
+
"learning_rate": 4.879725085910653e-05,
|
| 67 |
+
"loss": 0.2397,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.75,
|
| 72 |
+
"learning_rate": 4.8625429553264604e-05,
|
| 73 |
+
"loss": 0.168,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.82,
|
| 78 |
+
"learning_rate": 4.845360824742268e-05,
|
| 79 |
+
"loss": 0.1745,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.88,
|
| 84 |
+
"learning_rate": 4.828178694158076e-05,
|
| 85 |
+
"loss": 0.202,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.95,
|
| 90 |
+
"learning_rate": 4.810996563573883e-05,
|
| 91 |
+
"loss": 0.1819,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 1.0,
|
| 96 |
+
"eval_accuracy": 0.9879781420765027,
|
| 97 |
+
"eval_loss": 0.04708021134138107,
|
| 98 |
+
"eval_runtime": 29.2491,
|
| 99 |
+
"eval_samples_per_second": 31.283,
|
| 100 |
+
"eval_steps_per_second": 3.932,
|
| 101 |
+
"step": 147
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 1.02,
|
| 105 |
+
"learning_rate": 4.793814432989691e-05,
|
| 106 |
+
"loss": 0.1651,
|
| 107 |
+
"step": 150
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.09,
|
| 111 |
+
"learning_rate": 4.776632302405499e-05,
|
| 112 |
+
"loss": 0.1301,
|
| 113 |
+
"step": 160
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 1.16,
|
| 117 |
+
"learning_rate": 4.7594501718213055e-05,
|
| 118 |
+
"loss": 0.1177,
|
| 119 |
+
"step": 170
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.22,
|
| 123 |
+
"learning_rate": 4.7422680412371134e-05,
|
| 124 |
+
"loss": 0.1653,
|
| 125 |
+
"step": 180
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 1.29,
|
| 129 |
+
"learning_rate": 4.725085910652921e-05,
|
| 130 |
+
"loss": 0.1691,
|
| 131 |
+
"step": 190
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 1.36,
|
| 135 |
+
"learning_rate": 4.7079037800687284e-05,
|
| 136 |
+
"loss": 0.1266,
|
| 137 |
+
"step": 200
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.43,
|
| 141 |
+
"learning_rate": 4.690721649484536e-05,
|
| 142 |
+
"loss": 0.1334,
|
| 143 |
+
"step": 210
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 1.5,
|
| 147 |
+
"learning_rate": 4.673539518900344e-05,
|
| 148 |
+
"loss": 0.1716,
|
| 149 |
+
"step": 220
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.56,
|
| 153 |
+
"learning_rate": 4.656357388316151e-05,
|
| 154 |
+
"loss": 0.1662,
|
| 155 |
+
"step": 230
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 1.63,
|
| 159 |
+
"learning_rate": 4.639175257731959e-05,
|
| 160 |
+
"loss": 0.1527,
|
| 161 |
+
"step": 240
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 1.7,
|
| 165 |
+
"learning_rate": 4.621993127147767e-05,
|
| 166 |
+
"loss": 0.1171,
|
| 167 |
+
"step": 250
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 1.77,
|
| 171 |
+
"learning_rate": 4.6048109965635736e-05,
|
| 172 |
+
"loss": 0.1966,
|
| 173 |
+
"step": 260
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 1.84,
|
| 177 |
+
"learning_rate": 4.5876288659793814e-05,
|
| 178 |
+
"loss": 0.1274,
|
| 179 |
+
"step": 270
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.9,
|
| 183 |
+
"learning_rate": 4.570446735395189e-05,
|
| 184 |
+
"loss": 0.1589,
|
| 185 |
+
"step": 280
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.97,
|
| 189 |
+
"learning_rate": 4.5532646048109965e-05,
|
| 190 |
+
"loss": 0.1431,
|
| 191 |
+
"step": 290
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 2.0,
|
| 195 |
+
"eval_accuracy": 0.9890710382513661,
|
| 196 |
+
"eval_loss": 0.045684538781642914,
|
| 197 |
+
"eval_runtime": 29.1141,
|
| 198 |
+
"eval_samples_per_second": 31.428,
|
| 199 |
+
"eval_steps_per_second": 3.95,
|
| 200 |
+
"step": 294
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 2.04,
|
| 204 |
+
"learning_rate": 4.536082474226804e-05,
|
| 205 |
+
"loss": 0.1321,
|
| 206 |
+
"step": 300
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 2.11,
|
| 210 |
+
"learning_rate": 4.518900343642612e-05,
|
| 211 |
+
"loss": 0.1454,
|
| 212 |
+
"step": 310
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 2.18,
|
| 216 |
+
"learning_rate": 4.5017182130584194e-05,
|
| 217 |
+
"loss": 0.1106,
|
| 218 |
+
"step": 320
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 2.24,
|
| 222 |
+
"learning_rate": 4.484536082474227e-05,
|
| 223 |
+
"loss": 0.1536,
|
| 224 |
+
"step": 330
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 2.31,
|
| 228 |
+
"learning_rate": 4.467353951890035e-05,
|
| 229 |
+
"loss": 0.1166,
|
| 230 |
+
"step": 340
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 2.38,
|
| 234 |
+
"learning_rate": 4.4501718213058416e-05,
|
| 235 |
+
"loss": 0.1172,
|
| 236 |
+
"step": 350
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 2.45,
|
| 240 |
+
"learning_rate": 4.4329896907216494e-05,
|
| 241 |
+
"loss": 0.1021,
|
| 242 |
+
"step": 360
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 2.52,
|
| 246 |
+
"learning_rate": 4.415807560137457e-05,
|
| 247 |
+
"loss": 0.1234,
|
| 248 |
+
"step": 370
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 2.59,
|
| 252 |
+
"learning_rate": 4.3986254295532645e-05,
|
| 253 |
+
"loss": 0.101,
|
| 254 |
+
"step": 380
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 2.65,
|
| 258 |
+
"learning_rate": 4.3814432989690723e-05,
|
| 259 |
+
"loss": 0.1254,
|
| 260 |
+
"step": 390
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 2.72,
|
| 264 |
+
"learning_rate": 4.36426116838488e-05,
|
| 265 |
+
"loss": 0.1676,
|
| 266 |
+
"step": 400
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 2.79,
|
| 270 |
+
"learning_rate": 4.3470790378006874e-05,
|
| 271 |
+
"loss": 0.1297,
|
| 272 |
+
"step": 410
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.86,
|
| 276 |
+
"learning_rate": 4.329896907216495e-05,
|
| 277 |
+
"loss": 0.1346,
|
| 278 |
+
"step": 420
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 2.93,
|
| 282 |
+
"learning_rate": 4.312714776632303e-05,
|
| 283 |
+
"loss": 0.1123,
|
| 284 |
+
"step": 430
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 2.99,
|
| 288 |
+
"learning_rate": 4.2955326460481096e-05,
|
| 289 |
+
"loss": 0.1001,
|
| 290 |
+
"step": 440
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 3.0,
|
| 294 |
+
"eval_accuracy": 0.9890710382513661,
|
| 295 |
+
"eval_loss": 0.039226751774549484,
|
| 296 |
+
"eval_runtime": 29.509,
|
| 297 |
+
"eval_samples_per_second": 31.008,
|
| 298 |
+
"eval_steps_per_second": 3.897,
|
| 299 |
+
"step": 441
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 3.06,
|
| 303 |
+
"learning_rate": 4.2783505154639175e-05,
|
| 304 |
+
"loss": 0.1042,
|
| 305 |
+
"step": 450
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 3.13,
|
| 309 |
+
"learning_rate": 4.261168384879725e-05,
|
| 310 |
+
"loss": 0.1238,
|
| 311 |
+
"step": 460
