End of training
Browse files- all_results.json +13 -0
- eval_results.json +8 -0
- runs/Jul16_09-38-00_4bb7fc5f194e/events.out.tfevents.1689509130.4bb7fc5f194e.338.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +1756 -0
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 7.0,
|
| 3 |
+
"eval_accuracy": 0.9943422913719944,
|
| 4 |
+
"eval_loss": 0.022870095446705818,
|
| 5 |
+
"eval_runtime": 102.4976,
|
| 6 |
+
"eval_samples_per_second": 20.693,
|
| 7 |
+
"eval_steps_per_second": 2.595,
|
| 8 |
+
"total_flos": 1.0256650917539217e+19,
|
| 9 |
+
"train_loss": 0.15325962240578392,
|
| 10 |
+
"train_runtime": 8636.3959,
|
| 11 |
+
"train_samples_per_second": 5.157,
|
| 12 |
+
"train_steps_per_second": 0.161
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 7.0,
|
| 3 |
+
"eval_accuracy": 0.9943422913719944,
|
| 4 |
+
"eval_loss": 0.022870095446705818,
|
| 5 |
+
"eval_runtime": 102.4976,
|
| 6 |
+
"eval_samples_per_second": 20.693,
|
| 7 |
+
"eval_steps_per_second": 2.595
|
| 8 |
+
}
|
runs/Jul16_09-38-00_4bb7fc5f194e/events.out.tfevents.1689509130.4bb7fc5f194e.338.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:af0454a953e1d36b0829e1aa197d7fdb92d1c2163c7a92ae31c3fa61dc5df53b
|
| 3 |
+
size 411
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 7.0,
|
| 3 |
+
"total_flos": 1.0256650917539217e+19,
|
| 4 |
+
"train_loss": 0.15325962240578392,
|
| 5 |
+
"train_runtime": 8636.3959,
|
| 6 |
+
"train_samples_per_second": 5.157,
|
| 7 |
+
"train_steps_per_second": 0.161
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1756 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9943422913719944,
|
| 3 |
+
"best_model_checkpoint": "swinv2-large-patch4-window12to16-192to256-22kto1k-ft-finetuned-BreastCancer-BreakHis-AH-60-20-20/checkpoint-1194",
|
| 4 |
+
"epoch": 7.0,
|
| 5 |
+
"global_step": 1393,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.03,
|
| 12 |
+
"learning_rate": 3.586800573888092e-07,
|
| 13 |
+
"loss": 0.7746,
|
| 14 |
+
"step": 5
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.05,
|
| 18 |
+
"learning_rate": 7.173601147776184e-07,
|
| 19 |
+
"loss": 0.739,
|
| 20 |
+
"step": 10
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.08,
|
| 24 |
+
"learning_rate": 1.0760401721664278e-06,
|
| 25 |
+
"loss": 0.7228,
|
| 26 |
+
"step": 15
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.1,
|
| 30 |
+
"learning_rate": 1.4347202295552369e-06,
|
| 31 |
+
"loss": 0.7082,
|
| 32 |
+
"step": 20
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.13,
|
| 36 |
+
"learning_rate": 1.793400286944046e-06,
|
| 37 |
+
"loss": 0.6802,
|
| 38 |
+
"step": 25
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.15,
|
| 42 |
+
"learning_rate": 2.1520803443328555e-06,
|
| 43 |
+
"loss": 0.6662,
|
| 44 |
+
"step": 30
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.18,
|
| 48 |
+
"learning_rate": 2.5107604017216644e-06,
|
| 49 |
+
"loss": 0.631,
|
| 50 |
+
"step": 35
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.2,
|
| 54 |
+
"learning_rate": 2.8694404591104738e-06,
|
| 55 |
+
"loss": 0.6168,
|
| 56 |
+
"step": 40
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.23,
|
| 60 |
+
"learning_rate": 3.2281205164992827e-06,
|
| 61 |
+
"loss": 0.5649,
|
| 62 |
+
"step": 45
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.25,
|
| 66 |
+
"learning_rate": 3.586800573888092e-06,
|
| 67 |
+
"loss": 0.5661,
|
| 68 |
+
"step": 50
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.28,
|
| 72 |
+
"learning_rate": 3.945480631276901e-06,
|
| 73 |
+
"loss": 0.4748,
|
| 74 |
+
"step": 55
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.3,
|
| 78 |
+
"learning_rate": 4.304160688665711e-06,
|
| 79 |
+
"loss": 0.4842,
|
| 80 |
+
"step": 60
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.33,
|
| 84 |
+
"learning_rate": 4.66284074605452e-06,
|
| 85 |
+
"loss": 0.4239,
|
| 86 |
+
"step": 65
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.35,
|
| 90 |
+
"learning_rate": 5.021520803443329e-06,
|
| 91 |
+
"loss": 0.4256,
|
| 92 |
+
"step": 70
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.38,
|
| 96 |
+
"learning_rate": 5.380200860832138e-06,
|
| 97 |
+
"loss": 0.3675,
|
| 98 |
+
"step": 75
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.4,
|
| 102 |
+
"learning_rate": 5.7388809182209475e-06,
|
| 103 |
+
"loss": 0.3646,
|
| 104 |
+
"step": 80
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.43,
|
| 108 |
+
"learning_rate": 6.0975609756097564e-06,
|
| 109 |
+
"loss": 0.3939,
|
| 110 |
+
"step": 85
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.45,
|
| 114 |
+
"learning_rate": 6.456241032998565e-06,
|
| 115 |
+
"loss": 0.3783,
|
| 116 |
+
"step": 90
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.48,
|
| 120 |
+
"learning_rate": 6.814921090387374e-06,
|
| 121 |
+
"loss": 0.4425,
|
| 122 |
+
"step": 95
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.5,
|
| 126 |
+
"learning_rate": 7.173601147776184e-06,
|
| 127 |
+
"loss": 0.348,
|
| 128 |
+
"step": 100
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.53,
|
| 132 |
+
"learning_rate": 7.532281205164993e-06,
|
| 133 |
+
"loss": 0.3337,
|
| 134 |
+
"step": 105
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.55,
|
| 138 |
+
"learning_rate": 7.890961262553803e-06,
|
| 139 |
+
"loss": 0.2808,
|
| 140 |
+
"step": 110
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.58,
|
| 144 |
+
"learning_rate": 8.249641319942612e-06,
|
| 145 |
+
"loss": 0.3154,
|
| 146 |
+
"step": 115
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.6,
|
| 150 |
+
"learning_rate": 8.608321377331422e-06,
|
| 151 |
+
"loss": 0.2637,
|
| 152 |
+
"step": 120
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.63,
|
| 156 |
+
"learning_rate": 8.967001434720231e-06,
|
| 157 |
+
"loss": 0.2857,
|
| 158 |
+
"step": 125
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.65,
|
| 162 |
+
"learning_rate": 9.32568149210904e-06,
|
| 163 |
+
"loss": 0.2633,
|
| 164 |
+
"step": 130
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.68,
|
| 168 |
+
"learning_rate": 9.684361549497849e-06,
|
| 169 |
+
"loss": 0.3059,
|
| 170 |
+
"step": 135
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.7,
|
| 174 |
+
"learning_rate": 1.0043041606886658e-05,
|
| 175 |
+
"loss": 0.2524,
|
| 176 |
+
"step": 140
