End of training
Browse files- all_results.json +11 -0
- eval_results.json +7 -0
- pytorch_model.bin +1 -1
- train_results.json +7 -0
- trainer_state.json +3949 -0
all_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_loss": 0.19287119805812836,
|
| 4 |
+
"eval_runtime": 77.3332,
|
| 5 |
+
"eval_samples_per_second": 115.617,
|
| 6 |
+
"eval_steps_per_second": 1.81,
|
| 7 |
+
"train_loss": 0.13081293349950235,
|
| 8 |
+
"train_runtime": 214172.6098,
|
| 9 |
+
"train_samples_per_second": 76.308,
|
| 10 |
+
"train_steps_per_second": 0.298
|
| 11 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_loss": 0.19287119805812836,
|
| 4 |
+
"eval_runtime": 77.3332,
|
| 5 |
+
"eval_samples_per_second": 115.617,
|
| 6 |
+
"eval_steps_per_second": 1.81
|
| 7 |
+
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 290455265
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:831395f95548a7f966a4eec1ba50ef2a39c811fb27afbda6ff8eb22d4f71a98e
|
| 3 |
size 290455265
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"train_loss": 0.13081293349950235,
|
| 4 |
+
"train_runtime": 214172.6098,
|
| 5 |
+
"train_samples_per_second": 76.308,
|
| 6 |
+
"train_steps_per_second": 0.298
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3949 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.1957029402256012,
|
| 3 |
+
"best_model_checkpoint": "whisper-base-full-data-language-v2-20ep/checkpoint-60000",
|
| 4 |
+
"epoch": 20.003133322888925,
|
| 5 |
+
"global_step": 63840,
|
| 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": 2.9999999999999997e-06,
|
| 13 |
+
"loss": 1.6223,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.06,
|
| 18 |
+
"learning_rate": 5.999999999999999e-06,
|
| 19 |
+
"loss": 0.6381,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.09,
|
| 24 |
+
"learning_rate": 8.999999999999999e-06,
|
| 25 |
+
"loss": 0.5142,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.13,
|
| 30 |
+
"learning_rate": 1.1999999999999999e-05,
|
| 31 |
+
"loss": 0.4612,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.16,
|
| 36 |
+
"learning_rate": 1.4999999999999999e-05,
|
| 37 |
+
"loss": 0.4227,
|
| 38 |
+
"step": 500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.19,
|
| 42 |
+
"learning_rate": 1.7999999999999997e-05,
|
| 43 |
+
"loss": 0.4016,
|
| 44 |
+
"step": 600
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.22,
|
| 48 |
+
"learning_rate": 2.1e-05,
|
| 49 |
+
"loss": 0.3848,
|
| 50 |
+
"step": 700
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.25,
|
| 54 |
+
"learning_rate": 2.3999999999999997e-05,
|
| 55 |
+
"loss": 0.3705,
|
| 56 |
+
"step": 800
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.28,
|
| 60 |
+
"learning_rate": 2.6999999999999996e-05,
|
| 61 |
+
"loss": 0.3638,
|
| 62 |
+
"step": 900
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.31,
|
| 66 |
+
"learning_rate": 2.9999999999999997e-05,
|
| 67 |
+
"loss": 0.3519,
|
| 68 |
+
"step": 1000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.34,
|
| 72 |
+
"learning_rate": 3.2999999999999996e-05,
|
| 73 |
+
"loss": 0.3435,
|
| 74 |
+
"step": 1100
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.38,
|
| 78 |
+
"learning_rate": 3.5999999999999994e-05,
|
| 79 |
+
"loss": 0.3356,
|
| 80 |
+
"step": 1200
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.41,
|
| 84 |
+
"learning_rate": 3.9e-05,
|
| 85 |
+
"loss": 0.331,
|
| 86 |
+
"step": 1300
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.44,
|
| 90 |
+
"learning_rate": 4.2e-05,
|
| 91 |
+
"loss": 0.3284,
|
| 92 |
+
"step": 1400
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.47,
|
| 96 |
+
"learning_rate": 4.4999999999999996e-05,
|
| 97 |
+
"loss": 0.3212,
|
| 98 |
+
"step": 1500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.5,
|
| 102 |
+
"learning_rate": 4.7999999999999994e-05,
|
| 103 |
+
"loss": 0.3194,
|
| 104 |
+
"step": 1600
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.53,
|
| 108 |
+
"learning_rate": 5.1e-05,
|
| 109 |
+
"loss": 0.3167,
|
| 110 |
+
"step": 1700
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.56,
|
| 114 |
+
"learning_rate": 5.399999999999999e-05,
|
| 115 |
+
"loss": 0.3167,
|
| 116 |
+
"step": 1800
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.6,
|
| 120 |
+
"learning_rate": 5.6999999999999996e-05,
|
| 121 |
+
"loss": 0.311,
|
| 122 |
+
"step": 1900
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.63,
|
| 126 |
+
"learning_rate": 5.9999999999999995e-05,
|
| 127 |
+
"loss": 0.3116,
|
| 128 |
+
"step": 2000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.66,
|
| 132 |
+
"learning_rate": 6.299999999999999e-05,
|
| 133 |
+
"loss": 0.3051,
|
| 134 |
+
"step": 2100
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.69,
|
| 138 |
+
"learning_rate": 6.599999999999999e-05,
|
| 139 |
+
"loss": 0.3057,
|
| 140 |
+
"step": 2200
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.72,
|
| 144 |
+
"learning_rate": 6.9e-05,
|
| 145 |
+
"loss": 0.304,
|
| 146 |
+
"step": 2300
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.75,
|
| 150 |
+
"learning_rate": 7.199999999999999e-05,
|
| 151 |
+
"loss": 0.3119,
|
| 152 |
+
"step": 2400
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.78,
|
| 156 |
+
"learning_rate": 7.5e-05,
|
| 157 |
+
"loss": 0.3044,
|
| 158 |
+
"step": 2500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.81,
|
| 162 |
+
"learning_rate": 7.8e-05,
|
| 163 |
+
"loss": 0.3095,
|
| 164 |
+
"step": 2600
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.85,
|
| 168 |
+
"learning_rate": 8.1e-05,
|
| 169 |
+
"loss": 0.306,
|
| 170 |
+
"step": 2700
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.88,
|
| 174 |
+
"learning_rate": 8.4e-05,
|
| 175 |
+
"loss": 0.3081,
|
| 176 |
+
"step": 2800
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.91,
|
| 180 |
+
"learning_rate": 8.699999999999999e-05,
|
| 181 |
+
"loss": 0.3062,
|
| 182 |
+
"step": 2900
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.94,
|
| 186 |
+
"learning_rate": 8.999999999999999e-05,
|
| 187 |
+
"loss": 0.3022,
|
| 188 |
+
"step": 3000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.97,
|
| 192 |
+
"learning_rate": 9.3e-05,
|
| 193 |
+
"loss": 0.3037,
|
| 194 |
+
"step": 3100
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 1.0,
|
| 198 |
+
"learning_rate": 9.599999999999999e-05,
|
| 199 |
+
"loss": 0.306,
|
| 200 |
+
"step": 3200
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.03,
|
| 204 |
+
"learning_rate": 9.9e-05,
|
| 205 |
+
"loss": 0.2986,
|
| 206 |
+
"step": 3300
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.07,
|
| 210 |
+
"learning_rate": 0.000102,
|
| 211 |
+
"loss": 0.2994,
|
| 212 |
+
"step": 3400
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.1,
|
| 216 |
+
"learning_rate": 0.00010499999999999999,
|
| 217 |
+
"loss": 0.2966,
|
| 218 |
+
"step": 3500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 1.13,
|
| 222 |
+
"learning_rate": 0.00010799999999999998,
|
| 223 |
+
"loss": 0.299,
|
| 224 |
+
"step": 3600
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 1.16,
|
| 228 |
+
"learning_rate": 0.00011099999999999999,
|
| 229 |
+
"loss": 0.2994,
|
| 230 |
+
"step": 3700
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 1.19,
|
| 234 |
+
"learning_rate": 0.00011399999999999999,
|
| 235 |
+
"loss": 0.3021,
|
| 236 |
+
"step": 3800
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 1.22,
|
| 240 |
+
"learning_rate": 0.000117,
|
| 241 |
+
"loss": 0.3037,
|
| 242 |
+
"step": 3900
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.25,
|
| 246 |
+
"learning_rate": 0.00011999999999999999,
|
| 247 |
+
"loss": 0.3035,
|
| 248 |
+
"step": 4000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 1.28,
|
| 252 |
+
"learning_rate": 0.00012299999999999998,
|
| 253 |
+
"loss": 0.307,
|
| 254 |
+
"step": 4100
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.32,
|
| 258 |
+
"learning_rate": 0.00012599999999999997,
|
| 259 |
+
"loss": 0.3035,