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 3.2,
|
| 315 |
+
"learning_rate": 4.2439862542955325e-05,
|
| 316 |
+
"loss": 0.0994,
|
| 317 |
+
"step": 470
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 3.27,
|
| 321 |
+
"learning_rate": 4.2268041237113404e-05,
|
| 322 |
+
"loss": 0.1055,
|
| 323 |
+
"step": 480
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 3.33,
|
| 327 |
+
"learning_rate": 4.209621993127148e-05,
|
| 328 |
+
"loss": 0.1545,
|
| 329 |
+
"step": 490
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 3.4,
|
| 333 |
+
"learning_rate": 4.1924398625429554e-05,
|
| 334 |
+
"loss": 0.0874,
|
| 335 |
+
"step": 500
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 3.47,
|
| 339 |
+
"learning_rate": 4.175257731958763e-05,
|
| 340 |
+
"loss": 0.1033,
|
| 341 |
+
"step": 510
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 3.54,
|
| 345 |
+
"learning_rate": 4.158075601374571e-05,
|
| 346 |
+
"loss": 0.1123,
|
| 347 |
+
"step": 520
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 3.61,
|
| 351 |
+
"learning_rate": 4.140893470790378e-05,
|
| 352 |
+
"loss": 0.0974,
|
| 353 |
+
"step": 530
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 3.67,
|
| 357 |
+
"learning_rate": 4.1237113402061855e-05,
|
| 358 |
+
"loss": 0.1005,
|
| 359 |
+
"step": 540
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 3.74,
|
| 363 |
+
"learning_rate": 4.1065292096219934e-05,
|
| 364 |
+
"loss": 0.095,
|
| 365 |
+
"step": 550
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 3.81,
|
| 369 |
+
"learning_rate": 4.0893470790378005e-05,
|
| 370 |
+
"loss": 0.1275,
|
| 371 |
+
"step": 560
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 3.88,
|
| 375 |
+
"learning_rate": 4.0721649484536084e-05,
|
| 376 |
+
"loss": 0.1174,
|
| 377 |
+
"step": 570
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 3.95,
|
| 381 |
+
"learning_rate": 4.054982817869416e-05,
|
| 382 |
+
"loss": 0.116,
|
| 383 |
+
"step": 580
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 4.0,
|
| 387 |
+
"eval_accuracy": 0.9879781420765027,
|
| 388 |
+
"eval_loss": 0.045129090547561646,
|
| 389 |
+
"eval_runtime": 29.719,
|
| 390 |
+
"eval_samples_per_second": 30.788,
|
| 391 |
+
"eval_steps_per_second": 3.87,
|
| 392 |
+
"step": 588
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 4.01,
|
| 396 |
+
"learning_rate": 4.0378006872852234e-05,
|
| 397 |
+
"loss": 0.0961,
|
| 398 |
+
"step": 590
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 4.08,
|
| 402 |
+
"learning_rate": 4.020618556701031e-05,
|
| 403 |
+
"loss": 0.0953,
|
| 404 |
+
"step": 600
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 4.15,
|
| 408 |
+
"learning_rate": 4.003436426116839e-05,
|
| 409 |
+
"loss": 0.0921,
|
| 410 |
+
"step": 610
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 4.22,
|
| 414 |
+
"learning_rate": 3.9862542955326463e-05,
|
| 415 |
+
"loss": 0.1213,
|
| 416 |
+
"step": 620
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 4.29,
|
| 420 |
+
"learning_rate": 3.9690721649484535e-05,
|
| 421 |
+
"loss": 0.1133,
|
| 422 |
+
"step": 630
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 4.35,
|
| 426 |
+
"learning_rate": 3.9518900343642614e-05,
|
| 427 |
+
"loss": 0.1442,
|
| 428 |
+
"step": 640
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 4.42,
|
| 432 |
+
"learning_rate": 3.9347079037800686e-05,
|
| 433 |
+
"loss": 0.0827,
|
| 434 |
+
"step": 650
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 4.49,
|
| 438 |
+
"learning_rate": 3.9175257731958764e-05,
|
| 439 |
+
"loss": 0.0651,
|
| 440 |
+
"step": 660
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 4.56,
|
| 444 |
+
"learning_rate": 3.900343642611684e-05,
|
| 445 |
+
"loss": 0.0906,
|
| 446 |
+
"step": 670
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 4.63,
|
| 450 |
+
"learning_rate": 3.8831615120274915e-05,
|
| 451 |
+
"loss": 0.0607,
|
| 452 |
+
"step": 680
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 4.69,
|
| 456 |
+
"learning_rate": 3.865979381443299e-05,
|
| 457 |
+
"loss": 0.0998,
|
| 458 |
+
"step": 690
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 4.76,
|
| 462 |
+
"learning_rate": 3.848797250859107e-05,
|
| 463 |
+
"loss": 0.0882,
|
| 464 |
+
"step": 700
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 4.83,
|
| 468 |
+
"learning_rate": 3.8316151202749144e-05,
|
| 469 |
+
"loss": 0.0704,
|
| 470 |
+
"step": 710
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 4.9,
|
| 474 |
+
"learning_rate": 3.8144329896907216e-05,
|
| 475 |
+
"loss": 0.0689,
|
| 476 |
+
"step": 720
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 4.97,
|
| 480 |
+
"learning_rate": 3.7972508591065294e-05,
|
| 481 |
+
"loss": 0.1144,
|
| 482 |
+
"step": 730
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 5.0,
|
| 486 |
+
"eval_accuracy": 0.9901639344262295,
|
| 487 |
+
"eval_loss": 0.03983403742313385,
|
| 488 |
+
"eval_runtime": 29.5173,
|
| 489 |
+
"eval_samples_per_second": 30.999,
|
| 490 |
+
"eval_steps_per_second": 3.896,
|
| 491 |
+
"step": 735
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 5.03,
|
| 495 |
+
"learning_rate": 3.7800687285223366e-05,
|
| 496 |
+
"loss": 0.0698,
|
| 497 |
+
"step": 740
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 5.1,
|
| 501 |
+
"learning_rate": 3.7628865979381445e-05,
|
| 502 |
+
"loss": 0.0875,
|
| 503 |
+
"step": 750
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 5.17,
|
| 507 |
+
"learning_rate": 3.745704467353952e-05,
|
| 508 |
+
"loss": 0.0932,
|
| 509 |
+
"step": 760
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 5.24,
|
| 513 |
+
"learning_rate": 3.7285223367697595e-05,
|
| 514 |
+
"loss": 0.0789,
|
| 515 |
+
"step": 770
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 5.31,
|
| 519 |
+
"learning_rate": 3.7113402061855674e-05,
|
| 520 |
+
"loss": 0.0793,
|
| 521 |
+
"step": 780
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 5.37,
|
| 525 |
+
"learning_rate": 3.694158075601375e-05,
|
| 526 |
+
"loss": 0.1196,
|
| 527 |
+
"step": 790
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 5.44,
|
| 531 |
+
"learning_rate": 3.6769759450171824e-05,
|
| 532 |
+
"loss": 0.0964,
|
| 533 |
+
"step": 800
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 5.51,
|
| 537 |
+
"learning_rate": 3.6597938144329896e-05,
|
| 538 |
+
"loss": 0.1047,
|
| 539 |
+
"step": 810
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 5.58,
|
| 543 |
+
"learning_rate": 3.6426116838487974e-05,
|
| 544 |
+
"loss": 0.0932,
|
| 545 |
+
"step": 820
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 5.65,
|
| 549 |
+
"learning_rate": 3.6254295532646046e-05,
|
| 550 |
+
"loss": 0.0789,
|
| 551 |
+
"step": 830
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 5.71,
|
| 555 |
+