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.73,
|
| 180 |
+
"learning_rate": 1.0401721664275467e-05,
|
| 181 |
+
"loss": 0.2154,
|
| 182 |
+
"step": 145
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.75,
|
| 186 |
+
"learning_rate": 1.0760401721664276e-05,
|
| 187 |
+
"loss": 0.3399,
|
| 188 |
+
"step": 150
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.78,
|
| 192 |
+
"learning_rate": 1.1119081779053084e-05,
|
| 193 |
+
"loss": 0.2305,
|
| 194 |
+
"step": 155
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.8,
|
| 198 |
+
"learning_rate": 1.1477761836441895e-05,
|
| 199 |
+
"loss": 0.2263,
|
| 200 |
+
"step": 160
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.83,
|
| 204 |
+
"learning_rate": 1.1836441893830704e-05,
|
| 205 |
+
"loss": 0.1931,
|
| 206 |
+
"step": 165
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.85,
|
| 210 |
+
"learning_rate": 1.2195121951219513e-05,
|
| 211 |
+
"loss": 0.2317,
|
| 212 |
+
"step": 170
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.88,
|
| 216 |
+
"learning_rate": 1.2553802008608323e-05,
|
| 217 |
+
"loss": 0.2118,
|
| 218 |
+
"step": 175
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.9,
|
| 222 |
+
"learning_rate": 1.291248206599713e-05,
|
| 223 |
+
"loss": 0.1408,
|
| 224 |
+
"step": 180
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.93,
|
| 228 |
+
"learning_rate": 1.3271162123385941e-05,
|
| 229 |
+
"loss": 0.24,
|
| 230 |
+
"step": 185
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.95,
|
| 234 |
+
"learning_rate": 1.3629842180774748e-05,
|
| 235 |
+
"loss": 0.2579,
|
| 236 |
+
"step": 190
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.98,
|
| 240 |
+
"learning_rate": 1.3988522238163559e-05,
|
| 241 |
+
"loss": 0.2053,
|
| 242 |
+
"step": 195
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.0,
|
| 246 |
+
"eval_accuracy": 0.9495520980669495,
|
| 247 |
+
"eval_loss": 0.1227400153875351,
|
| 248 |
+
"eval_runtime": 524.4475,
|
| 249 |
+
"eval_samples_per_second": 4.044,
|
| 250 |
+
"eval_steps_per_second": 0.507,
|
| 251 |
+
"step": 199
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.01,
|
| 255 |
+
"learning_rate": 1.4347202295552368e-05,
|
| 256 |
+
"loss": 0.2,
|
| 257 |
+
"step": 200
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.03,
|
| 261 |
+
"learning_rate": 1.4705882352941177e-05,
|
| 262 |
+
"loss": 0.2002,
|
| 263 |
+
"step": 205
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.06,
|
| 267 |
+
"learning_rate": 1.5064562410329986e-05,
|
| 268 |
+
"loss": 0.2141,
|
| 269 |
+
"step": 210
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.08,
|
| 273 |
+
"learning_rate": 1.5423242467718796e-05,
|
| 274 |
+
"loss": 0.2358,
|
| 275 |
+
"step": 215
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.11,
|
| 279 |
+
"learning_rate": 1.5781922525107605e-05,
|
| 280 |
+
"loss": 0.1788,
|
| 281 |
+
"step": 220
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 1.13,
|
| 285 |
+
"learning_rate": 1.6140602582496414e-05,
|
| 286 |
+
"loss": 0.0829,
|
| 287 |
+
"step": 225
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 1.16,
|
| 291 |
+
"learning_rate": 1.6499282639885223e-05,
|
| 292 |
+
"loss": 0.205,
|
| 293 |
+
"step": 230
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 1.18,
|
| 297 |
+
"learning_rate": 1.6857962697274032e-05,
|
| 298 |
+
"loss": 0.1433,
|
| 299 |
+
"step": 235
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 1.21,
|
| 303 |
+
"learning_rate": 1.7216642754662844e-05,
|
| 304 |
+
"loss": 0.162,
|
| 305 |
+
"step": 240
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 1.23,
|
| 309 |
+
"learning_rate": 1.757532281205165e-05,
|
| 310 |
+
"loss": 0.2635,
|
| 311 |
+
"step": 245
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 1.26,
|
| 315 |
+
"learning_rate": 1.7934002869440462e-05,
|
| 316 |
+
"loss": 0.2703,
|
| 317 |
+
"step": 250
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 1.28,
|
| 321 |
+
"learning_rate": 1.8292682926829268e-05,
|
| 322 |
+
"loss": 0.239,
|
| 323 |
+
"step": 255
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 1.31,
|
| 327 |
+
"learning_rate": 1.865136298421808e-05,
|
| 328 |
+
"loss": 0.2087,
|
| 329 |
+
"step": 260
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 1.33,
|
| 333 |
+
"learning_rate": 1.9010043041606885e-05,
|
| 334 |
+
"loss": 0.206,
|
| 335 |
+
"step": 265
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 1.36,
|
| 339 |
+
"learning_rate": 1.9368723098995698e-05,
|
| 340 |
+
"loss": 0.1916,
|
| 341 |
+
"step": 270
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 1.38,
|
| 345 |
+
"learning_rate": 1.9727403156384507e-05,
|
| 346 |
+
"loss": 0.2102,
|
| 347 |
+
"step": 275
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 1.41,
|
| 351 |
+
"learning_rate": 2.0086083213773316e-05,
|
| 352 |
+
"loss": 0.1304,
|
| 353 |
+
"step": 280
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.43,
|
| 357 |
+
"learning_rate": 2.0444763271162124e-05,
|
| 358 |
+
"loss": 0.1707,
|
| 359 |
+
"step": 285
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 1.46,
|
| 363 |
+
"learning_rate": 2.0803443328550933e-05,
|
| 364 |
+
"loss": 0.0719,
|
| 365 |
+
"step": 290
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 1.48,
|
| 369 |
+
"learning_rate": 2.1162123385939742e-05,
|
| 370 |
+
"loss": 0.22,
|
| 371 |
+
"step": 295
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 1.51,
|
| 375 |
+
"learning_rate": 2.152080344332855e-05,
|
| 376 |
+
"loss": 0.2759,
|
| 377 |
+
"step": 300
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 1.53,
|
| 381 |
+
"learning_rate": 2.187948350071736e-05,
|
| 382 |
+
"loss": 0.2507,
|
| 383 |
+
"step": 305
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 1.56,
|
| 387 |
+
"learning_rate": 2.223816355810617e-05,
|
| 388 |
+
"loss": 0.2151,
|
| 389 |
+
"step": 310
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 1.58,
|
| 393 |
+
"learning_rate": 2.259684361549498e-05,
|
| 394 |
+
"loss": 0.1902,
|
| 395 |
+
"step": 315
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 1.61,
|
| 399 |
+
"learning_rate": 2.295552367288379e-05,
|
| 400 |
+
"loss": 0.3037,
|
| 401 |
+
"step": 320
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.63,
|
| 405 |
+
"learning_rate": 2.33142037302726e-05,
|
| 406 |
+