|
| 260 |
+
"step": 4200
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 1.35,
|
| 264 |
+
"learning_rate": 0.000129,
|
| 265 |
+
"loss": 0.3038,
|
| 266 |
+
"step": 4300
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 1.38,
|
| 270 |
+
"learning_rate": 0.00013199999999999998,
|
| 271 |
+
"loss": 0.3063,
|
| 272 |
+
"step": 4400
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 1.41,
|
| 276 |
+
"learning_rate": 0.000135,
|
| 277 |
+
"loss": 0.3072,
|
| 278 |
+
"step": 4500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 1.44,
|
| 282 |
+
"learning_rate": 0.000138,
|
| 283 |
+
"loss": 0.3093,
|
| 284 |
+
"step": 4600
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 1.47,
|
| 288 |
+
"learning_rate": 0.00014099999999999998,
|
| 289 |
+
"loss": 0.3142,
|
| 290 |
+
"step": 4700
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.5,
|
| 294 |
+
"learning_rate": 0.00014399999999999998,
|
| 295 |
+
"loss": 0.3097,
|
| 296 |
+
"step": 4800
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 1.54,
|
| 300 |
+
"learning_rate": 0.000147,
|
| 301 |
+
"loss": 0.309,
|
| 302 |
+
"step": 4900
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 1.57,
|
| 306 |
+
"learning_rate": 0.00015,
|
| 307 |
+
"loss": 0.3116,
|
| 308 |
+
"step": 5000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 1.57,
|
| 312 |
+
"eval_loss": 0.5301366448402405,
|
| 313 |
+
"eval_runtime": 103.8143,
|
| 314 |
+
"eval_samples_per_second": 86.125,
|
| 315 |
+
"eval_steps_per_second": 1.349,
|
| 316 |
+
"step": 5000
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 1.6,
|
| 320 |
+
"learning_rate": 0.0001497450713800136,
|
| 321 |
+
"loss": 0.3111,
|
| 322 |
+
"step": 5100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 1.63,
|
| 326 |
+
"learning_rate": 0.00014949014276002717,
|
| 327 |
+
"loss": 0.3085,
|
| 328 |
+
"step": 5200
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 1.66,
|
| 332 |
+
"learning_rate": 0.00014923521414004077,
|
| 333 |
+
"loss": 0.306,
|
| 334 |
+
"step": 5300
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 1.69,
|
| 338 |
+
"learning_rate": 0.00014898028552005438,
|
| 339 |
+
"loss": 0.3046,
|
| 340 |
+
"step": 5400
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 1.72,
|
| 344 |
+
"learning_rate": 0.00014872535690006795,
|
| 345 |
+
"loss": 0.3016,
|
| 346 |
+
"step": 5500
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 1.75,
|
| 350 |
+
"learning_rate": 0.00014847042828008156,
|
| 351 |
+
"loss": 0.2954,
|
| 352 |
+
"step": 5600
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 1.79,
|
| 356 |
+
"learning_rate": 0.00014821549966009516,
|
| 357 |
+
"loss": 0.2919,
|
| 358 |
+
"step": 5700
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 1.82,
|
| 362 |
+
"learning_rate": 0.00014796057104010877,
|
| 363 |
+
"loss": 0.293,
|
| 364 |
+
"step": 5800
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 1.85,
|
| 368 |
+
"learning_rate": 0.00014770564242012234,
|
| 369 |
+
"loss": 0.2893,
|
| 370 |
+
"step": 5900
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 1.88,
|
| 374 |
+
"learning_rate": 0.00014745071380013595,
|
| 375 |
+
"loss": 0.2824,
|
| 376 |
+
"step": 6000
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 1.91,
|
| 380 |
+
"learning_rate": 0.00014719578518014955,
|
| 381 |
+
"loss": 0.2868,
|
| 382 |
+
"step": 6100
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 1.94,
|
| 386 |
+
"learning_rate": 0.00014694085656016313,
|
| 387 |
+
"loss": 0.2816,
|
| 388 |
+
"step": 6200
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 1.97,
|
| 392 |
+
"learning_rate": 0.00014668592794017673,
|
| 393 |
+
"loss": 0.2765,
|
| 394 |
+
"step": 6300
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 2.01,
|
| 398 |
+
"learning_rate": 0.00014643099932019034,
|
| 399 |
+
"loss": 0.2782,
|
| 400 |
+
"step": 6400
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 2.04,
|
| 404 |
+
"learning_rate": 0.00014617607070020392,
|
| 405 |
+
"loss": 0.2669,
|
| 406 |
+
"step": 6500
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 2.07,
|
| 410 |
+
"learning_rate": 0.00014592114208021755,
|
| 411 |
+
"loss": 0.2622,
|
| 412 |
+
"step": 6600
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 2.1,
|
| 416 |
+
"learning_rate": 0.00014566621346023112,
|
| 417 |
+
"loss": 0.2651,
|
| 418 |
+
"step": 6700
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 2.13,
|
| 422 |
+
"learning_rate": 0.00014541128484024473,
|
| 423 |
+
"loss": 0.2624,
|
| 424 |
+
"step": 6800
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 2.16,
|
| 428 |
+
"learning_rate": 0.0001451563562202583,
|
| 429 |
+
"loss": 0.2581,
|
| 430 |
+
"step": 6900
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 2.19,
|
| 434 |
+
"learning_rate": 0.0001449014276002719,
|
| 435 |
+
"loss": 0.2604,
|
| 436 |
+
"step": 7000
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 2.22,
|
| 440 |
+
"learning_rate": 0.00014464649898028551,
|
| 441 |
+
"loss": 0.2531,
|
| 442 |
+
"step": 7100
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 2.26,
|
| 446 |
+
"learning_rate": 0.0001443915703602991,
|
| 447 |
+
"loss": 0.2584,
|
| 448 |
+
"step": 7200
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 2.29,
|
| 452 |
+
"learning_rate": 0.0001441366417403127,
|
| 453 |
+
"loss": 0.2528,
|
| 454 |
+
"step": 7300
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 2.32,
|
| 458 |
+
"learning_rate": 0.0001438817131203263,
|
| 459 |
+
"loss": 0.2522,
|
| 460 |
+
"step": 7400
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 2.35,
|
| 464 |
+
"learning_rate": 0.00014362678450033988,
|
| 465 |
+
"loss": 0.248,
|
| 466 |
+
"step": 7500
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 2.38,
|
| 470 |
+
"learning_rate": 0.00014337185588035348,
|
| 471 |
+
"loss": 0.2503,
|
| 472 |
+
"step": 7600
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 2.41,
|
| 476 |
+
"learning_rate": 0.00014311692726036708,
|
| 477 |
+
"loss": 0.2491,
|
| 478 |
+
"step": 7700
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 2.44,
|
| 482 |
+
"learning_rate": 0.0001428619986403807,
|
| 483 |
+
"loss": 0.2468,
|
| 484 |
+
"step": 7800
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 2.48,
|
| 488 |
+
"learning_rate": 0.0001426070700203943,
|
| 489 |
+
"loss": 0.2429,
|
| 490 |
+
"step": 7900
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 2.51,
|
| 494 |
+
"learning_rate": 0.00014235214140040787,
|
| 495 |
+
"loss": 0.2443,
|
| 496 |
+
"step": 8000
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 2.54,
|
| 500 |
+
"learning_rate": 0.00014209721278042147,
|
| 501 |
+
"loss": 0.2432,
|
| 502 |
+
"step": 8100
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 2.57,
|
| 506 |
+
"learning_rate": 0.00014184228416043508,
|
| 507 |
+
"loss": 0.2395,
|
| 508 |
+
"step": 8200
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 2.6,
|
| 512 |
+
"learning_rate": 0.00014158735554044866,
|
| 513 |
+
"loss": 0.2356,
|
| 514 |
+
"step": 8300
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 2.63,
|
| 518 |
+
"learning_rate": 0.00014133242692046226,
|
| 519 |
+
"loss": 0.2394,
|
| 520 |
+
"step": 8400
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 2.66,
|
| 524 |
+
"learning_rate": 0.00014107749830047584,
|
| 525 |
+
"loss": 0.2354,
|
| 526 |
+
"step": 8500
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 2.69,
|
| 530 |
+
"learning_rate": 0.00014082256968048944,
|
| 531 |
+
"loss": 0.2369,
|
| 532 |
+
"step": 8600
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 2.73,
|
| 536 |
+
"learning_rate": 0.00014056764106050305,
|
| 537 |
+
"loss": 0.2323,
|
| 538 |
+
"step": 8700
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 2.76,
|
| 542 |
+
"learning_rate": 0.00014031271244051665,
|
| 543 |
+
"loss": 0.233,
|
| 544 |
+
"step": 8800
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"epoch": 2.79,
|
| 548 |
+
"learning_rate": 0.00014005778382053025,
|