"learning_rate": 3.6082474226804125e-05,
|
| 556 |
+
"loss": 0.0619,
|
| 557 |
+
"step": 840
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 5.78,
|
| 561 |
+
"learning_rate": 3.5910652920962203e-05,
|
| 562 |
+
"loss": 0.0654,
|
| 563 |
+
"step": 850
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 5.85,
|
| 567 |
+
"learning_rate": 3.5738831615120275e-05,
|
| 568 |
+
"loss": 0.1124,
|
| 569 |
+
"step": 860
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 5.92,
|
| 573 |
+
"learning_rate": 3.5567010309278354e-05,
|
| 574 |
+
"loss": 0.1295,
|
| 575 |
+
"step": 870
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 5.99,
|
| 579 |
+
"learning_rate": 3.539518900343643e-05,
|
| 580 |
+
"loss": 0.0787,
|
| 581 |
+
"step": 880
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 6.0,
|
| 585 |
+
"eval_accuracy": 0.9901639344262295,
|
| 586 |
+
"eval_loss": 0.04411642625927925,
|
| 587 |
+
"eval_runtime": 29.3848,
|
| 588 |
+
"eval_samples_per_second": 31.139,
|
| 589 |
+
"eval_steps_per_second": 3.914,
|
| 590 |
+
"step": 882
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 6.05,
|
| 594 |
+
"learning_rate": 3.5223367697594504e-05,
|
| 595 |
+
"loss": 0.0664,
|
| 596 |
+
"step": 890
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 6.12,
|
| 600 |
+
"learning_rate": 3.5051546391752576e-05,
|
| 601 |
+
"loss": 0.1042,
|
| 602 |
+
"step": 900
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 6.19,
|
| 606 |
+
"learning_rate": 3.4879725085910655e-05,
|
| 607 |
+
"loss": 0.0516,
|
| 608 |
+
"step": 910
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 6.26,
|
| 612 |
+
"learning_rate": 3.4707903780068726e-05,
|
| 613 |
+
"loss": 0.1106,
|
| 614 |
+
"step": 920
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 6.33,
|
| 618 |
+
"learning_rate": 3.4536082474226805e-05,
|
| 619 |
+
"loss": 0.1176,
|
| 620 |
+
"step": 930
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 6.39,
|
| 624 |
+
"learning_rate": 3.4364261168384884e-05,
|
| 625 |
+
"loss": 0.1239,
|
| 626 |
+
"step": 940
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 6.46,
|
| 630 |
+
"learning_rate": 3.4192439862542955e-05,
|
| 631 |
+
"loss": 0.0996,
|
| 632 |
+
"step": 950
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 6.53,
|
| 636 |
+
"learning_rate": 3.4020618556701034e-05,
|
| 637 |
+
"loss": 0.0705,
|
| 638 |
+
"step": 960
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 6.6,
|
| 642 |
+
"learning_rate": 3.384879725085911e-05,
|
| 643 |
+
"loss": 0.0674,
|
| 644 |
+
"step": 970
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 6.67,
|
| 648 |
+
"learning_rate": 3.3676975945017185e-05,
|
| 649 |
+
"loss": 0.0857,
|
| 650 |
+
"step": 980
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 6.73,
|
| 654 |
+
"learning_rate": 3.3505154639175256e-05,
|
| 655 |
+
"loss": 0.1103,
|
| 656 |
+
"step": 990
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 6.8,
|
| 660 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 661 |
+
"loss": 0.0778,
|
| 662 |
+
"step": 1000
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 6.87,
|
| 666 |
+
"learning_rate": 3.316151202749141e-05,
|
| 667 |
+
"loss": 0.0963,
|
| 668 |
+
"step": 1010
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 6.94,
|
| 672 |
+
"learning_rate": 3.2989690721649485e-05,
|
| 673 |
+
"loss": 0.0998,
|
| 674 |
+
"step": 1020
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 7.0,
|
| 678 |
+
"eval_accuracy": 0.9901639344262295,
|
| 679 |
+
"eval_loss": 0.03199444338679314,
|
| 680 |
+
"eval_runtime": 30.1676,
|
| 681 |
+
"eval_samples_per_second": 30.331,
|
| 682 |
+
"eval_steps_per_second": 3.812,
|
| 683 |
+
"step": 1029
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 7.01,
|
| 687 |
+
"learning_rate": 3.2817869415807564e-05,
|
| 688 |
+
"loss": 0.0896,
|
| 689 |
+
"step": 1030
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 7.07,
|
| 693 |
+
"learning_rate": 3.2646048109965636e-05,
|
| 694 |
+
"loss": 0.0835,
|
| 695 |
+
"step": 1040
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 7.14,
|
| 699 |
+
"learning_rate": 3.2474226804123714e-05,
|
| 700 |
+
"loss": 0.0665,
|
| 701 |
+
"step": 1050
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 7.21,
|
| 705 |
+
"learning_rate": 3.230240549828179e-05,
|
| 706 |
+
"loss": 0.0733,
|
| 707 |
+
"step": 1060
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 7.28,
|
| 711 |
+
"learning_rate": 3.2130584192439865e-05,
|
| 712 |
+
"loss": 0.0852,
|
| 713 |
+
"step": 1070
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 7.35,
|
| 717 |
+
"learning_rate": 3.1958762886597937e-05,
|
| 718 |
+
"loss": 0.0618,
|
| 719 |
+
"step": 1080
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 7.41,
|
| 723 |
+
"learning_rate": 3.1786941580756015e-05,
|
| 724 |
+
"loss": 0.1017,
|
| 725 |
+
"step": 1090
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 7.48,
|
| 729 |
+
"learning_rate": 3.161512027491409e-05,
|
| 730 |
+
"loss": 0.0451,
|
| 731 |
+
"step": 1100
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 7.55,
|
| 735 |
+
"learning_rate": 3.1443298969072166e-05,
|
| 736 |
+
"loss": 0.0663,
|
| 737 |
+
"step": 1110
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 7.62,
|
| 741 |
+
"learning_rate": 3.1271477663230244e-05,
|
| 742 |
+
"loss": 0.1017,
|
| 743 |
+
"step": 1120
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 7.69,
|
| 747 |
+
"learning_rate": 3.1099656357388316e-05,
|
| 748 |
+
"loss": 0.1026,
|
| 749 |
+
"step": 1130
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 7.76,
|
| 753 |
+
"learning_rate": 3.0927835051546395e-05,
|
| 754 |
+
"loss": 0.0812,
|
| 755 |
+
"step": 1140
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 7.82,
|
| 759 |
+
"learning_rate": 3.075601374570447e-05,
|
| 760 |
+
"loss": 0.0913,
|
| 761 |
+
"step": 1150
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 7.89,
|
| 765 |
+
"learning_rate": 3.0584192439862545e-05,
|
| 766 |
+
"loss": 0.0783,
|
| 767 |
+
"step": 1160
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 7.96,
|
| 771 |
+
"learning_rate": 3.0412371134020617e-05,
|
| 772 |
+
"loss": 0.124,
|
| 773 |
+
"step": 1170
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 8.0,
|
| 777 |
+
"eval_accuracy": 0.9901639344262295,
|
| 778 |
+
"eval_loss": 0.0364375114440918,
|
| 779 |
+
"eval_runtime": 29.8491,
|
| 780 |
+
"eval_samples_per_second": 30.654,
|
| 781 |
+
"eval_steps_per_second": 3.853,
|
| 782 |
+
"step": 1176
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 8.03,
|
| 786 |
+
"learning_rate": 3.0240549828178692e-05,
|
| 787 |
+
"loss": 0.049,
|
| 788 |
+
"step": 1180
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 8.1,
|
| 792 |
+
"learning_rate": 3.006872852233677e-05,
|
| 793 |
+
"loss": 0.0884,