"loss": 0.2278,
|
| 407 |
+
"step": 325
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 1.66,
|
| 411 |
+
"learning_rate": 2.3672883787661408e-05,
|
| 412 |
+
"loss": 0.2791,
|
| 413 |
+
"step": 330
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.68,
|
| 417 |
+
"learning_rate": 2.4031563845050217e-05,
|
| 418 |
+
"loss": 0.1363,
|
| 419 |
+
"step": 335
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 1.71,
|
| 423 |
+
"learning_rate": 2.4390243902439026e-05,
|
| 424 |
+
"loss": 0.2639,
|
| 425 |
+
"step": 340
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 1.73,
|
| 429 |
+
"learning_rate": 2.4748923959827835e-05,
|
| 430 |
+
"loss": 0.108,
|
| 431 |
+
"step": 345
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 1.76,
|
| 435 |
+
"learning_rate": 2.5107604017216647e-05,
|
| 436 |
+
"loss": 0.1188,
|
| 437 |
+
"step": 350
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 1.78,
|
| 441 |
+
"learning_rate": 2.5466284074605452e-05,
|
| 442 |
+
"loss": 0.1908,
|
| 443 |
+
"step": 355
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 1.81,
|
| 447 |
+
"learning_rate": 2.582496413199426e-05,
|
| 448 |
+
"loss": 0.1316,
|
| 449 |
+
"step": 360
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.83,
|
| 453 |
+
"learning_rate": 2.6183644189383074e-05,
|
| 454 |
+
"loss": 0.2164,
|
| 455 |
+
"step": 365
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 1.86,
|
| 459 |
+
"learning_rate": 2.6542324246771883e-05,
|
| 460 |
+
"loss": 0.1956,
|
| 461 |
+
"step": 370
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 1.88,
|
| 465 |
+
"learning_rate": 2.6901004304160688e-05,
|
| 466 |
+
"loss": 0.1178,
|
| 467 |
+
"step": 375
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 1.91,
|
| 471 |
+
"learning_rate": 2.7259684361549497e-05,
|
| 472 |
+
"loss": 0.2018,
|
| 473 |
+
"step": 380
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.93,
|
| 477 |
+
"learning_rate": 2.761836441893831e-05,
|
| 478 |
+
"loss": 0.1793,
|
| 479 |
+
"step": 385
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.96,
|
| 483 |
+
"learning_rate": 2.7977044476327118e-05,
|
| 484 |
+
"loss": 0.1478,
|
| 485 |
+
"step": 390
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.98,
|
| 489 |
+
"learning_rate": 2.833572453371593e-05,
|
| 490 |
+
"loss": 0.1302,
|
| 491 |
+
"step": 395
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 2.0,
|
| 495 |
+
"eval_accuracy": 0.9735973597359736,
|
| 496 |
+
"eval_loss": 0.06645132601261139,
|
| 497 |
+
"eval_runtime": 102.5426,
|
| 498 |
+
"eval_samples_per_second": 20.684,
|
| 499 |
+
"eval_steps_per_second": 2.594,
|
| 500 |
+
"step": 398
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 2.01,
|
| 504 |
+
"learning_rate": 2.8694404591104736e-05,
|
| 505 |
+
"loss": 0.137,
|
| 506 |
+
"step": 400
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 2.04,
|
| 510 |
+
"learning_rate": 2.9053084648493545e-05,
|
| 511 |
+
"loss": 0.1309,
|
| 512 |
+
"step": 405
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 2.06,
|
| 516 |
+
"learning_rate": 2.9411764705882354e-05,
|
| 517 |
+
"loss": 0.1257,
|
| 518 |
+
"step": 410
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 2.09,
|
| 522 |
+
"learning_rate": 2.9770444763271166e-05,
|
| 523 |
+
"loss": 0.1457,
|
| 524 |
+
"step": 415
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 2.11,
|
| 528 |
+
"learning_rate": 3.012912482065997e-05,
|
| 529 |
+
"loss": 0.259,
|
| 530 |
+
"step": 420
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 2.14,
|
| 534 |
+
"learning_rate": 3.048780487804878e-05,
|
| 535 |
+
"loss": 0.1984,
|
| 536 |
+
"step": 425
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 2.16,
|
| 540 |
+
"learning_rate": 3.084648493543759e-05,
|
| 541 |
+
"loss": 0.1493,
|
| 542 |
+
"step": 430
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 2.19,
|
| 546 |
+
"learning_rate": 3.1205164992826405e-05,
|
| 547 |
+
"loss": 0.1391,
|
| 548 |
+
"step": 435
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 2.21,
|
| 552 |
+
"learning_rate": 3.156384505021521e-05,
|
| 553 |
+
"loss": 0.1312,
|
| 554 |
+
"step": 440
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 2.24,
|
| 558 |
+
"learning_rate": 3.1922525107604016e-05,
|
| 559 |
+
"loss": 0.1164,
|
| 560 |
+
"step": 445
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 2.26,
|
| 564 |
+
"learning_rate": 3.228120516499283e-05,
|
| 565 |
+
"loss": 0.1189,
|
| 566 |
+
"step": 450
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 2.29,
|
| 570 |
+
"learning_rate": 3.263988522238164e-05,
|
| 571 |
+
"loss": 0.1296,
|
| 572 |
+
"step": 455
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 2.31,
|
| 576 |
+
"learning_rate": 3.2998565279770446e-05,
|
| 577 |
+
"loss": 0.078,
|
| 578 |
+
"step": 460
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 2.34,
|
| 582 |
+
"learning_rate": 3.335724533715925e-05,
|
| 583 |
+
"loss": 0.0943,
|
| 584 |
+
"step": 465
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 2.36,
|
| 588 |
+
"learning_rate": 3.3715925394548064e-05,
|
| 589 |
+
"loss": 0.0932,
|
| 590 |
+
"step": 470
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 2.39,
|
| 594 |
+
"learning_rate": 3.4074605451936876e-05,
|
| 595 |
+
"loss": 0.1722,
|
| 596 |
+
"step": 475
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 2.41,
|
| 600 |
+
"learning_rate": 3.443328550932569e-05,
|
| 601 |
+
"loss": 0.2213,
|
| 602 |
+
"step": 480
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 2.44,
|
| 606 |
+
"learning_rate": 3.479196556671449e-05,
|
| 607 |
+
"loss": 0.1573,
|
| 608 |
+
"step": 485
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 2.46,
|
| 612 |
+
"learning_rate": 3.51506456241033e-05,
|
| 613 |
+
"loss": 0.2181,
|
| 614 |
+
"step": 490
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 2.49,
|
| 618 |
+
"learning_rate": 3.550932568149211e-05,
|
| 619 |
+
"loss": 0.1229,
|
| 620 |
+
"step": 495
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 2.51,
|
| 624 |
+
"learning_rate": 3.5868005738880924e-05,
|
| 625 |
+
"loss": 0.0817,
|
| 626 |
+
"step": 500
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 2.54,
|
| 630 |
+
"learning_rate": 3.622668579626973e-05,
|
| 631 |
+
"loss": 0.0617,
|
| 632 |
+
"step": 505
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 2.56,