| 549 |
+
"loss": 0.2324,
|
| 550 |
+
"step": 8900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 2.82,
|
| 554 |
+
"learning_rate": 0.00013980285520054383,
|
| 555 |
+
"loss": 0.2335,
|
| 556 |
+
"step": 9000
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 2.85,
|
| 560 |
+
"learning_rate": 0.00013954792658055744,
|
| 561 |
+
"loss": 0.2311,
|
| 562 |
+
"step": 9100
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 2.88,
|
| 566 |
+
"learning_rate": 0.00013929299796057104,
|
| 567 |
+
"loss": 0.2302,
|
| 568 |
+
"step": 9200
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 2.91,
|
| 572 |
+
"learning_rate": 0.00013903806934058462,
|
| 573 |
+
"loss": 0.2264,
|
| 574 |
+
"step": 9300
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 2.95,
|
| 578 |
+
"learning_rate": 0.00013878314072059822,
|
| 579 |
+
"loss": 0.226,
|
| 580 |
+
"step": 9400
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 2.98,
|
| 584 |
+
"learning_rate": 0.00013852821210061183,
|
| 585 |
+
"loss": 0.2214,
|
| 586 |
+
"step": 9500
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 3.01,
|
| 590 |
+
"learning_rate": 0.0001382732834806254,
|
| 591 |
+
"loss": 0.2218,
|
| 592 |
+
"step": 9600
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 3.04,
|
| 596 |
+
"learning_rate": 0.000138018354860639,
|
| 597 |
+
"loss": 0.2141,
|
| 598 |
+
"step": 9700
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 3.07,
|
| 602 |
+
"learning_rate": 0.0001377634262406526,
|
| 603 |
+
"loss": 0.2132,
|
| 604 |
+
"step": 9800
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 3.1,
|
| 608 |
+
"learning_rate": 0.00013750849762066622,
|
| 609 |
+
"loss": 0.2131,
|
| 610 |
+
"step": 9900
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 3.13,
|
| 614 |
+
"learning_rate": 0.0001372535690006798,
|
| 615 |
+
"loss": 0.2104,
|
| 616 |
+
"step": 10000
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 3.13,
|
| 620 |
+
"eval_loss": 0.40657177567481995,
|
| 621 |
+
"eval_runtime": 67.9841,
|
| 622 |
+
"eval_samples_per_second": 131.516,
|
| 623 |
+
"eval_steps_per_second": 2.059,
|
| 624 |
+
"step": 10000
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 3.16,
|
| 628 |
+
"learning_rate": 0.0001369986403806934,
|
| 629 |
+
"loss": 0.2127,
|
| 630 |
+
"step": 10100
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 3.2,
|
| 634 |
+
"learning_rate": 0.000136743711760707,
|
| 635 |
+
"loss": 0.2071,
|
| 636 |
+
"step": 10200
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 3.23,
|
| 640 |
+
"learning_rate": 0.00013648878314072058,
|
| 641 |
+
"loss": 0.2125,
|
| 642 |
+
"step": 10300
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 3.26,
|
| 646 |
+
"learning_rate": 0.00013623385452073418,
|
| 647 |
+
"loss": 0.2089,
|
| 648 |
+
"step": 10400
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 3.29,
|
| 652 |
+
"learning_rate": 0.0001359789259007478,
|
| 653 |
+
"loss": 0.209,
|
| 654 |
+
"step": 10500
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 3.32,
|
| 658 |
+
"learning_rate": 0.00013572399728076136,
|
| 659 |
+
"loss": 0.2109,
|
| 660 |
+
"step": 10600
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 3.35,
|
| 664 |
+
"learning_rate": 0.00013546906866077497,
|
| 665 |
+
"loss": 0.2105,
|
| 666 |
+
"step": 10700
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 3.38,
|
| 670 |
+
"learning_rate": 0.00013521414004078857,
|
| 671 |
+
"loss": 0.2092,
|
| 672 |
+
"step": 10800
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 3.42,
|
| 676 |
+
"learning_rate": 0.00013495921142080218,
|
| 677 |
+
"loss": 0.2093,
|
| 678 |
+
"step": 10900
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 3.45,
|
| 682 |
+
"learning_rate": 0.00013470428280081575,
|
| 683 |
+
"loss": 0.206,
|
| 684 |
+
"step": 11000
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 3.48,
|
| 688 |
+
"learning_rate": 0.00013444935418082936,
|
| 689 |
+
"loss": 0.2032,
|
| 690 |
+
"step": 11100
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 3.51,
|
| 694 |
+
"learning_rate": 0.00013419442556084296,
|
| 695 |
+
"loss": 0.204,
|
| 696 |
+
"step": 11200
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 3.54,
|
| 700 |
+
"learning_rate": 0.00013393949694085654,
|
| 701 |
+
"loss": 0.2005,
|
| 702 |
+
"step": 11300
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 3.57,
|
| 706 |
+
"learning_rate": 0.00013368456832087014,
|
| 707 |
+
"loss": 0.1988,
|
| 708 |
+
"step": 11400
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 3.6,
|
| 712 |
+
"learning_rate": 0.00013342963970088375,
|
| 713 |
+
"loss": 0.1981,
|
| 714 |
+
"step": 11500
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 3.63,
|
| 718 |
+
"learning_rate": 0.00013317471108089733,
|
| 719 |
+
"loss": 0.2017,
|
| 720 |
+
"step": 11600
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 3.67,
|
| 724 |
+
"learning_rate": 0.00013291978246091093,
|
| 725 |
+
"loss": 0.1989,
|
| 726 |
+
"step": 11700
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 3.7,
|
| 730 |
+
"learning_rate": 0.00013266485384092453,
|
| 731 |
+
"loss": 0.2011,
|
| 732 |
+
"step": 11800
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 3.73,
|
| 736 |
+
"learning_rate": 0.0001324099252209381,
|
| 737 |
+
"loss": 0.197,
|
| 738 |
+
"step": 11900
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 3.76,
|
| 742 |
+
"learning_rate": 0.00013215499660095172,
|
| 743 |
+
"loss": 0.1975,
|
| 744 |
+
"step": 12000
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 3.79,
|
| 748 |
+
"learning_rate": 0.00013190006798096532,
|
| 749 |
+
"loss": 0.1982,
|
| 750 |
+
"step": 12100
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 3.82,
|
| 754 |
+
"learning_rate": 0.00013164513936097892,
|
| 755 |
+
"loss": 0.1983,
|
| 756 |
+
"step": 12200
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 3.85,
|
| 760 |
+
"learning_rate": 0.0001313902107409925,
|
| 761 |
+
"loss": 0.1992,
|
| 762 |
+
"step": 12300
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 3.89,
|
| 766 |
+
"learning_rate": 0.0001311352821210061,
|
| 767 |
+
"loss": 0.1943,
|
| 768 |
+
"step": 12400
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 3.92,
|
| 772 |
+
"learning_rate": 0.0001308803535010197,
|
| 773 |
+
"loss": 0.1958,
|
| 774 |
+
"step": 12500
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 3.95,
|
| 778 |
+
"learning_rate": 0.0001306254248810333,
|
| 779 |
+
"loss": 0.1927,
|
| 780 |
+
"step": 12600
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 3.98,
|
| 784 |
+
"learning_rate": 0.0001303704962610469,
|
| 785 |
+
"loss": 0.1927,
|
| 786 |
+
"step": 12700
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 4.01,
|
| 790 |
+
"learning_rate": 0.0001301155676410605,
|
| 791 |
+
"loss": 0.1882,
|
| 792 |
+
"step": 12800
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 4.04,
|
| 796 |
+
"learning_rate": 0.00012986063902107407,
|
| 797 |
+
"loss": 0.1854,
|
| 798 |
+
"step": 12900
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 4.07,
|
| 802 |
+
"learning_rate": 0.0001296057104010877,
|
| 803 |
+
"loss": 0.1842,
|
| 804 |
+
"step": 13000
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 4.1,
|
| 808 |
+
"learning_rate": 0.00012935078178110128,
|
| 809 |
+
"loss": 0.1846,
|
| 810 |
+
"step": 13100
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 4.14,
|
| 814 |
+
"learning_rate": 0.00012909585316111489,
|
| 815 |
+
"loss": 0.1834,
|
| 816 |
+
"step": 13200
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 4.17,
|
| 820 |
+
"learning_rate": 0.0001288409245411285,
|
| 821 |
+
"loss": 0.1833,
|
| 822 |
+
"step": 13300
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 4.2,
|
| 826 |
+
"learning_rate": 0.00012858599592114207,
|
| 827 |
+
"loss": 0.1826,
|
| 828 |
+
"step": 13400
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 4.23,
|
| 832 |
+
"learning_rate": 0.00012833106730115567,
|
| 833 |
+
"loss": 0.1807,
|
| 834 |
+
"step": 13500