|
| 794 |
+
"step": 1190
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 8.16,
|
| 798 |
+
"learning_rate": 2.9896907216494846e-05,
|
| 799 |
+
"loss": 0.0843,
|
| 800 |
+
"step": 1200
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 8.23,
|
| 804 |
+
"learning_rate": 2.972508591065292e-05,
|
| 805 |
+
"loss": 0.0892,
|
| 806 |
+
"step": 1210
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 8.3,
|
| 810 |
+
"learning_rate": 2.9553264604811e-05,
|
| 811 |
+
"loss": 0.0468,
|
| 812 |
+
"step": 1220
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 8.37,
|
| 816 |
+
"learning_rate": 2.9381443298969075e-05,
|
| 817 |
+
"loss": 0.0669,
|
| 818 |
+
"step": 1230
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 8.44,
|
| 822 |
+
"learning_rate": 2.920962199312715e-05,
|
| 823 |
+
"loss": 0.067,
|
| 824 |
+
"step": 1240
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 8.5,
|
| 828 |
+
"learning_rate": 2.903780068728523e-05,
|
| 829 |
+
"loss": 0.0585,
|
| 830 |
+
"step": 1250
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 8.57,
|
| 834 |
+
"learning_rate": 2.8865979381443297e-05,
|
| 835 |
+
"loss": 0.032,
|
| 836 |
+
"step": 1260
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 8.64,
|
| 840 |
+
"learning_rate": 2.8694158075601372e-05,
|
| 841 |
+
"loss": 0.1195,
|
| 842 |
+
"step": 1270
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 8.71,
|
| 846 |
+
"learning_rate": 2.852233676975945e-05,
|
| 847 |
+
"loss": 0.0733,
|
| 848 |
+
"step": 1280
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 8.78,
|
| 852 |
+
"learning_rate": 2.8350515463917526e-05,
|
| 853 |
+
"loss": 0.0664,
|
| 854 |
+
"step": 1290
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 8.84,
|
| 858 |
+
"learning_rate": 2.81786941580756e-05,
|
| 859 |
+
"loss": 0.0792,
|
| 860 |
+
"step": 1300
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 8.91,
|
| 864 |
+
"learning_rate": 2.800687285223368e-05,
|
| 865 |
+
"loss": 0.0866,
|
| 866 |
+
"step": 1310
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 8.98,
|
| 870 |
+
"learning_rate": 2.7835051546391755e-05,
|
| 871 |
+
"loss": 0.103,
|
| 872 |
+
"step": 1320
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 9.0,
|
| 876 |
+
"eval_accuracy": 0.9879781420765027,
|
| 877 |
+
"eval_loss": 0.039538685232400894,
|
| 878 |
+
"eval_runtime": 30.6734,
|
| 879 |
+
"eval_samples_per_second": 29.83,
|
| 880 |
+
"eval_steps_per_second": 3.749,
|
| 881 |
+
"step": 1323
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 9.05,
|
| 885 |
+
"learning_rate": 2.766323024054983e-05,
|
| 886 |
+
"loss": 0.0559,
|
| 887 |
+
"step": 1330
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 9.12,
|
| 891 |
+
"learning_rate": 2.749140893470791e-05,
|
| 892 |
+
"loss": 0.0774,
|
| 893 |
+
"step": 1340
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 9.18,
|
| 897 |
+
"learning_rate": 2.7319587628865977e-05,
|
| 898 |
+
"loss": 0.0809,
|
| 899 |
+
"step": 1350
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 9.25,
|
| 903 |
+
"learning_rate": 2.7147766323024053e-05,
|
| 904 |
+
"loss": 0.0984,
|
| 905 |
+
"step": 1360
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 9.32,
|
| 909 |
+
"learning_rate": 2.697594501718213e-05,
|
| 910 |
+
"loss": 0.0838,
|
| 911 |
+
"step": 1370
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 9.39,
|
| 915 |
+
"learning_rate": 2.6804123711340206e-05,
|
| 916 |
+
"loss": 0.0894,
|
| 917 |
+
"step": 1380
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 9.46,
|
| 921 |
+
"learning_rate": 2.663230240549828e-05,
|
| 922 |
+
"loss": 0.0501,
|
| 923 |
+
"step": 1390
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 9.52,
|
| 927 |
+
"learning_rate": 2.646048109965636e-05,
|
| 928 |
+
"loss": 0.0829,
|
| 929 |
+
"step": 1400
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 9.59,
|
| 933 |
+
"learning_rate": 2.6288659793814435e-05,
|
| 934 |
+
"loss": 0.0885,
|
| 935 |
+
"step": 1410
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 9.66,
|
| 939 |
+
"learning_rate": 2.611683848797251e-05,
|
| 940 |
+
"loss": 0.0925,
|
| 941 |
+
"step": 1420
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 9.73,
|
| 945 |
+
"learning_rate": 2.594501718213059e-05,
|
| 946 |
+
"loss": 0.0645,
|
| 947 |
+
"step": 1430
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 9.8,
|
| 951 |
+
"learning_rate": 2.5773195876288658e-05,
|
| 952 |
+
"loss": 0.091,
|
| 953 |
+
"step": 1440
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 9.86,
|
| 957 |
+
"learning_rate": 2.5601374570446733e-05,
|
| 958 |
+
"loss": 0.0785,
|
| 959 |
+
"step": 1450
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 9.93,
|
| 963 |
+
"learning_rate": 2.542955326460481e-05,
|
| 964 |
+
"loss": 0.0794,
|
| 965 |
+
"step": 1460
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 10.0,
|
| 969 |
+
"learning_rate": 2.5257731958762887e-05,
|
| 970 |
+
"loss": 0.0591,
|
| 971 |
+
"step": 1470
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 10.0,
|
| 975 |
+
"eval_accuracy": 0.9912568306010929,
|
| 976 |
+
"eval_loss": 0.029904531314969063,
|
| 977 |
+
"eval_runtime": 31.3335,
|
| 978 |
+
"eval_samples_per_second": 29.202,
|
| 979 |
+
"eval_steps_per_second": 3.67,
|
| 980 |
+
"step": 1470
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 10.07,
|
| 984 |
+
"learning_rate": 2.5085910652920962e-05,
|
| 985 |
+
"loss": 0.0475,
|
| 986 |
+
"step": 1480
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 10.14,
|
| 990 |
+
"learning_rate": 2.491408934707904e-05,
|
| 991 |
+
"loss": 0.0637,
|
| 992 |
+
"step": 1490
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 10.2,
|
| 996 |
+
"learning_rate": 2.4742268041237116e-05,
|
| 997 |
+
"loss": 0.0445,
|
| 998 |
+
"step": 1500
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 10.27,
|
| 1002 |
+
"learning_rate": 2.4570446735395188e-05,
|
| 1003 |
+
"loss": 0.0421,
|
| 1004 |
+
"step": 1510
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 10.34,
|
| 1008 |
+
"learning_rate": 2.4398625429553266e-05,
|
| 1009 |
+
"loss": 0.0807,
|
| 1010 |
+
"step": 1520
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 10.41,
|
| 1014 |
+
"learning_rate": 2.422680412371134e-05,
|
| 1015 |
+
"loss": 0.0552,
|
| 1016 |
+
"step": 1530
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 10.48,
|
| 1020 |
+
"learning_rate": 2.4054982817869417e-05,
|
| 1021 |
+
"loss": 0.0447,
|
| 1022 |
+
"step": 1540
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 10.54,
|
| 1026 |
+
"learning_rate": 2.3883161512027495e-05,
|
| 1027 |
+
"loss": 0.054,
|
| 1028 |
+
"step": 1550
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 10.61,
|
| 1032 |
+
"learning_rate": 2.3711340206185567e-05,
|
| 1033 |
+
"loss": 0.0445,
|
| 1034 |