|
| 636 |
+
"learning_rate": 3.6585365853658535e-05,
|
| 637 |
+
"loss": 0.2366,
|
| 638 |
+
"step": 510
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 2.59,
|
| 642 |
+
"learning_rate": 3.694404591104735e-05,
|
| 643 |
+
"loss": 0.1142,
|
| 644 |
+
"step": 515
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 2.61,
|
| 648 |
+
"learning_rate": 3.730272596843616e-05,
|
| 649 |
+
"loss": 0.1394,
|
| 650 |
+
"step": 520
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 2.64,
|
| 654 |
+
"learning_rate": 3.7661406025824965e-05,
|
| 655 |
+
"loss": 0.1163,
|
| 656 |
+
"step": 525
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 2.66,
|
| 660 |
+
"learning_rate": 3.802008608321377e-05,
|
| 661 |
+
"loss": 0.1146,
|
| 662 |
+
"step": 530
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 2.69,
|
| 666 |
+
"learning_rate": 3.837876614060258e-05,
|
| 667 |
+
"loss": 0.1014,
|
| 668 |
+
"step": 535
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 2.71,
|
| 672 |
+
"learning_rate": 3.8737446197991395e-05,
|
| 673 |
+
"loss": 0.0707,
|
| 674 |
+
"step": 540
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 2.74,
|
| 678 |
+
"learning_rate": 3.909612625538021e-05,
|
| 679 |
+
"loss": 0.1651,
|
| 680 |
+
"step": 545
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 2.76,
|
| 684 |
+
"learning_rate": 3.945480631276901e-05,
|
| 685 |
+
"loss": 0.3627,
|
| 686 |
+
"step": 550
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 2.79,
|
| 690 |
+
"learning_rate": 3.981348637015782e-05,
|
| 691 |
+
"loss": 0.1615,
|
| 692 |
+
"step": 555
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 2.81,
|
| 696 |
+
"learning_rate": 4.017216642754663e-05,
|
| 697 |
+
"loss": 0.2595,
|
| 698 |
+
"step": 560
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 2.84,
|
| 702 |
+
"learning_rate": 4.053084648493544e-05,
|
| 703 |
+
"loss": 0.1507,
|
| 704 |
+
"step": 565
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 2.86,
|
| 708 |
+
"learning_rate": 4.088952654232425e-05,
|
| 709 |
+
"loss": 0.1336,
|
| 710 |
+
"step": 570
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 2.89,
|
| 714 |
+
"learning_rate": 4.1248206599713054e-05,
|
| 715 |
+
"loss": 0.093,
|
| 716 |
+
"step": 575
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 2.91,
|
| 720 |
+
"learning_rate": 4.160688665710187e-05,
|
| 721 |
+
"loss": 0.1895,
|
| 722 |
+
"step": 580
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 2.94,
|
| 726 |
+
"learning_rate": 4.196556671449068e-05,
|
| 727 |
+
"loss": 0.1971,
|
| 728 |
+
"step": 585
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 2.96,
|
| 732 |
+
"learning_rate": 4.2324246771879484e-05,
|
| 733 |
+
"loss": 0.1536,
|
| 734 |
+
"step": 590
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 2.99,
|
| 738 |
+
"learning_rate": 4.26829268292683e-05,
|
| 739 |
+
"loss": 0.0784,
|
| 740 |
+
"step": 595
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 3.0,
|
| 744 |
+
"eval_accuracy": 0.9778406412069779,
|
| 745 |
+
"eval_loss": 0.05999346077442169,
|
| 746 |
+
"eval_runtime": 102.3498,
|
| 747 |
+
"eval_samples_per_second": 20.723,
|
| 748 |
+
"eval_steps_per_second": 2.599,
|
| 749 |
+
"step": 597
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 3.02,
|
| 753 |
+
"learning_rate": 4.30416068866571e-05,
|
| 754 |
+
"loss": 0.1106,
|
| 755 |
+
"step": 600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 3.04,
|
| 759 |
+
"learning_rate": 4.3400286944045915e-05,
|
| 760 |
+
"loss": 0.1983,
|
| 761 |
+
"step": 605
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 3.07,
|
| 765 |
+
"learning_rate": 4.375896700143472e-05,
|
| 766 |
+
"loss": 0.1241,
|
| 767 |
+
"step": 610
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 3.09,
|
| 771 |
+
"learning_rate": 4.411764705882353e-05,
|
| 772 |
+
"loss": 0.1141,
|
| 773 |
+
"step": 615
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 3.12,
|
| 777 |
+
"learning_rate": 4.447632711621234e-05,
|
| 778 |
+
"loss": 0.1416,
|
| 779 |
+
"step": 620
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 3.14,
|
| 783 |
+
"learning_rate": 4.483500717360115e-05,
|
| 784 |
+
"loss": 0.1494,
|
| 785 |
+
"step": 625
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 3.17,
|
| 789 |
+
"learning_rate": 4.519368723098996e-05,
|
| 790 |
+
"loss": 0.0762,
|
| 791 |
+
"step": 630
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 3.19,
|
| 795 |
+
"learning_rate": 4.555236728837877e-05,
|
| 796 |
+
"loss": 0.1448,
|
| 797 |
+
"step": 635
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 3.22,
|
| 801 |
+
"learning_rate": 4.591104734576758e-05,
|
| 802 |
+
"loss": 0.1665,
|
| 803 |
+
"step": 640
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 3.24,
|
| 807 |
+
"learning_rate": 4.6269727403156386e-05,
|
| 808 |
+
"loss": 0.0737,
|
| 809 |
+
"step": 645
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 3.27,
|
| 813 |
+
"learning_rate": 4.66284074605452e-05,
|
| 814 |
+
"loss": 0.0924,
|
| 815 |
+
"step": 650
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 3.29,
|
| 819 |
+
"learning_rate": 4.6987087517934004e-05,
|
| 820 |
+
"loss": 0.1181,
|
| 821 |
+
"step": 655
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 3.32,
|
| 825 |
+
"learning_rate": 4.7345767575322816e-05,
|
| 826 |
+
"loss": 0.2936,
|
| 827 |
+
"step": 660
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 3.34,
|
| 831 |
+
"learning_rate": 4.770444763271162e-05,
|
| 832 |
+
"loss": 0.1374,
|
| 833 |
+
"step": 665
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 3.37,
|
| 837 |
+
"learning_rate": 4.8063127690100434e-05,
|
| 838 |
+
"loss": 0.1504,
|
| 839 |
+
"step": 670
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 3.39,
|
| 843 |
+
"learning_rate": 4.842180774748924e-05,
|
| 844 |
+
"loss": 0.2001,
|
| 845 |
+
"step": 675
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 3.42,
|
| 849 |
+
"learning_rate": 4.878048780487805e-05,
|
| 850 |
+
"loss": 0.0628,
|
| 851 |
+
"step": 680
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 3.44,
|
| 855 |
+
"learning_rate": 4.9139167862266864e-05,
|
| 856 |
+
"loss": 0.116,
|
| 857 |
+
"step": 685
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 3.47,
|
| 861 |
+
"learning_rate": 4.949784791965567e-05,
|
| 862 |
+
"loss": 0.1345,
|