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 4.26,
|
| 838 |
+
"learning_rate": 0.00012807613868116925,
|
| 839 |
+
"loss": 0.1805,
|
| 840 |
+
"step": 13600
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 4.29,
|
| 844 |
+
"learning_rate": 0.00012782121006118285,
|
| 845 |
+
"loss": 0.1808,
|
| 846 |
+
"step": 13700
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 4.32,
|
| 850 |
+
"learning_rate": 0.00012756628144119646,
|
| 851 |
+
"loss": 0.1791,
|
| 852 |
+
"step": 13800
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 4.36,
|
| 856 |
+
"learning_rate": 0.00012731135282121003,
|
| 857 |
+
"loss": 0.181,
|
| 858 |
+
"step": 13900
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 4.39,
|
| 862 |
+
"learning_rate": 0.00012705642420122364,
|
| 863 |
+
"loss": 0.1816,
|
| 864 |
+
"step": 14000
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 4.42,
|
| 868 |
+
"learning_rate": 0.00012680149558123724,
|
| 869 |
+
"loss": 0.1805,
|
| 870 |
+
"step": 14100
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 4.45,
|
| 874 |
+
"learning_rate": 0.00012654656696125085,
|
| 875 |
+
"loss": 0.1784,
|
| 876 |
+
"step": 14200
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 4.48,
|
| 880 |
+
"learning_rate": 0.00012629163834126445,
|
| 881 |
+
"loss": 0.1767,
|
| 882 |
+
"step": 14300
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 4.51,
|
| 886 |
+
"learning_rate": 0.00012603670972127803,
|
| 887 |
+
"loss": 0.18,
|
| 888 |
+
"step": 14400
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 4.54,
|
| 892 |
+
"learning_rate": 0.00012578178110129163,
|
| 893 |
+
"loss": 0.1773,
|
| 894 |
+
"step": 14500
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 4.57,
|
| 898 |
+
"learning_rate": 0.00012552685248130524,
|
| 899 |
+
"loss": 0.1794,
|
| 900 |
+
"step": 14600
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 4.61,
|
| 904 |
+
"learning_rate": 0.00012527192386131881,
|
| 905 |
+
"loss": 0.1776,
|
| 906 |
+
"step": 14700
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 4.64,
|
| 910 |
+
"learning_rate": 0.00012501699524133242,
|
| 911 |
+
"loss": 0.1749,
|
| 912 |
+
"step": 14800
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 4.67,
|
| 916 |
+
"learning_rate": 0.000124762066621346,
|
| 917 |
+
"loss": 0.1746,
|
| 918 |
+
"step": 14900
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 4.7,
|
| 922 |
+
"learning_rate": 0.0001245071380013596,
|
| 923 |
+
"loss": 0.1729,
|
| 924 |
+
"step": 15000
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 4.7,
|
| 928 |
+
"eval_loss": 0.3554973900318146,
|
| 929 |
+
"eval_runtime": 74.4919,
|
| 930 |
+
"eval_samples_per_second": 120.026,
|
| 931 |
+
"eval_steps_per_second": 1.879,
|
| 932 |
+
"step": 15000
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 4.73,
|
| 936 |
+
"learning_rate": 0.0001242522093813732,
|
| 937 |
+
"loss": 0.1768,
|
| 938 |
+
"step": 15100
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 4.76,
|
| 942 |
+
"learning_rate": 0.0001239972807613868,
|
| 943 |
+
"loss": 0.1727,
|
| 944 |
+
"step": 15200
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 4.79,
|
| 948 |
+
"learning_rate": 0.0001237423521414004,
|
| 949 |
+
"loss": 0.1742,
|
| 950 |
+
"step": 15300
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 4.83,
|
| 954 |
+
"learning_rate": 0.000123487423521414,
|
| 955 |
+
"loss": 0.1697,
|
| 956 |
+
"step": 15400
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 4.86,
|
| 960 |
+
"learning_rate": 0.0001232324949014276,
|
| 961 |
+
"loss": 0.1755,
|
| 962 |
+
"step": 15500
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 4.89,
|
| 966 |
+
"learning_rate": 0.0001229775662814412,
|
| 967 |
+
"loss": 0.1709,
|
| 968 |
+
"step": 15600
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 4.92,
|
| 972 |
+
"learning_rate": 0.00012272263766145478,
|
| 973 |
+
"loss": 0.1735,
|
| 974 |
+
"step": 15700
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 4.95,
|
| 978 |
+
"learning_rate": 0.00012246770904146838,
|
| 979 |
+
"loss": 0.1692,
|
| 980 |
+
"step": 15800
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 4.98,
|
| 984 |
+
"learning_rate": 0.00012221278042148198,
|
| 985 |
+
"loss": 0.1706,
|
| 986 |
+
"step": 15900
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 5.01,
|
| 990 |
+
"learning_rate": 0.00012195785180149556,
|
| 991 |
+
"loss": 0.1665,
|
| 992 |
+
"step": 16000
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 5.04,
|
| 996 |
+
"learning_rate": 0.00012170292318150917,
|
| 997 |
+
"loss": 0.1601,
|
| 998 |
+
"step": 16100
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 5.08,
|
| 1002 |
+
"learning_rate": 0.00012144799456152277,
|
| 1003 |
+
"loss": 0.1619,
|
| 1004 |
+
"step": 16200
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 5.11,
|
| 1008 |
+
"learning_rate": 0.00012119306594153636,
|
| 1009 |
+
"loss": 0.1615,
|
| 1010 |
+
"step": 16300
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 5.14,
|
| 1014 |
+
"learning_rate": 0.00012093813732154996,
|
| 1015 |
+
"loss": 0.1613,
|
| 1016 |
+
"step": 16400
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 5.17,
|
| 1020 |
+
"learning_rate": 0.00012068320870156356,
|
| 1021 |
+
"loss": 0.1603,
|
| 1022 |
+
"step": 16500
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 5.2,
|
| 1026 |
+
"learning_rate": 0.00012042828008157715,
|
| 1027 |
+
"loss": 0.1611,
|
| 1028 |
+
"step": 16600
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 5.23,
|
| 1032 |
+
"learning_rate": 0.00012017335146159075,
|
| 1033 |
+
"loss": 0.16,
|
| 1034 |
+
"step": 16700
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 5.26,
|
| 1038 |
+
"learning_rate": 0.00011991842284160434,
|
| 1039 |
+
"loss": 0.1616,
|
| 1040 |
+
"step": 16800
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 5.3,
|
| 1044 |
+
"learning_rate": 0.00011966349422161793,
|
| 1045 |
+
"loss": 0.1628,
|
| 1046 |
+
"step": 16900
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 5.33,
|
| 1050 |
+
"learning_rate": 0.00011940856560163152,
|
| 1051 |
+
"loss": 0.1652,
|
| 1052 |
+
"step": 17000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 5.36,
|
| 1056 |
+
"learning_rate": 0.00011915363698164513,
|
| 1057 |
+
"loss": 0.1593,
|
| 1058 |
+
"step": 17100
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 5.39,
|
| 1062 |
+
"learning_rate": 0.00011889870836165872,
|
| 1063 |
+
"loss": 0.1589,
|
| 1064 |
+
"step": 17200
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 5.42,
|
| 1068 |
+
"learning_rate": 0.00011864377974167233,
|
| 1069 |
+
"loss": 0.1593,
|
| 1070 |
+
"step": 17300
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 5.45,
|
| 1074 |
+
"learning_rate": 0.00011838885112168593,
|
| 1075 |
+
"loss": 0.159,
|
| 1076 |
+
"step": 17400
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 5.48,
|
| 1080 |
+
"learning_rate": 0.00011813392250169952,
|
| 1081 |
+
"loss": 0.1599,
|
| 1082 |
+
"step": 17500
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 5.51,
|
| 1086 |
+
"learning_rate": 0.00011787899388171311,
|
| 1087 |
+
"loss": 0.1603,
|
| 1088 |
+
"step": 17600
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 5.55,
|
| 1092 |
+
"learning_rate": 0.00011762406526172671,
|
| 1093 |
+
"loss": 0.1549,
|
| 1094 |
+
"step": 17700
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 5.58,
|
| 1098 |
+
"learning_rate": 0.0001173691366417403,
|
| 1099 |
+
"loss": 0.1574,
|
| 1100 |
+
"step": 17800
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 5.61,
|
| 1104 |
+
"learning_rate": 0.00011711420802175389,
|
| 1105 |
+
"loss": 0.1575,
|
| 1106 |
+
"step": 17900
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 5.64,
|
| 1110 |
+
"learning_rate": 0.0001168592794017675,
|
| 1111 |
+
"loss": 0.1583,
|
| 1112 |
+
"step": 18000
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 5.67,
|
| 1116 |
+
"learning_rate": 0.00011660435078178109,
|
| 1117 |
+
"loss": 0.1542,
|