+
"step": 1560
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 10.68,
|
| 1038 |
+
"learning_rate": 2.3539518900343642e-05,
|
| 1039 |
+
"loss": 0.0747,
|
| 1040 |
+
"step": 1570
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 10.75,
|
| 1044 |
+
"learning_rate": 2.336769759450172e-05,
|
| 1045 |
+
"loss": 0.0705,
|
| 1046 |
+
"step": 1580
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 10.82,
|
| 1050 |
+
"learning_rate": 2.3195876288659796e-05,
|
| 1051 |
+
"loss": 0.0846,
|
| 1052 |
+
"step": 1590
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 10.88,
|
| 1056 |
+
"learning_rate": 2.3024054982817868e-05,
|
| 1057 |
+
"loss": 0.0832,
|
| 1058 |
+
"step": 1600
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 10.95,
|
| 1062 |
+
"learning_rate": 2.2852233676975946e-05,
|
| 1063 |
+
"loss": 0.0445,
|
| 1064 |
+
"step": 1610
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 11.0,
|
| 1068 |
+
"eval_accuracy": 0.9912568306010929,
|
| 1069 |
+
"eval_loss": 0.030193105340003967,
|
| 1070 |
+
"eval_runtime": 29.7828,
|
| 1071 |
+
"eval_samples_per_second": 30.722,
|
| 1072 |
+
"eval_steps_per_second": 3.861,
|
| 1073 |
+
"step": 1617
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 11.02,
|
| 1077 |
+
"learning_rate": 2.268041237113402e-05,
|
| 1078 |
+
"loss": 0.0648,
|
| 1079 |
+
"step": 1620
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 11.09,
|
| 1083 |
+
"learning_rate": 2.2508591065292097e-05,
|
| 1084 |
+
"loss": 0.0355,
|
| 1085 |
+
"step": 1630
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 11.16,
|
| 1089 |
+
"learning_rate": 2.2336769759450175e-05,
|
| 1090 |
+
"loss": 0.0566,
|
| 1091 |
+
"step": 1640
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 11.22,
|
| 1095 |
+
"learning_rate": 2.2164948453608247e-05,
|
| 1096 |
+
"loss": 0.0849,
|
| 1097 |
+
"step": 1650
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 11.29,
|
| 1101 |
+
"learning_rate": 2.1993127147766322e-05,
|
| 1102 |
+
"loss": 0.0469,
|
| 1103 |
+
"step": 1660
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 11.36,
|
| 1107 |
+
"learning_rate": 2.18213058419244e-05,
|
| 1108 |
+
"loss": 0.0702,
|
| 1109 |
+
"step": 1670
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 11.43,
|
| 1113 |
+
"learning_rate": 2.1649484536082476e-05,
|
| 1114 |
+
"loss": 0.0442,
|
| 1115 |
+
"step": 1680
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 11.5,
|
| 1119 |
+
"learning_rate": 2.1477663230240548e-05,
|
| 1120 |
+
"loss": 0.0591,
|
| 1121 |
+
"step": 1690
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 11.56,
|
| 1125 |
+
"learning_rate": 2.1305841924398627e-05,
|
| 1126 |
+
"loss": 0.0651,
|
| 1127 |
+
"step": 1700
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 11.63,
|
| 1131 |
+
"learning_rate": 2.1134020618556702e-05,
|
| 1132 |
+
"loss": 0.042,
|
| 1133 |
+
"step": 1710
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 11.7,
|
| 1137 |
+
"learning_rate": 2.0962199312714777e-05,
|
| 1138 |
+
"loss": 0.0726,
|
| 1139 |
+
"step": 1720
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 11.77,
|
| 1143 |
+
"learning_rate": 2.0790378006872856e-05,
|
| 1144 |
+
"loss": 0.0792,
|
| 1145 |
+
"step": 1730
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 11.84,
|
| 1149 |
+
"learning_rate": 2.0618556701030927e-05,
|
| 1150 |
+
"loss": 0.0599,
|
| 1151 |
+
"step": 1740
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 11.9,
|
| 1155 |
+
"learning_rate": 2.0446735395189003e-05,
|
| 1156 |
+
"loss": 0.0702,
|
| 1157 |
+
"step": 1750
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 11.97,
|
| 1161 |
+
"learning_rate": 2.027491408934708e-05,
|
| 1162 |
+
"loss": 0.0684,
|
| 1163 |
+
"step": 1760
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 12.0,
|
| 1167 |
+
"eval_accuracy": 0.9879781420765027,
|
| 1168 |
+
"eval_loss": 0.03504302725195885,
|
| 1169 |
+
"eval_runtime": 29.3732,
|
| 1170 |
+
"eval_samples_per_second": 31.151,
|
| 1171 |
+
"eval_steps_per_second": 3.915,
|
| 1172 |
+
"step": 1764
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 12.04,
|
| 1176 |
+
"learning_rate": 2.0103092783505157e-05,
|
| 1177 |
+
"loss": 0.0412,
|
| 1178 |
+
"step": 1770
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 12.11,
|
| 1182 |
+
"learning_rate": 1.9931271477663232e-05,
|
| 1183 |
+
"loss": 0.0772,
|
| 1184 |
+
"step": 1780
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 12.18,
|
| 1188 |
+
"learning_rate": 1.9759450171821307e-05,
|
| 1189 |
+
"loss": 0.0542,
|
| 1190 |
+
"step": 1790
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 12.24,
|
| 1194 |
+
"learning_rate": 1.9587628865979382e-05,
|
| 1195 |
+
"loss": 0.0628,
|
| 1196 |
+
"step": 1800
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 12.31,
|
| 1200 |
+
"learning_rate": 1.9415807560137457e-05,
|
| 1201 |
+
"loss": 0.0542,
|
| 1202 |
+
"step": 1810
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 12.38,
|
| 1206 |
+
"learning_rate": 1.9243986254295536e-05,
|
| 1207 |
+
"loss": 0.061,
|
| 1208 |
+
"step": 1820
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 12.45,
|
| 1212 |
+
"learning_rate": 1.9072164948453608e-05,
|
| 1213 |
+
"loss": 0.0461,
|
| 1214 |
+
"step": 1830
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 12.52,
|
| 1218 |
+
"learning_rate": 1.8900343642611683e-05,
|
| 1219 |
+
"loss": 0.0551,
|
| 1220 |
+
"step": 1840
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 12.59,
|
| 1224 |
+
"learning_rate": 1.872852233676976e-05,
|
| 1225 |
+
"loss": 0.0588,
|
| 1226 |
+
"step": 1850
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 12.65,
|
| 1230 |
+
"learning_rate": 1.8556701030927837e-05,
|
| 1231 |
+
"loss": 0.0463,
|
| 1232 |
+
"step": 1860
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 12.72,
|
| 1236 |
+
"learning_rate": 1.8384879725085912e-05,
|
| 1237 |
+
"loss": 0.0401,
|
| 1238 |
+
"step": 1870
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 12.79,
|
| 1242 |
+
"learning_rate": 1.8213058419243987e-05,
|
| 1243 |
+
"loss": 0.0597,
|
| 1244 |
+
"step": 1880
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 12.86,
|
| 1248 |
+
"learning_rate": 1.8041237113402062e-05,
|
| 1249 |
+
"loss": 0.0449,
|
| 1250 |
+
"step": 1890
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 12.93,
|
| 1254 |
+
"learning_rate": 1.7869415807560138e-05,
|
| 1255 |
+
"loss": 0.0586,
|
| 1256 |
+
"step": 1900
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 12.99,
|
| 1260 |
+
"learning_rate": 1.7697594501718216e-05,
|
| 1261 |
+
"loss": 0.0358,
|
| 1262 |
+
"step": 1910
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 13.0,
|
| 1266 |
+
"eval_accuracy": 0.9890710382513661,
|
| 1267 |
+
"eval_loss": 0.040750984102487564,
|
| 1268 |
+