| 863 |
+
"step": 690
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 3.49,
|
| 867 |
+
"learning_rate": 4.9856527977044475e-05,
|
| 868 |
+
"loss": 0.119,
|
| 869 |
+
"step": 695
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 3.52,
|
| 873 |
+
"learning_rate": 4.978448275862069e-05,
|
| 874 |
+
"loss": 0.0882,
|
| 875 |
+
"step": 700
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 3.54,
|
| 879 |
+
"learning_rate": 4.9425287356321845e-05,
|
| 880 |
+
"loss": 0.0853,
|
| 881 |
+
"step": 705
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 3.57,
|
| 885 |
+
"learning_rate": 4.906609195402299e-05,
|
| 886 |
+
"loss": 0.2363,
|
| 887 |
+
"step": 710
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 3.59,
|
| 891 |
+
"learning_rate": 4.870689655172414e-05,
|
| 892 |
+
"loss": 0.1878,
|
| 893 |
+
"step": 715
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 3.62,
|
| 897 |
+
"learning_rate": 4.834770114942529e-05,
|
| 898 |
+
"loss": 0.0954,
|
| 899 |
+
"step": 720
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 3.64,
|
| 903 |
+
"learning_rate": 4.798850574712644e-05,
|
| 904 |
+
"loss": 0.1891,
|
| 905 |
+
"step": 725
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 3.67,
|
| 909 |
+
"learning_rate": 4.762931034482759e-05,
|
| 910 |
+
"loss": 0.1562,
|
| 911 |
+
"step": 730
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 3.69,
|
| 915 |
+
"learning_rate": 4.7270114942528734e-05,
|
| 916 |
+
"loss": 0.1207,
|
| 917 |
+
"step": 735
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 3.72,
|
| 921 |
+
"learning_rate": 4.6910919540229886e-05,
|
| 922 |
+
"loss": 0.096,
|
| 923 |
+
"step": 740
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 3.74,
|
| 927 |
+
"learning_rate": 4.655172413793104e-05,
|
| 928 |
+
"loss": 0.1072,
|
| 929 |
+
"step": 745
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 3.77,
|
| 933 |
+
"learning_rate": 4.619252873563218e-05,
|
| 934 |
+
"loss": 0.0718,
|
| 935 |
+
"step": 750
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 3.79,
|
| 939 |
+
"learning_rate": 4.5833333333333334e-05,
|
| 940 |
+
"loss": 0.0622,
|
| 941 |
+
"step": 755
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 3.82,
|
| 945 |
+
"learning_rate": 4.5474137931034485e-05,
|
| 946 |
+
"loss": 0.1661,
|
| 947 |
+
"step": 760
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 3.84,
|
| 951 |
+
"learning_rate": 4.511494252873563e-05,
|
| 952 |
+
"loss": 0.1559,
|
| 953 |
+
"step": 765
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 3.87,
|
| 957 |
+
"learning_rate": 4.475574712643678e-05,
|
| 958 |
+
"loss": 0.1195,
|
| 959 |
+
"step": 770
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 3.89,
|
| 963 |
+
"learning_rate": 4.4396551724137933e-05,
|
| 964 |
+
"loss": 0.105,
|
| 965 |
+
"step": 775
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 3.92,
|
| 969 |
+
"learning_rate": 4.4037356321839085e-05,
|
| 970 |
+
"loss": 0.0764,
|
| 971 |
+
"step": 780
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 3.94,
|
| 975 |
+
"learning_rate": 4.367816091954024e-05,
|
| 976 |
+
"loss": 0.1322,
|
| 977 |
+
"step": 785
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 3.97,
|
| 981 |
+
"learning_rate": 4.331896551724138e-05,
|
| 982 |
+
"loss": 0.0829,
|
| 983 |
+
"step": 790
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 3.99,
|
| 987 |
+
"learning_rate": 4.295977011494253e-05,
|
| 988 |
+
"loss": 0.1181,
|
| 989 |
+
"step": 795
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 4.0,
|
| 993 |
+
"eval_accuracy": 0.9849127769919849,
|
| 994 |
+
"eval_loss": 0.04494262859225273,
|
| 995 |
+
"eval_runtime": 102.4081,
|
| 996 |
+
"eval_samples_per_second": 20.711,
|
| 997 |
+
"eval_steps_per_second": 2.597,
|
| 998 |
+
"step": 796
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 4.02,
|
| 1002 |
+
"learning_rate": 4.2600574712643685e-05,
|
| 1003 |
+
"loss": 0.1042,
|
| 1004 |
+
"step": 800
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 4.05,
|
| 1008 |
+
"learning_rate": 4.224137931034483e-05,
|
| 1009 |
+
"loss": 0.0089,
|
| 1010 |
+
"step": 805
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 4.07,
|
| 1014 |
+
"learning_rate": 4.188218390804598e-05,
|
| 1015 |
+
"loss": 0.1135,
|
| 1016 |
+
"step": 810
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 4.1,
|
| 1020 |
+
"learning_rate": 4.1522988505747126e-05,
|
| 1021 |
+
"loss": 0.0473,
|
| 1022 |
+
"step": 815
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 4.12,
|
| 1026 |
+
"learning_rate": 4.116379310344828e-05,
|
| 1027 |
+
"loss": 0.2587,
|
| 1028 |
+
"step": 820
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 4.15,
|
| 1032 |
+
"learning_rate": 4.080459770114943e-05,
|
| 1033 |
+
"loss": 0.1912,
|
| 1034 |
+
"step": 825
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 4.17,
|
| 1038 |
+
"learning_rate": 4.0445402298850574e-05,
|
| 1039 |
+
"loss": 0.1441,
|
| 1040 |
+
"step": 830
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 4.2,
|
| 1044 |
+
"learning_rate": 4.0086206896551726e-05,
|
| 1045 |
+
"loss": 0.0334,
|
| 1046 |
+
"step": 835
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 4.22,
|
| 1050 |
+
"learning_rate": 3.972701149425288e-05,
|
| 1051 |
+
"loss": 0.0537,
|
| 1052 |
+
"step": 840
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 4.25,
|
| 1056 |
+
"learning_rate": 3.936781609195402e-05,
|
| 1057 |
+
"loss": 0.1477,
|
| 1058 |
+
"step": 845
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 4.27,
|
| 1062 |
+
"learning_rate": 3.9008620689655174e-05,
|
| 1063 |
+
"loss": 0.0594,
|
| 1064 |
+
"step": 850
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 4.3,
|
| 1068 |
+
"learning_rate": 3.8649425287356325e-05,
|
| 1069 |
+
"loss": 0.0833,
|
| 1070 |
+
"step": 855
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 4.32,
|
| 1074 |
+
"learning_rate": 3.829022988505747e-05,
|
| 1075 |
+
"loss": 0.1532,
|
| 1076 |
+
"step": 860
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 4.35,
|
| 1080 |
+
"learning_rate": 3.793103448275862e-05,
|
| 1081 |
+
"loss": 0.1087,
|
| 1082 |
+
"step": 865
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 4.37,
|
| 1086 |
+
"learning_rate": 3.7571839080459766e-05,
|
| 1087 |
+
"loss": 0.1735,
|
| 1088 |
+
"step": 870