| 1118 |
+
"step": 18100
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 5.7,
|
| 1122 |
+
"learning_rate": 0.00011634942216179468,
|
| 1123 |
+
"loss": 0.1559,
|
| 1124 |
+
"step": 18200
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 5.73,
|
| 1128 |
+
"learning_rate": 0.00011609449354180827,
|
| 1129 |
+
"loss": 0.157,
|
| 1130 |
+
"step": 18300
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 5.77,
|
| 1134 |
+
"learning_rate": 0.00011583956492182189,
|
| 1135 |
+
"loss": 0.1568,
|
| 1136 |
+
"step": 18400
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 5.8,
|
| 1140 |
+
"learning_rate": 0.00011558463630183548,
|
| 1141 |
+
"loss": 0.155,
|
| 1142 |
+
"step": 18500
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 5.83,
|
| 1146 |
+
"learning_rate": 0.00011532970768184908,
|
| 1147 |
+
"loss": 0.1571,
|
| 1148 |
+
"step": 18600
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 5.86,
|
| 1152 |
+
"learning_rate": 0.00011507477906186267,
|
| 1153 |
+
"loss": 0.1526,
|
| 1154 |
+
"step": 18700
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 5.89,
|
| 1158 |
+
"learning_rate": 0.00011481985044187626,
|
| 1159 |
+
"loss": 0.1513,
|
| 1160 |
+
"step": 18800
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 5.92,
|
| 1164 |
+
"learning_rate": 0.00011456492182188985,
|
| 1165 |
+
"loss": 0.1556,
|
| 1166 |
+
"step": 18900
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 5.95,
|
| 1170 |
+
"learning_rate": 0.00011430999320190346,
|
| 1171 |
+
"loss": 0.1542,
|
| 1172 |
+
"step": 19000
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 5.98,
|
| 1176 |
+
"learning_rate": 0.00011405506458191705,
|
| 1177 |
+
"loss": 0.1519,
|
| 1178 |
+
"step": 19100
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 6.02,
|
| 1182 |
+
"learning_rate": 0.00011380013596193064,
|
| 1183 |
+
"loss": 0.1488,
|
| 1184 |
+
"step": 19200
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 6.05,
|
| 1188 |
+
"learning_rate": 0.00011354520734194424,
|
| 1189 |
+
"loss": 0.1453,
|
| 1190 |
+
"step": 19300
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 6.08,
|
| 1194 |
+
"learning_rate": 0.00011329027872195785,
|
| 1195 |
+
"loss": 0.1458,
|
| 1196 |
+
"step": 19400
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 6.11,
|
| 1200 |
+
"learning_rate": 0.00011303535010197144,
|
| 1201 |
+
"loss": 0.1431,
|
| 1202 |
+
"step": 19500
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 6.14,
|
| 1206 |
+
"learning_rate": 0.00011278042148198504,
|
| 1207 |
+
"loss": 0.1453,
|
| 1208 |
+
"step": 19600
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 6.17,
|
| 1212 |
+
"learning_rate": 0.00011252549286199863,
|
| 1213 |
+
"loss": 0.1416,
|
| 1214 |
+
"step": 19700
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 6.2,
|
| 1218 |
+
"learning_rate": 0.00011227056424201222,
|
| 1219 |
+
"loss": 0.1465,
|
| 1220 |
+
"step": 19800
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 6.24,
|
| 1224 |
+
"learning_rate": 0.00011201563562202583,
|
| 1225 |
+
"loss": 0.1451,
|
| 1226 |
+
"step": 19900
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 6.27,
|
| 1230 |
+
"learning_rate": 0.00011176070700203942,
|
| 1231 |
+
"loss": 0.1472,
|
| 1232 |
+
"step": 20000
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 6.27,
|
| 1236 |
+
"eval_loss": 0.3207816481590271,
|
| 1237 |
+
"eval_runtime": 71.6607,
|
| 1238 |
+
"eval_samples_per_second": 124.769,
|
| 1239 |
+
"eval_steps_per_second": 1.954,
|
| 1240 |
+
"step": 20000
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 6.3,
|
| 1244 |
+
"learning_rate": 0.00011150577838205301,
|
| 1245 |
+
"loss": 0.143,
|
| 1246 |
+
"step": 20100
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 6.33,
|
| 1250 |
+
"learning_rate": 0.0001112508497620666,
|
| 1251 |
+
"loss": 0.1418,
|
| 1252 |
+
"step": 20200
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 6.36,
|
| 1256 |
+
"learning_rate": 0.0001109959211420802,
|
| 1257 |
+
"loss": 0.1461,
|
| 1258 |
+
"step": 20300
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 6.39,
|
| 1262 |
+
"learning_rate": 0.0001107409925220938,
|
| 1263 |
+
"loss": 0.1414,
|
| 1264 |
+
"step": 20400
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"epoch": 6.42,
|
| 1268 |
+
"learning_rate": 0.00011048606390210741,
|
| 1269 |
+
"loss": 0.1418,
|
| 1270 |
+
"step": 20500
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 6.45,
|
| 1274 |
+
"learning_rate": 0.000110231135282121,
|
| 1275 |
+
"loss": 0.144,
|
| 1276 |
+
"step": 20600
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 6.49,
|
| 1280 |
+
"learning_rate": 0.0001099762066621346,
|
| 1281 |
+
"loss": 0.1427,
|
| 1282 |
+
"step": 20700
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 6.52,
|
| 1286 |
+
"learning_rate": 0.00010972127804214819,
|
| 1287 |
+
"loss": 0.1434,
|
| 1288 |
+
"step": 20800
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 6.55,
|
| 1292 |
+
"learning_rate": 0.00010946634942216179,
|
| 1293 |
+
"loss": 0.1437,
|
| 1294 |
+
"step": 20900
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 6.58,
|
| 1298 |
+
"learning_rate": 0.00010921142080217538,
|
| 1299 |
+
"loss": 0.1426,
|
| 1300 |
+
"step": 21000
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 6.61,
|
| 1304 |
+
"learning_rate": 0.00010895649218218897,
|
| 1305 |
+
"loss": 0.1403,
|
| 1306 |
+
"step": 21100
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 6.64,
|
| 1310 |
+
"learning_rate": 0.00010870156356220258,
|
| 1311 |
+
"loss": 0.1399,
|
| 1312 |
+
"step": 21200
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 6.67,
|
| 1316 |
+
"learning_rate": 0.00010844663494221617,
|
| 1317 |
+
"loss": 0.1454,
|
| 1318 |
+
"step": 21300
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 6.71,
|
| 1322 |
+
"learning_rate": 0.00010819170632222976,
|
| 1323 |
+
"loss": 0.1402,
|
| 1324 |
+
"step": 21400
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 6.74,
|
| 1328 |
+
"learning_rate": 0.00010793677770224335,
|
| 1329 |
+
"loss": 0.1403,
|
| 1330 |
+
"step": 21500
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 6.77,
|
| 1334 |
+
"learning_rate": 0.00010768184908225697,
|
| 1335 |
+
"loss": 0.1401,
|
| 1336 |
+
"step": 21600
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 6.8,
|
| 1340 |
+
"learning_rate": 0.00010742692046227056,
|
| 1341 |
+
"loss": 0.1408,
|
| 1342 |
+
"step": 21700
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 6.83,
|
| 1346 |
+
"learning_rate": 0.00010717199184228416,
|
| 1347 |
+
"loss": 0.1388,
|
| 1348 |
+
"step": 21800
|
| 1349 |
+
},
|
| 1350 |
+
{
|
| 1351 |
+
"epoch": 6.86,
|
| 1352 |
+
"learning_rate": 0.00010691706322229775,
|
| 1353 |
+
"loss": 0.1431,
|
| 1354 |
+
"step": 21900
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 6.89,
|
| 1358 |
+
"learning_rate": 0.00010666213460231134,
|
| 1359 |
+
"loss": 0.1384,
|
| 1360 |
+
"step": 22000
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 6.92,
|
| 1364 |
+
"learning_rate": 0.00010640720598232493,
|
| 1365 |
+
"loss": 0.1381,
|
| 1366 |
+
"step": 22100
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 6.96,
|
| 1370 |
+
"learning_rate": 0.00010615227736233854,
|
| 1371 |
+
"loss": 0.1367,
|
| 1372 |
+
"step": 22200
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"epoch": 6.99,
|
| 1376 |
+
"learning_rate": 0.00010589734874235213,
|
| 1377 |
+
"loss": 0.1424,
|
| 1378 |
+
"step": 22300
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 7.02,
|
| 1382 |
+
"learning_rate": 0.00010564242012236572,
|
| 1383 |
+
"loss": 0.1347,
|
| 1384 |
+
"step": 22400
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 7.05,
|
| 1388 |
+
"learning_rate": 0.00010538749150237932,
|
| 1389 |
+
"loss": 0.1297,
|
| 1390 |
+
"step": 22500
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 7.08,
|
| 1394 |
+
"learning_rate": 0.00010513256288239293,
|
| 1395 |
+
"loss": 0.1323,
|
| 1396 |
+
"step": 22600