"eval_runtime": 29.934,
|
| 1269 |
+
"eval_samples_per_second": 30.567,
|
| 1270 |
+
"eval_steps_per_second": 3.842,
|
| 1271 |
+
"step": 1911
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 13.06,
|
| 1275 |
+
"learning_rate": 1.7525773195876288e-05,
|
| 1276 |
+
"loss": 0.0635,
|
| 1277 |
+
"step": 1920
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 13.13,
|
| 1281 |
+
"learning_rate": 1.7353951890034363e-05,
|
| 1282 |
+
"loss": 0.0545,
|
| 1283 |
+
"step": 1930
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 13.2,
|
| 1287 |
+
"learning_rate": 1.7182130584192442e-05,
|
| 1288 |
+
"loss": 0.0378,
|
| 1289 |
+
"step": 1940
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 13.27,
|
| 1293 |
+
"learning_rate": 1.7010309278350517e-05,
|
| 1294 |
+
"loss": 0.0744,
|
| 1295 |
+
"step": 1950
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 13.33,
|
| 1299 |
+
"learning_rate": 1.6838487972508592e-05,
|
| 1300 |
+
"loss": 0.0406,
|
| 1301 |
+
"step": 1960
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 13.4,
|
| 1305 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 1306 |
+
"loss": 0.0506,
|
| 1307 |
+
"step": 1970
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 13.47,
|
| 1311 |
+
"learning_rate": 1.6494845360824743e-05,
|
| 1312 |
+
"loss": 0.0676,
|
| 1313 |
+
"step": 1980
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 13.54,
|
| 1317 |
+
"learning_rate": 1.6323024054982818e-05,
|
| 1318 |
+
"loss": 0.0553,
|
| 1319 |
+
"step": 1990
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 13.61,
|
| 1323 |
+
"learning_rate": 1.6151202749140896e-05,
|
| 1324 |
+
"loss": 0.0594,
|
| 1325 |
+
"step": 2000
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 13.67,
|
| 1329 |
+
"learning_rate": 1.5979381443298968e-05,
|
| 1330 |
+
"loss": 0.0378,
|
| 1331 |
+
"step": 2010
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 13.74,
|
| 1335 |
+
"learning_rate": 1.5807560137457044e-05,
|
| 1336 |
+
"loss": 0.0576,
|
| 1337 |
+
"step": 2020
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 13.81,
|
| 1341 |
+
"learning_rate": 1.5635738831615122e-05,
|
| 1342 |
+
"loss": 0.0621,
|
| 1343 |
+
"step": 2030
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 13.88,
|
| 1347 |
+
"learning_rate": 1.5463917525773197e-05,
|
| 1348 |
+
"loss": 0.0611,
|
| 1349 |
+
"step": 2040
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 13.95,
|
| 1353 |
+
"learning_rate": 1.5292096219931273e-05,
|
| 1354 |
+
"loss": 0.0548,
|
| 1355 |
+
"step": 2050
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 14.0,
|
| 1359 |
+
"eval_accuracy": 0.9901639344262295,
|
| 1360 |
+
"eval_loss": 0.038173872977495193,
|
| 1361 |
+
"eval_runtime": 30.2908,
|
| 1362 |
+
"eval_samples_per_second": 30.207,
|
| 1363 |
+
"eval_steps_per_second": 3.797,
|
| 1364 |
+
"step": 2058
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 14.01,
|
| 1368 |
+
"learning_rate": 1.5120274914089346e-05,
|
| 1369 |
+
"loss": 0.0702,
|
| 1370 |
+
"step": 2060
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 14.08,
|
| 1374 |
+
"learning_rate": 1.4948453608247423e-05,
|
| 1375 |
+
"loss": 0.0667,
|
| 1376 |
+
"step": 2070
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 14.15,
|
| 1380 |
+
"learning_rate": 1.47766323024055e-05,
|
| 1381 |
+
"loss": 0.0563,
|
| 1382 |
+
"step": 2080
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 14.22,
|
| 1386 |
+
"learning_rate": 1.4604810996563575e-05,
|
| 1387 |
+
"loss": 0.0601,
|
| 1388 |
+
"step": 2090
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 14.29,
|
| 1392 |
+
"learning_rate": 1.4432989690721649e-05,
|
| 1393 |
+
"loss": 0.0535,
|
| 1394 |
+
"step": 2100
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 14.35,
|
| 1398 |
+
"learning_rate": 1.4261168384879725e-05,
|
| 1399 |
+
"loss": 0.0753,
|
| 1400 |
+
"step": 2110
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 14.42,
|
| 1404 |
+
"learning_rate": 1.40893470790378e-05,
|
| 1405 |
+
"loss": 0.0677,
|
| 1406 |
+
"step": 2120
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 14.49,
|
| 1410 |
+
"learning_rate": 1.3917525773195878e-05,
|
| 1411 |
+
"loss": 0.0451,
|
| 1412 |
+
"step": 2130
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 14.56,
|
| 1416 |
+
"learning_rate": 1.3745704467353954e-05,
|
| 1417 |
+
"loss": 0.0631,
|
| 1418 |
+
"step": 2140
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 14.63,
|
| 1422 |
+
"learning_rate": 1.3573883161512026e-05,
|
| 1423 |
+
"loss": 0.0177,
|
| 1424 |
+
"step": 2150
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 14.69,
|
| 1428 |
+
"learning_rate": 1.3402061855670103e-05,
|
| 1429 |
+
"loss": 0.0559,
|
| 1430 |
+
"step": 2160
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 14.76,
|
| 1434 |
+
"learning_rate": 1.323024054982818e-05,
|
| 1435 |
+
"loss": 0.0352,
|
| 1436 |
+
"step": 2170
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 14.83,
|
| 1440 |
+
"learning_rate": 1.3058419243986255e-05,
|
| 1441 |
+
"loss": 0.0398,
|
| 1442 |
+
"step": 2180
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 14.9,
|
| 1446 |
+
"learning_rate": 1.2886597938144329e-05,
|
| 1447 |
+
"loss": 0.0134,
|
| 1448 |
+
"step": 2190
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 14.97,
|
| 1452 |
+
"learning_rate": 1.2714776632302406e-05,
|
| 1453 |
+
"loss": 0.0611,
|
| 1454 |
+
"step": 2200
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 15.0,
|
| 1458 |
+
"eval_accuracy": 0.9923497267759562,
|
| 1459 |
+
"eval_loss": 0.03305086866021156,
|
| 1460 |
+
"eval_runtime": 29.8991,
|
| 1461 |
+
"eval_samples_per_second": 30.603,
|
| 1462 |
+
"eval_steps_per_second": 3.846,
|
| 1463 |
+
"step": 2205
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 15.03,
|
| 1467 |
+
"learning_rate": 1.2542955326460481e-05,
|
| 1468 |
+
"loss": 0.0722,
|
| 1469 |
+
"step": 2210
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 15.1,
|
| 1473 |
+
"learning_rate": 1.2371134020618558e-05,
|
| 1474 |
+
"loss": 0.0704,
|
| 1475 |
+
"step": 2220
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 15.17,
|
| 1479 |
+
"learning_rate": 1.2199312714776633e-05,
|
| 1480 |
+
"loss": 0.0528,
|
| 1481 |
+
"step": 2230
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 15.24,
|
| 1485 |
+
"learning_rate": 1.2027491408934708e-05,
|
| 1486 |
+
"loss": 0.043,
|
| 1487 |
+
"step": 2240
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 15.31,
|
| 1491 |
+
"learning_rate": 1.1855670103092783e-05,
|
| 1492 |
+
"loss": 0.0529,
|
| 1493 |
+
"step": 2250
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 15.37,
|
| 1497 |
+
"learning_rate": 1.168384879725086e-05,
|
| 1498 |
+
"loss": 0.0297,
|
| 1499 |
+
"step": 2260
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 15.44,
|