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 4.4,
|
| 1092 |
+
"learning_rate": 3.721264367816092e-05,
|
| 1093 |
+
"loss": 0.1434,
|
| 1094 |
+
"step": 875
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 4.42,
|
| 1098 |
+
"learning_rate": 3.685344827586207e-05,
|
| 1099 |
+
"loss": 0.1227,
|
| 1100 |
+
"step": 880
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 4.45,
|
| 1104 |
+
"learning_rate": 3.649425287356322e-05,
|
| 1105 |
+
"loss": 0.1714,
|
| 1106 |
+
"step": 885
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 4.47,
|
| 1110 |
+
"learning_rate": 3.613505747126437e-05,
|
| 1111 |
+
"loss": 0.1588,
|
| 1112 |
+
"step": 890
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 4.5,
|
| 1116 |
+
"learning_rate": 3.5775862068965524e-05,
|
| 1117 |
+
"loss": 0.1283,
|
| 1118 |
+
"step": 895
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 4.52,
|
| 1122 |
+
"learning_rate": 3.541666666666667e-05,
|
| 1123 |
+
"loss": 0.0578,
|
| 1124 |
+
"step": 900
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 4.55,
|
| 1128 |
+
"learning_rate": 3.505747126436782e-05,
|
| 1129 |
+
"loss": 0.119,
|
| 1130 |
+
"step": 905
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 4.57,
|
| 1134 |
+
"learning_rate": 3.4698275862068966e-05,
|
| 1135 |
+
"loss": 0.0986,
|
| 1136 |
+
"step": 910
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 4.6,
|
| 1140 |
+
"learning_rate": 3.433908045977012e-05,
|
| 1141 |
+
"loss": 0.0577,
|
| 1142 |
+
"step": 915
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 4.62,
|
| 1146 |
+
"learning_rate": 3.397988505747127e-05,
|
| 1147 |
+
"loss": 0.0148,
|
| 1148 |
+
"step": 920
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 4.65,
|
| 1152 |
+
"learning_rate": 3.3620689655172414e-05,
|
| 1153 |
+
"loss": 0.0293,
|
| 1154 |
+
"step": 925
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 4.67,
|
| 1158 |
+
"learning_rate": 3.3261494252873565e-05,
|
| 1159 |
+
"loss": 0.0276,
|
| 1160 |
+
"step": 930
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 4.7,
|
| 1164 |
+
"learning_rate": 3.290229885057472e-05,
|
| 1165 |
+
"loss": 0.0033,
|
| 1166 |
+
"step": 935
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 4.72,
|
| 1170 |
+
"learning_rate": 3.254310344827586e-05,
|
| 1171 |
+
"loss": 0.0977,
|
| 1172 |
+
"step": 940
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 4.75,
|
| 1176 |
+
"learning_rate": 3.218390804597701e-05,
|
| 1177 |
+
"loss": 0.0749,
|
| 1178 |
+
"step": 945
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 4.77,
|
| 1182 |
+
"learning_rate": 3.1824712643678165e-05,
|
| 1183 |
+
"loss": 0.0323,
|
| 1184 |
+
"step": 950
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 4.8,
|
| 1188 |
+
"learning_rate": 3.146551724137931e-05,
|
| 1189 |
+
"loss": 0.1322,
|
| 1190 |
+
"step": 955
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 4.82,
|
| 1194 |
+
"learning_rate": 3.110632183908046e-05,
|
| 1195 |
+
"loss": 0.081,
|
| 1196 |
+
"step": 960
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 4.85,
|
| 1200 |
+
"learning_rate": 3.0747126436781606e-05,
|
| 1201 |
+
"loss": 0.1045,
|
| 1202 |
+
"step": 965
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 4.87,
|
| 1206 |
+
"learning_rate": 3.0387931034482758e-05,
|
| 1207 |
+
"loss": 0.1323,
|
| 1208 |
+
"step": 970
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 4.9,
|
| 1212 |
+
"learning_rate": 3.0028735632183906e-05,
|
| 1213 |
+
"loss": 0.2384,
|
| 1214 |
+
"step": 975
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 4.92,
|
| 1218 |
+
"learning_rate": 2.9669540229885058e-05,
|
| 1219 |
+
"loss": 0.0663,
|
| 1220 |
+
"step": 980
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 4.95,
|
| 1224 |
+
"learning_rate": 2.9310344827586206e-05,
|
| 1225 |
+
"loss": 0.0836,
|
| 1226 |
+
"step": 985
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 4.97,
|
| 1230 |
+
"learning_rate": 2.8951149425287354e-05,
|
| 1231 |
+
"loss": 0.1856,
|
| 1232 |
+
"step": 990
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 5.0,
|
| 1236 |
+
"learning_rate": 2.859195402298851e-05,
|
| 1237 |
+
"loss": 0.208,
|
| 1238 |
+
"step": 995
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 5.0,
|
| 1242 |
+
"eval_accuracy": 0.9886845827439887,
|
| 1243 |
+
"eval_loss": 0.03925405070185661,
|
| 1244 |
+
"eval_runtime": 102.5066,
|
| 1245 |
+
"eval_samples_per_second": 20.691,
|
| 1246 |
+
"eval_steps_per_second": 2.595,
|
| 1247 |
+
"step": 995
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 5.03,
|
| 1251 |
+
"learning_rate": 2.8232758620689657e-05,
|
| 1252 |
+
"loss": 0.0379,
|
| 1253 |
+
"step": 1000
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 5.05,
|
| 1257 |
+
"learning_rate": 2.787356321839081e-05,
|
| 1258 |
+
"loss": 0.1243,
|
| 1259 |
+
"step": 1005
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 5.08,
|
| 1263 |
+
"learning_rate": 2.7514367816091957e-05,
|
| 1264 |
+
"loss": 0.0983,
|
| 1265 |
+
"step": 1010
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 5.1,
|
| 1269 |
+
"learning_rate": 2.7155172413793105e-05,
|
| 1270 |
+
"loss": 0.0448,
|
| 1271 |
+
"step": 1015
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 5.13,
|
| 1275 |
+
"learning_rate": 2.6795977011494257e-05,
|
| 1276 |
+
"loss": 0.1149,
|
| 1277 |
+
"step": 1020
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 5.15,
|
| 1281 |
+
"learning_rate": 2.6436781609195405e-05,
|
| 1282 |
+
"loss": 0.0619,
|
| 1283 |
+
"step": 1025
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 5.18,
|
| 1287 |
+
"learning_rate": 2.6077586206896553e-05,
|
| 1288 |
+
"loss": 0.1441,
|
| 1289 |
+
"step": 1030
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 5.2,
|
| 1293 |
+
"learning_rate": 2.57183908045977e-05,
|
| 1294 |
+
"loss": 0.0575,
|
| 1295 |
+
"step": 1035
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 5.23,
|
| 1299 |
+
"learning_rate": 2.5359195402298853e-05,
|
| 1300 |
+
"loss": 0.067,
|
| 1301 |
+
"step": 1040
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 5.25,
|
| 1305 |
+
"learning_rate": 2.5e-05,
|
| 1306 |
+
"loss": 0.0616,
|
| 1307 |
+
"step": 1045
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 5.28,
|
| 1311 |
+
"learning_rate": 2.464080459770115e-05,
|
| 1312 |
+
"loss": 0.0797,
|
| 1313 |
+
"step": 1050