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 7.11,
|
| 1400 |
+
"learning_rate": 0.00010487763426240652,
|
| 1401 |
+
"loss": 0.1292,
|
| 1402 |
+
"step": 22700
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 7.14,
|
| 1406 |
+
"learning_rate": 0.00010462270564242012,
|
| 1407 |
+
"loss": 0.1311,
|
| 1408 |
+
"step": 22800
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 7.18,
|
| 1412 |
+
"learning_rate": 0.00010436777702243371,
|
| 1413 |
+
"loss": 0.1328,
|
| 1414 |
+
"step": 22900
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 7.21,
|
| 1418 |
+
"learning_rate": 0.0001041128484024473,
|
| 1419 |
+
"loss": 0.1277,
|
| 1420 |
+
"step": 23000
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 7.24,
|
| 1424 |
+
"learning_rate": 0.00010385791978246091,
|
| 1425 |
+
"loss": 0.129,
|
| 1426 |
+
"step": 23100
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 7.27,
|
| 1430 |
+
"learning_rate": 0.0001036029911624745,
|
| 1431 |
+
"loss": 0.1304,
|
| 1432 |
+
"step": 23200
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 7.3,
|
| 1436 |
+
"learning_rate": 0.00010334806254248809,
|
| 1437 |
+
"loss": 0.1308,
|
| 1438 |
+
"step": 23300
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 7.33,
|
| 1442 |
+
"learning_rate": 0.00010309313392250168,
|
| 1443 |
+
"loss": 0.1285,
|
| 1444 |
+
"step": 23400
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 7.36,
|
| 1448 |
+
"learning_rate": 0.00010283820530251528,
|
| 1449 |
+
"loss": 0.1289,
|
| 1450 |
+
"step": 23500
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 7.39,
|
| 1454 |
+
"learning_rate": 0.00010258327668252887,
|
| 1455 |
+
"loss": 0.1318,
|
| 1456 |
+
"step": 23600
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"epoch": 7.43,
|
| 1460 |
+
"learning_rate": 0.00010232834806254249,
|
| 1461 |
+
"loss": 0.1291,
|
| 1462 |
+
"step": 23700
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 7.46,
|
| 1466 |
+
"learning_rate": 0.00010207341944255608,
|
| 1467 |
+
"loss": 0.1298,
|
| 1468 |
+
"step": 23800
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 7.49,
|
| 1472 |
+
"learning_rate": 0.00010181849082256967,
|
| 1473 |
+
"loss": 0.1287,
|
| 1474 |
+
"step": 23900
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 7.52,
|
| 1478 |
+
"learning_rate": 0.00010156356220258326,
|
| 1479 |
+
"loss": 0.1293,
|
| 1480 |
+
"step": 24000
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 7.55,
|
| 1484 |
+
"learning_rate": 0.00010130863358259687,
|
| 1485 |
+
"loss": 0.1277,
|
| 1486 |
+
"step": 24100
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 7.58,
|
| 1490 |
+
"learning_rate": 0.00010105370496261046,
|
| 1491 |
+
"loss": 0.1295,
|
| 1492 |
+
"step": 24200
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 7.61,
|
| 1496 |
+
"learning_rate": 0.00010079877634262405,
|
| 1497 |
+
"loss": 0.1289,
|
| 1498 |
+
"step": 24300
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 7.65,
|
| 1502 |
+
"learning_rate": 0.00010054384772263765,
|
| 1503 |
+
"loss": 0.1298,
|
| 1504 |
+
"step": 24400
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"epoch": 7.68,
|
| 1508 |
+
"learning_rate": 0.00010028891910265125,
|
| 1509 |
+
"loss": 0.1287,
|
| 1510 |
+
"step": 24500
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 7.71,
|
| 1514 |
+
"learning_rate": 0.00010003399048266484,
|
| 1515 |
+
"loss": 0.1286,
|
| 1516 |
+
"step": 24600
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 7.74,
|
| 1520 |
+
"learning_rate": 9.977906186267843e-05,
|
| 1521 |
+
"loss": 0.1294,
|
| 1522 |
+
"step": 24700
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 7.77,
|
| 1526 |
+
"learning_rate": 9.952413324269204e-05,
|
| 1527 |
+
"loss": 0.1287,
|
| 1528 |
+
"step": 24800
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 7.8,
|
| 1532 |
+
"learning_rate": 9.926920462270564e-05,
|
| 1533 |
+
"loss": 0.1295,
|
| 1534 |
+
"step": 24900
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 7.83,
|
| 1538 |
+
"learning_rate": 9.901427600271924e-05,
|
| 1539 |
+
"loss": 0.128,
|
| 1540 |
+
"step": 25000
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 7.83,
|
| 1544 |
+
"eval_loss": 0.29230204224586487,
|
| 1545 |
+
"eval_runtime": 77.9206,
|
| 1546 |
+
"eval_samples_per_second": 114.745,
|
| 1547 |
+
"eval_steps_per_second": 1.797,
|
| 1548 |
+
"step": 25000
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 7.86,
|
| 1552 |
+
"learning_rate": 9.875934738273283e-05,
|
| 1553 |
+
"loss": 0.1305,
|
| 1554 |
+
"step": 25100
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 7.9,
|
| 1558 |
+
"learning_rate": 9.850441876274642e-05,
|
| 1559 |
+
"loss": 0.1284,
|
| 1560 |
+
"step": 25200
|
| 1561 |
+
},
|
| 1562 |
+
{
|
| 1563 |
+
"epoch": 7.93,
|
| 1564 |
+
"learning_rate": 9.824949014276001e-05,
|
| 1565 |
+
"loss": 0.1275,
|
| 1566 |
+
"step": 25300
|
| 1567 |
+
},
|
| 1568 |
+
{
|
| 1569 |
+
"epoch": 7.96,
|
| 1570 |
+
"learning_rate": 9.799456152277362e-05,
|
| 1571 |
+
"loss": 0.1277,
|
| 1572 |
+
"step": 25400
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 7.99,
|
| 1576 |
+
"learning_rate": 9.77396329027872e-05,
|
| 1577 |
+
"loss": 0.1252,
|
| 1578 |
+
"step": 25500
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 8.02,
|
| 1582 |
+
"learning_rate": 9.74847042828008e-05,
|
| 1583 |
+
"loss": 0.1195,
|
| 1584 |
+
"step": 25600
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 8.05,
|
| 1588 |
+
"learning_rate": 9.72297756628144e-05,
|
| 1589 |
+
"loss": 0.1195,
|
| 1590 |
+
"step": 25700
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 8.08,
|
| 1594 |
+
"learning_rate": 9.6974847042828e-05,
|
| 1595 |
+
"loss": 0.119,
|
| 1596 |
+
"step": 25800
|
| 1597 |
+
},
|
| 1598 |
+
{
|
| 1599 |
+
"epoch": 8.12,
|
| 1600 |
+
"learning_rate": 9.67199184228416e-05,
|
| 1601 |
+
"loss": 0.1205,
|
| 1602 |
+
"step": 25900
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 8.15,
|
| 1606 |
+
"learning_rate": 9.64649898028552e-05,
|
| 1607 |
+
"loss": 0.1173,
|
| 1608 |
+
"step": 26000
|
| 1609 |
+
},
|
| 1610 |
+
{
|
| 1611 |
+
"epoch": 8.18,
|
| 1612 |
+
"learning_rate": 9.621006118286879e-05,
|
| 1613 |
+
"loss": 0.1175,
|
| 1614 |
+
"step": 26100
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 8.21,
|
| 1618 |
+
"learning_rate": 9.595513256288238e-05,
|
| 1619 |
+
"loss": 0.1188,
|
| 1620 |
+
"step": 26200
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 8.24,
|
| 1624 |
+
"learning_rate": 9.570020394289599e-05,
|
| 1625 |
+
"loss": 0.1177,
|
| 1626 |
+
"step": 26300
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 8.27,
|
| 1630 |
+
"learning_rate": 9.544527532290958e-05,
|
| 1631 |
+
"loss": 0.1185,
|
| 1632 |
+
"step": 26400
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 8.3,
|
| 1636 |
+
"learning_rate": 9.519034670292317e-05,
|
| 1637 |
+
"loss": 0.1199,
|
| 1638 |
+
"step": 26500
|
| 1639 |
+
},
|
| 1640 |
+
{
|
| 1641 |
+
"epoch": 8.33,
|
| 1642 |
+
"learning_rate": 9.493541808293676e-05,
|
| 1643 |
+
"loss": 0.1206,
|
| 1644 |
+
"step": 26600
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 8.37,
|
| 1648 |
+
"learning_rate": 9.468048946295036e-05,
|
| 1649 |
+
"loss": 0.1173,
|
| 1650 |
+
"step": 26700
|
| 1651 |
+
},
|
| 1652 |
+
{
|
| 1653 |
+
"epoch": 8.4,
|
| 1654 |
+
"learning_rate": 9.442556084296395e-05,
|
| 1655 |
+
"loss": 0.1172,
|
| 1656 |
+
"step": 26800
|
| 1657 |
+
},
|
| 1658 |
+
{
|
| 1659 |
+
"epoch": 8.43,
|
| 1660 |
+
"learning_rate": 9.417063222297757e-05,
|
| 1661 |
+
"loss": 0.1173,
|
| 1662 |
+
"step": 26900
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 8.46,
|
| 1666 |
+
"learning_rate": 9.391570360299116e-05,
|
| 1667 |
+
"loss": 0.121,
|
| 1668 |
+
"step": 27000
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 8.49,
|
| 1672 |
+
"learning_rate": 9.366077498300475e-05,
|
| 1673 |
+
"loss": 0.1173,
|
| 1674 |
+
"step": 27100
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 8.52,
|
| 1678 |
+
"learning_rate": 9.340584636301834e-05,