| 1503 |
+
"learning_rate": 1.1512027491408934e-05,
|
| 1504 |
+
"loss": 0.0522,
|
| 1505 |
+
"step": 2270
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 15.51,
|
| 1509 |
+
"learning_rate": 1.134020618556701e-05,
|
| 1510 |
+
"loss": 0.0401,
|
| 1511 |
+
"step": 2280
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 15.58,
|
| 1515 |
+
"learning_rate": 1.1168384879725088e-05,
|
| 1516 |
+
"loss": 0.0455,
|
| 1517 |
+
"step": 2290
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 15.65,
|
| 1521 |
+
"learning_rate": 1.0996563573883161e-05,
|
| 1522 |
+
"loss": 0.0303,
|
| 1523 |
+
"step": 2300
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 15.71,
|
| 1527 |
+
"learning_rate": 1.0824742268041238e-05,
|
| 1528 |
+
"loss": 0.0263,
|
| 1529 |
+
"step": 2310
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 15.78,
|
| 1533 |
+
"learning_rate": 1.0652920962199313e-05,
|
| 1534 |
+
"loss": 0.1067,
|
| 1535 |
+
"step": 2320
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 15.85,
|
| 1539 |
+
"learning_rate": 1.0481099656357389e-05,
|
| 1540 |
+
"loss": 0.049,
|
| 1541 |
+
"step": 2330
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 15.92,
|
| 1545 |
+
"learning_rate": 1.0309278350515464e-05,
|
| 1546 |
+
"loss": 0.044,
|
| 1547 |
+
"step": 2340
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 15.99,
|
| 1551 |
+
"learning_rate": 1.013745704467354e-05,
|
| 1552 |
+
"loss": 0.0231,
|
| 1553 |
+
"step": 2350
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 16.0,
|
| 1557 |
+
"eval_accuracy": 0.994535519125683,
|
| 1558 |
+
"eval_loss": 0.035519640892744064,
|
| 1559 |
+
"eval_runtime": 30.0972,
|
| 1560 |
+
"eval_samples_per_second": 30.401,
|
| 1561 |
+
"eval_steps_per_second": 3.821,
|
| 1562 |
+
"step": 2352
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 16.05,
|
| 1566 |
+
"learning_rate": 9.965635738831616e-06,
|
| 1567 |
+
"loss": 0.0456,
|
| 1568 |
+
"step": 2360
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 16.12,
|
| 1572 |
+
"learning_rate": 9.793814432989691e-06,
|
| 1573 |
+
"loss": 0.0517,
|
| 1574 |
+
"step": 2370
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 16.19,
|
| 1578 |
+
"learning_rate": 9.621993127147768e-06,
|
| 1579 |
+
"loss": 0.0601,
|
| 1580 |
+
"step": 2380
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 16.26,
|
| 1584 |
+
"learning_rate": 9.450171821305841e-06,
|
| 1585 |
+
"loss": 0.0695,
|
| 1586 |
+
"step": 2390
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 16.33,
|
| 1590 |
+
"learning_rate": 9.278350515463918e-06,
|
| 1591 |
+
"loss": 0.0709,
|
| 1592 |
+
"step": 2400
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 16.39,
|
| 1596 |
+
"learning_rate": 9.106529209621994e-06,
|
| 1597 |
+
"loss": 0.0691,
|
| 1598 |
+
"step": 2410
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 16.46,
|
| 1602 |
+
"learning_rate": 8.934707903780069e-06,
|
| 1603 |
+
"loss": 0.0514,
|
| 1604 |
+
"step": 2420
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 16.53,
|
| 1608 |
+
"learning_rate": 8.762886597938144e-06,
|
| 1609 |
+
"loss": 0.0507,
|
| 1610 |
+
"step": 2430
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 16.6,
|
| 1614 |
+
"learning_rate": 8.591065292096221e-06,
|
| 1615 |
+
"loss": 0.0497,
|
| 1616 |
+
"step": 2440
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 16.67,
|
| 1620 |
+
"learning_rate": 8.419243986254296e-06,
|
| 1621 |
+
"loss": 0.0539,
|
| 1622 |
+
"step": 2450
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 16.73,
|
| 1626 |
+
"learning_rate": 8.247422680412371e-06,
|
| 1627 |
+
"loss": 0.0144,
|
| 1628 |
+
"step": 2460
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 16.8,
|
| 1632 |
+
"learning_rate": 8.075601374570448e-06,
|
| 1633 |
+
"loss": 0.039,
|
| 1634 |
+
"step": 2470
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 16.87,
|
| 1638 |
+
"learning_rate": 7.903780068728522e-06,
|
| 1639 |
+
"loss": 0.0495,
|
| 1640 |
+
"step": 2480
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 16.94,
|
| 1644 |
+
"learning_rate": 7.731958762886599e-06,
|
| 1645 |
+
"loss": 0.046,
|
| 1646 |
+
"step": 2490
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 17.0,
|
| 1650 |
+
"eval_accuracy": 0.9934426229508196,
|
| 1651 |
+
"eval_loss": 0.03209972009062767,
|
| 1652 |
+
"eval_runtime": 30.1657,
|
| 1653 |
+
"eval_samples_per_second": 30.332,
|
| 1654 |
+
"eval_steps_per_second": 3.812,
|
| 1655 |
+
"step": 2499
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 17.01,
|
| 1659 |
+
"learning_rate": 7.560137457044673e-06,
|
| 1660 |
+
"loss": 0.0709,
|
| 1661 |
+
"step": 2500
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 17.07,
|
| 1665 |
+
"learning_rate": 7.38831615120275e-06,
|
| 1666 |
+
"loss": 0.0541,
|
| 1667 |
+
"step": 2510
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 17.14,
|
| 1671 |
+
"learning_rate": 7.216494845360824e-06,
|
| 1672 |
+
"loss": 0.0432,
|
| 1673 |
+
"step": 2520
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 17.21,
|
| 1677 |
+
"learning_rate": 7.0446735395189e-06,
|
| 1678 |
+
"loss": 0.0525,
|
| 1679 |
+
"step": 2530
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"epoch": 17.28,
|
| 1683 |
+
"learning_rate": 6.872852233676977e-06,
|
| 1684 |
+
"loss": 0.0454,
|
| 1685 |
+
"step": 2540
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 17.35,
|
| 1689 |
+
"learning_rate": 6.701030927835052e-06,
|
| 1690 |
+
"loss": 0.0283,
|
| 1691 |
+
"step": 2550
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 17.41,
|
| 1695 |
+
"learning_rate": 6.529209621993128e-06,
|
| 1696 |
+
"loss": 0.0578,
|
| 1697 |
+
"step": 2560
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 17.48,
|
| 1701 |
+
"learning_rate": 6.357388316151203e-06,
|
| 1702 |
+
"loss": 0.0652,
|
| 1703 |
+
"step": 2570
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 17.55,
|
| 1707 |
+
"learning_rate": 6.185567010309279e-06,
|
| 1708 |
+
"loss": 0.0418,
|
| 1709 |
+
"step": 2580
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 17.62,
|
| 1713 |
+
"learning_rate": 6.013745704467354e-06,
|
| 1714 |
+
"loss": 0.0795,
|
| 1715 |
+
"step": 2590
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 17.69,
|
| 1719 |
+
"learning_rate": 5.84192439862543e-06,
|
| 1720 |
+
"loss": 0.0294,
|
| 1721 |
+
"step": 2600
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 17.76,
|
| 1725 |
+
"learning_rate": 5.670103092783505e-06,
|
| 1726 |
+
"loss": 0.0702,
|
| 1727 |
+
"step": 2610
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 17.82,
|
| 1731 |
+
"learning_rate": 5.498281786941581e-06,
|
| 1732 |
+
"loss": 0.0267,
|
| 1733 |
+
"step": 2620
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 17.89,
|
| 1737 |
+
"learning_rate": 5.326460481099657e-06,