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 5.3,
|
| 1317 |
+
"learning_rate": 2.42816091954023e-05,
|
| 1318 |
+
"loss": 0.043,
|
| 1319 |
+
"step": 1055
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 5.33,
|
| 1323 |
+
"learning_rate": 2.392241379310345e-05,
|
| 1324 |
+
"loss": 0.0809,
|
| 1325 |
+
"step": 1060
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 5.35,
|
| 1329 |
+
"learning_rate": 2.3563218390804597e-05,
|
| 1330 |
+
"loss": 0.0548,
|
| 1331 |
+
"step": 1065
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 5.38,
|
| 1335 |
+
"learning_rate": 2.3204022988505746e-05,
|
| 1336 |
+
"loss": 0.0662,
|
| 1337 |
+
"step": 1070
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 5.4,
|
| 1341 |
+
"learning_rate": 2.2844827586206897e-05,
|
| 1342 |
+
"loss": 0.1148,
|
| 1343 |
+
"step": 1075
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 5.43,
|
| 1347 |
+
"learning_rate": 2.248563218390805e-05,
|
| 1348 |
+
"loss": 0.0548,
|
| 1349 |
+
"step": 1080
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 5.45,
|
| 1353 |
+
"learning_rate": 2.2126436781609197e-05,
|
| 1354 |
+
"loss": 0.0658,
|
| 1355 |
+
"step": 1085
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 5.48,
|
| 1359 |
+
"learning_rate": 2.1767241379310345e-05,
|
| 1360 |
+
"loss": 0.0686,
|
| 1361 |
+
"step": 1090
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 5.5,
|
| 1365 |
+
"learning_rate": 2.1408045977011497e-05,
|
| 1366 |
+
"loss": 0.1114,
|
| 1367 |
+
"step": 1095
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 5.53,
|
| 1371 |
+
"learning_rate": 2.1048850574712645e-05,
|
| 1372 |
+
"loss": 0.0262,
|
| 1373 |
+
"step": 1100
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 5.55,
|
| 1377 |
+
"learning_rate": 2.0689655172413793e-05,
|
| 1378 |
+
"loss": 0.0745,
|
| 1379 |
+
"step": 1105
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 5.58,
|
| 1383 |
+
"learning_rate": 2.033045977011494e-05,
|
| 1384 |
+
"loss": 0.1139,
|
| 1385 |
+
"step": 1110
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 5.6,
|
| 1389 |
+
"learning_rate": 1.9971264367816093e-05,
|
| 1390 |
+
"loss": 0.1325,
|
| 1391 |
+
"step": 1115
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 5.63,
|
| 1395 |
+
"learning_rate": 1.961206896551724e-05,
|
| 1396 |
+
"loss": 0.0565,
|
| 1397 |
+
"step": 1120
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 5.65,
|
| 1401 |
+
"learning_rate": 1.925287356321839e-05,
|
| 1402 |
+
"loss": 0.0274,
|
| 1403 |
+
"step": 1125
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 5.68,
|
| 1407 |
+
"learning_rate": 1.889367816091954e-05,
|
| 1408 |
+
"loss": 0.1295,
|
| 1409 |
+
"step": 1130
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 5.7,
|
| 1413 |
+
"learning_rate": 1.8534482758620693e-05,
|
| 1414 |
+
"loss": 0.0501,
|
| 1415 |
+
"step": 1135
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 5.73,
|
| 1419 |
+
"learning_rate": 1.817528735632184e-05,
|
| 1420 |
+
"loss": 0.0586,
|
| 1421 |
+
"step": 1140
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 5.75,
|
| 1425 |
+
"learning_rate": 1.781609195402299e-05,
|
| 1426 |
+
"loss": 0.045,
|
| 1427 |
+
"step": 1145
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 5.78,
|
| 1431 |
+
"learning_rate": 1.7456896551724137e-05,
|
| 1432 |
+
"loss": 0.0226,
|
| 1433 |
+
"step": 1150
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 5.8,
|
| 1437 |
+
"learning_rate": 1.709770114942529e-05,
|
| 1438 |
+
"loss": 0.0166,
|
| 1439 |
+
"step": 1155
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 5.83,
|
| 1443 |
+
"learning_rate": 1.6738505747126437e-05,
|
| 1444 |
+
"loss": 0.1241,
|
| 1445 |
+
"step": 1160
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 5.85,
|
| 1449 |
+
"learning_rate": 1.6379310344827585e-05,
|
| 1450 |
+
"loss": 0.0243,
|
| 1451 |
+
"step": 1165
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 5.88,
|
| 1455 |
+
"learning_rate": 1.6020114942528737e-05,
|
| 1456 |
+
"loss": 0.0681,
|
| 1457 |
+
"step": 1170
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 5.9,
|
| 1461 |
+
"learning_rate": 1.5660919540229885e-05,
|
| 1462 |
+
"loss": 0.0099,
|
| 1463 |
+
"step": 1175
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 5.93,
|
| 1467 |
+
"learning_rate": 1.5301724137931033e-05,
|
| 1468 |
+
"loss": 0.0196,
|
| 1469 |
+
"step": 1180
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 5.95,
|
| 1473 |
+
"learning_rate": 1.4942528735632185e-05,
|
| 1474 |
+
"loss": 0.1818,
|
| 1475 |
+
"step": 1185
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 5.98,
|
| 1479 |
+
"learning_rate": 1.4583333333333335e-05,
|
| 1480 |
+
"loss": 0.0057,
|
| 1481 |
+
"step": 1190
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 6.0,
|
| 1485 |
+
"eval_accuracy": 0.9943422913719944,
|
| 1486 |
+
"eval_loss": 0.022870095446705818,
|
| 1487 |
+
"eval_runtime": 102.5148,
|
| 1488 |
+
"eval_samples_per_second": 20.69,
|
| 1489 |
+
"eval_steps_per_second": 2.595,
|
| 1490 |
+
"step": 1194
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 6.01,
|
| 1494 |
+
"learning_rate": 1.4224137931034485e-05,
|
| 1495 |
+
"loss": 0.0652,
|
| 1496 |
+
"step": 1195
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 6.03,
|
| 1500 |
+
"learning_rate": 1.3864942528735633e-05,
|
| 1501 |
+
"loss": 0.0672,
|
| 1502 |
+
"step": 1200
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 6.06,
|
| 1506 |
+
"learning_rate": 1.3505747126436783e-05,
|
| 1507 |
+
"loss": 0.0322,
|
| 1508 |
+
"step": 1205
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 6.08,
|
| 1512 |
+
"learning_rate": 1.3146551724137931e-05,
|
| 1513 |
+
"loss": 0.0374,
|
| 1514 |
+
"step": 1210
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 6.11,
|
| 1518 |
+
"learning_rate": 1.2787356321839081e-05,
|
| 1519 |
+
"loss": 0.0112,
|
| 1520 |
+
"step": 1215
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 6.13,
|
| 1524 |
+
"learning_rate": 1.242816091954023e-05,
|
| 1525 |
+
"loss": 0.0284,
|
| 1526 |
+
"step": 1220
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 6.16,
|
| 1530 |
+
"learning_rate": 1.206896551724138e-05,
|
| 1531 |
+
"loss": 0.0688,
|
| 1532 |
+
"step": 1225
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 6.18,
|
| 1536 |
+
"learning_rate": 1.1709770114942529e-05,