|
| 1679 |
+
"loss": 0.1184,
|
| 1680 |
+
"step": 27200
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 8.55,
|
| 1684 |
+
"learning_rate": 9.315091774303195e-05,
|
| 1685 |
+
"loss": 0.1172,
|
| 1686 |
+
"step": 27300
|
| 1687 |
+
},
|
| 1688 |
+
{
|
| 1689 |
+
"epoch": 8.59,
|
| 1690 |
+
"learning_rate": 9.289598912304554e-05,
|
| 1691 |
+
"loss": 0.118,
|
| 1692 |
+
"step": 27400
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 8.62,
|
| 1696 |
+
"learning_rate": 9.264106050305913e-05,
|
| 1697 |
+
"loss": 0.1179,
|
| 1698 |
+
"step": 27500
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 8.65,
|
| 1702 |
+
"learning_rate": 9.238613188307273e-05,
|
| 1703 |
+
"loss": 0.115,
|
| 1704 |
+
"step": 27600
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 8.68,
|
| 1708 |
+
"learning_rate": 9.213120326308632e-05,
|
| 1709 |
+
"loss": 0.1159,
|
| 1710 |
+
"step": 27700
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 8.71,
|
| 1714 |
+
"learning_rate": 9.187627464309991e-05,
|
| 1715 |
+
"loss": 0.1152,
|
| 1716 |
+
"step": 27800
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 8.74,
|
| 1720 |
+
"learning_rate": 9.162134602311352e-05,
|
| 1721 |
+
"loss": 0.1166,
|
| 1722 |
+
"step": 27900
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 8.77,
|
| 1726 |
+
"learning_rate": 9.136641740312712e-05,
|
| 1727 |
+
"loss": 0.1184,
|
| 1728 |
+
"step": 28000
|
| 1729 |
+
},
|
| 1730 |
+
{
|
| 1731 |
+
"epoch": 8.8,
|
| 1732 |
+
"learning_rate": 9.111148878314071e-05,
|
| 1733 |
+
"loss": 0.1155,
|
| 1734 |
+
"step": 28100
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 8.84,
|
| 1738 |
+
"learning_rate": 9.085656016315432e-05,
|
| 1739 |
+
"loss": 0.1136,
|
| 1740 |
+
"step": 28200
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 8.87,
|
| 1744 |
+
"learning_rate": 9.060163154316791e-05,
|
| 1745 |
+
"loss": 0.117,
|
| 1746 |
+
"step": 28300
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 8.9,
|
| 1750 |
+
"learning_rate": 9.03467029231815e-05,
|
| 1751 |
+
"loss": 0.1142,
|
| 1752 |
+
"step": 28400
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 8.93,
|
| 1756 |
+
"learning_rate": 9.009177430319509e-05,
|
| 1757 |
+
"loss": 0.1168,
|
| 1758 |
+
"step": 28500
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 8.96,
|
| 1762 |
+
"learning_rate": 8.98368456832087e-05,
|
| 1763 |
+
"loss": 0.115,
|
| 1764 |
+
"step": 28600
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 8.99,
|
| 1768 |
+
"learning_rate": 8.958191706322229e-05,
|
| 1769 |
+
"loss": 0.1158,
|
| 1770 |
+
"step": 28700
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 9.02,
|
| 1774 |
+
"learning_rate": 8.932698844323588e-05,
|
| 1775 |
+
"loss": 0.1103,
|
| 1776 |
+
"step": 28800
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 9.06,
|
| 1780 |
+
"learning_rate": 8.907205982324948e-05,
|
| 1781 |
+
"loss": 0.1067,
|
| 1782 |
+
"step": 28900
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 9.09,
|
| 1786 |
+
"learning_rate": 8.881713120326308e-05,
|
| 1787 |
+
"loss": 0.1091,
|
| 1788 |
+
"step": 29000
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 9.12,
|
| 1792 |
+
"learning_rate": 8.856220258327668e-05,
|
| 1793 |
+
"loss": 0.108,
|
| 1794 |
+
"step": 29100
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 9.15,
|
| 1798 |
+
"learning_rate": 8.830727396329028e-05,
|
| 1799 |
+
"loss": 0.1082,
|
| 1800 |
+
"step": 29200
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 9.18,
|
| 1804 |
+
"learning_rate": 8.805234534330387e-05,
|
| 1805 |
+
"loss": 0.108,
|
| 1806 |
+
"step": 29300
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 9.21,
|
| 1810 |
+
"learning_rate": 8.779741672331746e-05,
|
| 1811 |
+
"loss": 0.1069,
|
| 1812 |
+
"step": 29400
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 9.24,
|
| 1816 |
+
"learning_rate": 8.754248810333107e-05,
|
| 1817 |
+
"loss": 0.1111,
|
| 1818 |
+
"step": 29500
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 9.27,
|
| 1822 |
+
"learning_rate": 8.728755948334466e-05,
|
| 1823 |
+
"loss": 0.11,
|
| 1824 |
+
"step": 29600
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 9.31,
|
| 1828 |
+
"learning_rate": 8.703263086335825e-05,
|
| 1829 |
+
"loss": 0.1085,
|
| 1830 |
+
"step": 29700
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 9.34,
|
| 1834 |
+
"learning_rate": 8.677770224337184e-05,
|
| 1835 |
+
"loss": 0.1094,
|
| 1836 |
+
"step": 29800
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 9.37,
|
| 1840 |
+
"learning_rate": 8.652277362338544e-05,
|
| 1841 |
+
"loss": 0.1111,
|
| 1842 |
+
"step": 29900
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 9.4,
|
| 1846 |
+
"learning_rate": 8.626784500339903e-05,
|
| 1847 |
+
"loss": 0.1065,
|
| 1848 |
+
"step": 30000
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 9.4,
|
| 1852 |
+
"eval_loss": 0.2719452977180481,
|
| 1853 |
+
"eval_runtime": 70.8391,
|
| 1854 |
+
"eval_samples_per_second": 126.216,
|
| 1855 |
+
"eval_steps_per_second": 1.976,
|
| 1856 |
+
"step": 30000
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 9.43,
|
| 1860 |
+
"learning_rate": 8.601291638341265e-05,
|
| 1861 |
+
"loss": 0.1072,
|
| 1862 |
+
"step": 30100
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 9.46,
|
| 1866 |
+
"learning_rate": 8.575798776342624e-05,
|
| 1867 |
+
"loss": 0.1075,
|
| 1868 |
+
"step": 30200
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 9.49,
|
| 1872 |
+
"learning_rate": 8.550305914343983e-05,
|
| 1873 |
+
"loss": 0.1076,
|
| 1874 |
+
"step": 30300
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 9.53,
|
| 1878 |
+
"learning_rate": 8.524813052345342e-05,
|
| 1879 |
+
"loss": 0.1056,
|
| 1880 |
+
"step": 30400
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 9.56,
|
| 1884 |
+
"learning_rate": 8.499320190346703e-05,
|
| 1885 |
+
"loss": 0.1079,
|
| 1886 |
+
"step": 30500
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 9.59,
|
| 1890 |
+
"learning_rate": 8.473827328348062e-05,
|
| 1891 |
+
"loss": 0.106,
|
| 1892 |
+
"step": 30600
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 9.62,
|
| 1896 |
+
"learning_rate": 8.448334466349421e-05,
|
| 1897 |
+
"loss": 0.1058,
|
| 1898 |
+
"step": 30700
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 9.65,
|
| 1902 |
+
"learning_rate": 8.422841604350781e-05,
|
| 1903 |
+
"loss": 0.1063,
|
| 1904 |
+
"step": 30800
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 9.68,
|
| 1908 |
+
"learning_rate": 8.39734874235214e-05,
|
| 1909 |
+
"loss": 0.1102,
|
| 1910 |
+
"step": 30900
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 9.71,
|
| 1914 |
+
"learning_rate": 8.3718558803535e-05,
|
| 1915 |
+
"loss": 0.1085,
|
| 1916 |
+
"step": 31000
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 9.74,
|
| 1920 |
+
"learning_rate": 8.34636301835486e-05,
|
| 1921 |
+
"loss": 0.1049,
|
| 1922 |
+
"step": 31100
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 9.78,
|
| 1926 |
+
"learning_rate": 8.32087015635622e-05,
|
| 1927 |
+
"loss": 0.1047,
|
| 1928 |
+
"step": 31200
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 9.81,
|
| 1932 |
+
"learning_rate": 8.295377294357579e-05,
|
| 1933 |
+
"loss": 0.1054,
|
| 1934 |
+
"step": 31300
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 9.84,
|
| 1938 |
+
"learning_rate": 8.26988443235894e-05,
|
| 1939 |
+
"loss": 0.1058,
|
| 1940 |
+
"step": 31400
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 9.87,
|
| 1944 |
+
"learning_rate": 8.244391570360299e-05,
|
| 1945 |
+
"loss": 0.1065,
|
| 1946 |
+
"step": 31500
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 9.9,
|
| 1950 |
+
"learning_rate": 8.218898708361658e-05,
|
| 1951 |
+
"loss": 0.1053,
|
| 1952 |
+
"step": 31600
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 9.93,
|
| 1956 |
+
"learning_rate": 8.193405846363017e-05,
|
| 1957 |
+
"loss": 0.1055,
|
| 1958 |
+
"step": 31700
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 9.96,