|
| 1738 |
+
"loss": 0.0397,
|
| 1739 |
+
"step": 2630
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 17.96,
|
| 1743 |
+
"learning_rate": 5.154639175257732e-06,
|
| 1744 |
+
"loss": 0.0648,
|
| 1745 |
+
"step": 2640
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 18.0,
|
| 1749 |
+
"eval_accuracy": 0.9923497267759562,
|
| 1750 |
+
"eval_loss": 0.0327494814991951,
|
| 1751 |
+
"eval_runtime": 30.289,
|
| 1752 |
+
"eval_samples_per_second": 30.209,
|
| 1753 |
+
"eval_steps_per_second": 3.797,
|
| 1754 |
+
"step": 2646
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 18.03,
|
| 1758 |
+
"learning_rate": 4.982817869415808e-06,
|
| 1759 |
+
"loss": 0.033,
|
| 1760 |
+
"step": 2650
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 18.1,
|
| 1764 |
+
"learning_rate": 4.810996563573884e-06,
|
| 1765 |
+
"loss": 0.0447,
|
| 1766 |
+
"step": 2660
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 18.16,
|
| 1770 |
+
"learning_rate": 4.639175257731959e-06,
|
| 1771 |
+
"loss": 0.0324,
|
| 1772 |
+
"step": 2670
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 18.23,
|
| 1776 |
+
"learning_rate": 4.467353951890034e-06,
|
| 1777 |
+
"loss": 0.0382,
|
| 1778 |
+
"step": 2680
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 18.3,
|
| 1782 |
+
"learning_rate": 4.2955326460481105e-06,
|
| 1783 |
+
"loss": 0.0628,
|
| 1784 |
+
"step": 2690
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 18.37,
|
| 1788 |
+
"learning_rate": 4.123711340206186e-06,
|
| 1789 |
+
"loss": 0.0808,
|
| 1790 |
+
"step": 2700
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 18.44,
|
| 1794 |
+
"learning_rate": 3.951890034364261e-06,
|
| 1795 |
+
"loss": 0.0463,
|
| 1796 |
+
"step": 2710
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 18.5,
|
| 1800 |
+
"learning_rate": 3.7800687285223365e-06,
|
| 1801 |
+
"loss": 0.0369,
|
| 1802 |
+
"step": 2720
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 18.57,
|
| 1806 |
+
"learning_rate": 3.608247422680412e-06,
|
| 1807 |
+
"loss": 0.0273,
|
| 1808 |
+
"step": 2730
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 18.64,
|
| 1812 |
+
"learning_rate": 3.4364261168384886e-06,
|
| 1813 |
+
"loss": 0.0327,
|
| 1814 |
+
"step": 2740
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 18.71,
|
| 1818 |
+
"learning_rate": 3.264604810996564e-06,
|
| 1819 |
+
"loss": 0.0331,
|
| 1820 |
+
"step": 2750
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 18.78,
|
| 1824 |
+
"learning_rate": 3.0927835051546395e-06,
|
| 1825 |
+
"loss": 0.0461,
|
| 1826 |
+
"step": 2760
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 18.84,
|
| 1830 |
+
"learning_rate": 2.920962199312715e-06,
|
| 1831 |
+
"loss": 0.0525,
|
| 1832 |
+
"step": 2770
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 18.91,
|
| 1836 |
+
"learning_rate": 2.7491408934707903e-06,
|
| 1837 |
+
"loss": 0.053,
|
| 1838 |
+
"step": 2780
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 18.98,
|
| 1842 |
+
"learning_rate": 2.577319587628866e-06,
|
| 1843 |
+
"loss": 0.0565,
|
| 1844 |
+
"step": 2790
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 19.0,
|
| 1848 |
+
"eval_accuracy": 0.9923497267759562,
|
| 1849 |
+
"eval_loss": 0.03197157755494118,
|
| 1850 |
+
"eval_runtime": 30.3146,
|
| 1851 |
+
"eval_samples_per_second": 30.184,
|
| 1852 |
+
"eval_steps_per_second": 3.794,
|
| 1853 |
+
"step": 2793
|
| 1854 |
+
},
|
| 1855 |
+
{
|
| 1856 |
+
"epoch": 19.05,
|
| 1857 |
+
"learning_rate": 2.405498281786942e-06,
|
| 1858 |
+
"loss": 0.0308,
|
| 1859 |
+
"step": 2800
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 19.12,
|
| 1863 |
+
"learning_rate": 2.233676975945017e-06,
|
| 1864 |
+
"loss": 0.0538,
|
| 1865 |
+
"step": 2810
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 19.18,
|
| 1869 |
+
"learning_rate": 2.061855670103093e-06,
|
| 1870 |
+
"loss": 0.0483,
|
| 1871 |
+
"step": 2820
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 19.25,
|
| 1875 |
+
"learning_rate": 1.8900343642611683e-06,
|
| 1876 |
+
"loss": 0.0546,
|
| 1877 |
+
"step": 2830
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 19.32,
|
| 1881 |
+
"learning_rate": 1.7182130584192443e-06,
|
| 1882 |
+
"loss": 0.0346,
|
| 1883 |
+
"step": 2840
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 19.39,
|
| 1887 |
+
"learning_rate": 1.5463917525773197e-06,
|
| 1888 |
+
"loss": 0.0423,
|
| 1889 |
+
"step": 2850
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 19.46,
|
| 1893 |
+
"learning_rate": 1.3745704467353952e-06,
|
| 1894 |
+
"loss": 0.0241,
|
| 1895 |
+
"step": 2860
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 19.52,
|
| 1899 |
+
"learning_rate": 1.202749140893471e-06,
|
| 1900 |
+
"loss": 0.0426,
|
| 1901 |
+
"step": 2870
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 19.59,
|
| 1905 |
+
"learning_rate": 1.0309278350515464e-06,
|
| 1906 |
+
"loss": 0.0409,
|
| 1907 |
+
"step": 2880
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 19.66,
|
| 1911 |
+
"learning_rate": 8.591065292096222e-07,
|
| 1912 |
+
"loss": 0.0323,
|
| 1913 |
+
"step": 2890
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 19.73,
|
| 1917 |
+
"learning_rate": 6.872852233676976e-07,
|
| 1918 |
+
"loss": 0.0261,
|
| 1919 |
+
"step": 2900
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 19.8,
|
| 1923 |
+
"learning_rate": 5.154639175257732e-07,
|
| 1924 |
+
"loss": 0.04,
|
| 1925 |
+
"step": 2910
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 19.86,
|
| 1929 |
+
"learning_rate": 3.436426116838488e-07,
|
| 1930 |
+
"loss": 0.0575,
|
| 1931 |
+
"step": 2920
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 19.93,
|
| 1935 |
+
"learning_rate": 1.718213058419244e-07,
|
| 1936 |
+
"loss": 0.0495,
|
| 1937 |
+
"step": 2930
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 20.0,
|
| 1941 |
+
"learning_rate": 0.0,
|
| 1942 |
+
"loss": 0.0413,
|
| 1943 |
+
"step": 2940
|
| 1944 |
+
},
|
| 1945 |
+
{
|
| 1946 |
+
"epoch": 20.0,
|
| 1947 |
+
"eval_accuracy": 0.9923497267759562,
|
| 1948 |
+
"eval_loss": 0.032683759927749634,
|
| 1949 |
+
"eval_runtime": 29.7008,
|
| 1950 |
+
"eval_samples_per_second": 30.807,
|
| 1951 |
+
"eval_steps_per_second": 3.872,
|
| 1952 |
+
"step": 2940
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 20.0,
|
| 1956 |
+
"step": 2940,
|
| 1957 |
+
"total_flos": 4.089918706717778e+18,
|
| 1958 |
+
"train_loss": 0.08860804434029423,
|
| 1959 |
+
"train_runtime": 6732.3279,
|
| 1960 |
+
"train_samples_per_second": 24.44,
|
| 1961 |
+
"train_steps_per_second": 0.437
|
| 1962 |
+
}
|
| 1963 |
+
],
|
| 1964 |
+
"max_steps": 2940,
|
| 1965 |
+
"num_train_epochs": 20,
|
| 1966 |
+
"total_flos": 4.089918706717778e+18,
|
| 1967 |
+
"trial_name": null,
|
| 1968 |
+
"trial_params": null
|
| 1969 |
+
}
|