|
| 1537 |
+
"loss": 0.0271,
|
| 1538 |
+
"step": 1230
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 6.21,
|
| 1542 |
+
"learning_rate": 1.1350574712643679e-05,
|
| 1543 |
+
"loss": 0.0048,
|
| 1544 |
+
"step": 1235
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 6.23,
|
| 1548 |
+
"learning_rate": 1.0991379310344827e-05,
|
| 1549 |
+
"loss": 0.0699,
|
| 1550 |
+
"step": 1240
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 6.26,
|
| 1554 |
+
"learning_rate": 1.0632183908045977e-05,
|
| 1555 |
+
"loss": 0.0951,
|
| 1556 |
+
"step": 1245
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 6.28,
|
| 1560 |
+
"learning_rate": 1.0272988505747127e-05,
|
| 1561 |
+
"loss": 0.0334,
|
| 1562 |
+
"step": 1250
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 6.31,
|
| 1566 |
+
"learning_rate": 9.913793103448277e-06,
|
| 1567 |
+
"loss": 0.061,
|
| 1568 |
+
"step": 1255
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 6.33,
|
| 1572 |
+
"learning_rate": 9.554597701149425e-06,
|
| 1573 |
+
"loss": 0.0025,
|
| 1574 |
+
"step": 1260
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 6.36,
|
| 1578 |
+
"learning_rate": 9.195402298850575e-06,
|
| 1579 |
+
"loss": 0.0532,
|
| 1580 |
+
"step": 1265
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 6.38,
|
| 1584 |
+
"learning_rate": 8.836206896551725e-06,
|
| 1585 |
+
"loss": 0.0359,
|
| 1586 |
+
"step": 1270
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 6.41,
|
| 1590 |
+
"learning_rate": 8.477011494252873e-06,
|
| 1591 |
+
"loss": 0.0034,
|
| 1592 |
+
"step": 1275
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 6.43,
|
| 1596 |
+
"learning_rate": 8.117816091954025e-06,
|
| 1597 |
+
"loss": 0.0309,
|
| 1598 |
+
"step": 1280
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 6.46,
|
| 1602 |
+
"learning_rate": 7.758620689655173e-06,
|
| 1603 |
+
"loss": 0.012,
|
| 1604 |
+
"step": 1285
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 6.48,
|
| 1608 |
+
"learning_rate": 7.399425287356322e-06,
|
| 1609 |
+
"loss": 0.0651,
|
| 1610 |
+
"step": 1290
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 6.51,
|
| 1614 |
+
"learning_rate": 7.040229885057471e-06,
|
| 1615 |
+
"loss": 0.0261,
|
| 1616 |
+
"step": 1295
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 6.53,
|
| 1620 |
+
"learning_rate": 6.68103448275862e-06,
|
| 1621 |
+
"loss": 0.0366,
|
| 1622 |
+
"step": 1300
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 6.56,
|
| 1626 |
+
"learning_rate": 6.321839080459771e-06,
|
| 1627 |
+
"loss": 0.0095,
|
| 1628 |
+
"step": 1305
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 6.58,
|
| 1632 |
+
"learning_rate": 5.96264367816092e-06,
|
| 1633 |
+
"loss": 0.0244,
|
| 1634 |
+
"step": 1310
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 6.61,
|
| 1638 |
+
"learning_rate": 5.603448275862069e-06,
|
| 1639 |
+
"loss": 0.0133,
|
| 1640 |
+
"step": 1315
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 6.63,
|
| 1644 |
+
"learning_rate": 5.244252873563219e-06,
|
| 1645 |
+
"loss": 0.0611,
|
| 1646 |
+
"step": 1320
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 6.66,
|
| 1650 |
+
"learning_rate": 4.885057471264369e-06,
|
| 1651 |
+
"loss": 0.0558,
|
| 1652 |
+
"step": 1325
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 6.68,
|
| 1656 |
+
"learning_rate": 4.525862068965518e-06,
|
| 1657 |
+
"loss": 0.0074,
|
| 1658 |
+
"step": 1330
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 6.71,
|
| 1662 |
+
"learning_rate": 4.166666666666667e-06,
|
| 1663 |
+
"loss": 0.0219,
|
| 1664 |
+
"step": 1335
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 6.73,
|
| 1668 |
+
"learning_rate": 3.8074712643678163e-06,
|
| 1669 |
+
"loss": 0.0106,
|
| 1670 |
+
"step": 1340
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 6.76,
|
| 1674 |
+
"learning_rate": 3.448275862068966e-06,
|
| 1675 |
+
"loss": 0.0009,
|
| 1676 |
+
"step": 1345
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 6.78,
|
| 1680 |
+
"learning_rate": 3.0890804597701153e-06,
|
| 1681 |
+
"loss": 0.0028,
|
| 1682 |
+
"step": 1350
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 6.81,
|
| 1686 |
+
"learning_rate": 2.729885057471265e-06,
|
| 1687 |
+
"loss": 0.018,
|
| 1688 |
+
"step": 1355
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 6.83,
|
| 1692 |
+
"learning_rate": 2.370689655172414e-06,
|
| 1693 |
+
"loss": 0.0383,
|
| 1694 |
+
"step": 1360
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 6.86,
|
| 1698 |
+
"learning_rate": 2.0114942528735633e-06,
|
| 1699 |
+
"loss": 0.0463,
|
| 1700 |
+
"step": 1365
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 6.88,
|
| 1704 |
+
"learning_rate": 1.6522988505747128e-06,
|
| 1705 |
+
"loss": 0.0113,
|
| 1706 |
+
"step": 1370
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 6.91,
|
| 1710 |
+
"learning_rate": 1.293103448275862e-06,
|
| 1711 |
+
"loss": 0.0684,
|
| 1712 |
+
"step": 1375
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 6.93,
|
| 1716 |
+
"learning_rate": 9.339080459770116e-07,
|
| 1717 |
+
"loss": 0.0319,
|
| 1718 |
+
"step": 1380
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 6.96,
|
| 1722 |
+
"learning_rate": 5.747126436781609e-07,
|
| 1723 |
+
"loss": 0.0689,
|
| 1724 |
+
"step": 1385
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 6.98,
|
| 1728 |
+
"learning_rate": 2.1551724137931036e-07,
|
| 1729 |
+
"loss": 0.0017,
|
| 1730 |
+
"step": 1390
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 7.0,
|
| 1734 |
+
"eval_accuracy": 0.9938708156529938,
|
| 1735 |
+
"eval_loss": 0.026258554309606552,
|
| 1736 |
+
"eval_runtime": 102.4925,
|
| 1737 |
+
"eval_samples_per_second": 20.694,
|
| 1738 |
+
"eval_steps_per_second": 2.595,
|
| 1739 |
+
"step": 1393
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 7.0,
|
| 1743 |
+
"step": 1393,
|
| 1744 |
+
"total_flos": 1.0256650917539217e+19,
|
| 1745 |
+
"train_loss": 0.15325962240578392,
|
| 1746 |
+
"train_runtime": 8636.3959,
|
| 1747 |
+
"train_samples_per_second": 5.157,
|
| 1748 |
+
"train_steps_per_second": 0.161
|
| 1749 |
+
}
|
| 1750 |
+
],
|
| 1751 |
+
"max_steps": 1393,
|
| 1752 |
+
"num_train_epochs": 7,
|
| 1753 |
+
"total_flos": 1.0256650917539217e+19,
|
| 1754 |
+
"trial_name": null,
|
| 1755 |
+
"trial_params": null
|
| 1756 |
+
}
|