|
| 1962 |
+
"learning_rate": 8.167912984364377e-05,
|
| 1963 |
+
"loss": 0.1048,
|
| 1964 |
+
"step": 31800
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 10.0,
|
| 1968 |
+
"learning_rate": 8.142420122365736e-05,
|
| 1969 |
+
"loss": 0.1041,
|
| 1970 |
+
"step": 31900
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 10.03,
|
| 1974 |
+
"learning_rate": 8.116927260367095e-05,
|
| 1975 |
+
"loss": 0.0979,
|
| 1976 |
+
"step": 32000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 10.06,
|
| 1980 |
+
"learning_rate": 8.091434398368456e-05,
|
| 1981 |
+
"loss": 0.0987,
|
| 1982 |
+
"step": 32100
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 10.09,
|
| 1986 |
+
"learning_rate": 8.065941536369816e-05,
|
| 1987 |
+
"loss": 0.1002,
|
| 1988 |
+
"step": 32200
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 10.12,
|
| 1992 |
+
"learning_rate": 8.040448674371175e-05,
|
| 1993 |
+
"loss": 0.0962,
|
| 1994 |
+
"step": 32300
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 10.15,
|
| 1998 |
+
"learning_rate": 8.014955812372536e-05,
|
| 1999 |
+
"loss": 0.0982,
|
| 2000 |
+
"step": 32400
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 10.18,
|
| 2004 |
+
"learning_rate": 7.989462950373895e-05,
|
| 2005 |
+
"loss": 0.0999,
|
| 2006 |
+
"step": 32500
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 10.21,
|
| 2010 |
+
"learning_rate": 7.963970088375254e-05,
|
| 2011 |
+
"loss": 0.0971,
|
| 2012 |
+
"step": 32600
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 10.25,
|
| 2016 |
+
"learning_rate": 7.938477226376614e-05,
|
| 2017 |
+
"loss": 0.0986,
|
| 2018 |
+
"step": 32700
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 10.28,
|
| 2022 |
+
"learning_rate": 7.912984364377973e-05,
|
| 2023 |
+
"loss": 0.0977,
|
| 2024 |
+
"step": 32800
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 10.31,
|
| 2028 |
+
"learning_rate": 7.887491502379333e-05,
|
| 2029 |
+
"loss": 0.0965,
|
| 2030 |
+
"step": 32900
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 10.34,
|
| 2034 |
+
"learning_rate": 7.861998640380692e-05,
|
| 2035 |
+
"loss": 0.1012,
|
| 2036 |
+
"step": 33000
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 10.37,
|
| 2040 |
+
"learning_rate": 7.836505778382052e-05,
|
| 2041 |
+
"loss": 0.0973,
|
| 2042 |
+
"step": 33100
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 10.4,
|
| 2046 |
+
"learning_rate": 7.811012916383411e-05,
|
| 2047 |
+
"loss": 0.0969,
|
| 2048 |
+
"step": 33200
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 10.43,
|
| 2052 |
+
"learning_rate": 7.785520054384773e-05,
|
| 2053 |
+
"loss": 0.0983,
|
| 2054 |
+
"step": 33300
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 10.47,
|
| 2058 |
+
"learning_rate": 7.760027192386132e-05,
|
| 2059 |
+
"loss": 0.0986,
|
| 2060 |
+
"step": 33400
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 10.5,
|
| 2064 |
+
"learning_rate": 7.734534330387491e-05,
|
| 2065 |
+
"loss": 0.1001,
|
| 2066 |
+
"step": 33500
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 10.53,
|
| 2070 |
+
"learning_rate": 7.70904146838885e-05,
|
| 2071 |
+
"loss": 0.0981,
|
| 2072 |
+
"step": 33600
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 10.56,
|
| 2076 |
+
"learning_rate": 7.68354860639021e-05,
|
| 2077 |
+
"loss": 0.098,
|
| 2078 |
+
"step": 33700
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 10.59,
|
| 2082 |
+
"learning_rate": 7.65805574439157e-05,
|
| 2083 |
+
"loss": 0.0963,
|
| 2084 |
+
"step": 33800
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 10.62,
|
| 2088 |
+
"learning_rate": 7.632562882392929e-05,
|
| 2089 |
+
"loss": 0.0971,
|
| 2090 |
+
"step": 33900
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 10.65,
|
| 2094 |
+
"learning_rate": 7.607070020394289e-05,
|
| 2095 |
+
"loss": 0.0994,
|
| 2096 |
+
"step": 34000
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 10.68,
|
| 2100 |
+
"learning_rate": 7.581577158395648e-05,
|
| 2101 |
+
"loss": 0.096,
|
| 2102 |
+
"step": 34100
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 10.72,
|
| 2106 |
+
"learning_rate": 7.556084296397007e-05,
|
| 2107 |
+
"loss": 0.0992,
|
| 2108 |
+
"step": 34200
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 10.75,
|
| 2112 |
+
"learning_rate": 7.530591434398368e-05,
|
| 2113 |
+
"loss": 0.0968,
|
| 2114 |
+
"step": 34300
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 10.78,
|
| 2118 |
+
"learning_rate": 7.505098572399728e-05,
|
| 2119 |
+
"loss": 0.0974,
|
| 2120 |
+
"step": 34400
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 10.81,
|
| 2124 |
+
"learning_rate": 7.479605710401087e-05,
|
| 2125 |
+
"loss": 0.0961,
|
| 2126 |
+
"step": 34500
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 10.84,
|
| 2130 |
+
"learning_rate": 7.454112848402448e-05,
|
| 2131 |
+
"loss": 0.0973,
|
| 2132 |
+
"step": 34600
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 10.87,
|
| 2136 |
+
"learning_rate": 7.428619986403807e-05,
|
| 2137 |
+
"loss": 0.0968,
|
| 2138 |
+
"step": 34700
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 10.9,
|
| 2142 |
+
"learning_rate": 7.403127124405166e-05,
|
| 2143 |
+
"loss": 0.0974,
|
| 2144 |
+
"step": 34800
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 10.94,
|
| 2148 |
+
"learning_rate": 7.377634262406525e-05,
|
| 2149 |
+
"loss": 0.0959,
|
| 2150 |
+
"step": 34900
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 10.97,
|
| 2154 |
+
"learning_rate": 7.352141400407885e-05,
|
| 2155 |
+
"loss": 0.0995,
|
| 2156 |
+
"step": 35000
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 10.97,
|
| 2160 |
+
"eval_loss": 0.2515884339809418,
|
| 2161 |
+
"eval_runtime": 88.1892,
|
| 2162 |
+
"eval_samples_per_second": 101.384,
|
| 2163 |
+
"eval_steps_per_second": 1.587,
|
| 2164 |
+
"step": 35000
|
| 2165 |
+
},
|
| 2166 |
+
{
|
| 2167 |
+
"epoch": 11.0,
|
| 2168 |
+
"learning_rate": 7.326648538409244e-05,
|
| 2169 |
+
"loss": 0.0975,
|
| 2170 |
+
"step": 35100
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 11.03,
|
| 2174 |
+
"learning_rate": 7.301155676410605e-05,
|
| 2175 |
+
"loss": 0.0904,
|
| 2176 |
+
"step": 35200
|
| 2177 |
+
},
|
| 2178 |
+
{
|
| 2179 |
+
"epoch": 11.06,
|
| 2180 |
+
"learning_rate": 7.275662814411964e-05,
|
| 2181 |
+
"loss": 0.0888,
|
| 2182 |
+
"step": 35300
|
| 2183 |
+
},
|
| 2184 |
+
{
|
| 2185 |
+
"epoch": 11.09,
|
| 2186 |
+
"learning_rate": 7.250169952413324e-05,
|
| 2187 |
+
"loss": 0.0902,
|
| 2188 |
+
"step": 35400
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 11.12,
|
| 2192 |
+
"learning_rate": 7.224677090414683e-05,
|
| 2193 |
+
"loss": 0.0908,
|
| 2194 |
+
"step": 35500
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 11.15,
|
| 2198 |
+
"learning_rate": 7.199184228416042e-05,
|
| 2199 |
+
"loss": 0.0885,
|
| 2200 |
+
"step": 35600
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 11.19,
|
| 2204 |
+
"learning_rate": 7.173691366417403e-05,
|
| 2205 |
+
"loss": 0.0908,
|
| 2206 |
+
"step": 35700
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 11.22,
|
| 2210 |
+
"learning_rate": 7.148198504418762e-05,
|
| 2211 |
+
"loss": 0.0892,
|
| 2212 |
+
"step": 35800
|
| 2213 |
+
},
|
| 2214 |
+
{
|
| 2215 |
+
"epoch": 11.25,
|
| 2216 |
+
"learning_rate": 7.122705642420122e-05,
|
| 2217 |
+
"loss": 0.0902,
|
| 2218 |
+
"step": 35900
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 11.28,
|
| 2222 |
+
"learning_rate": 7.097212780421481e-05,
|
| 2223 |
+
"loss": 0.0883,
|
| 2224 |
+
"step": 36000
|
| 2225 |
+
},
|
| 2226 |
+
{
|
| 2227 |
+
"epoch": 11.31,
|
| 2228 |
+
"learning_rate": 7.07171991842284e-05,
|
| 2229 |
+
"loss": 0.0898,
|
| 2230 |
+
"step": 36100
|
| 2231 |
+
},
|
| 2232 |
+
{
|
| 2233 |
+
"epoch": 11.34,
|
| 2234 |
+
"learning_rate": 7.046227056424201e-05,
|
| 2235 |
+
"loss": 0.0887,
|
| 2236 |
+
"step": 36200
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 11.37,
|
| 2240 |
+
"learning_rate": 7.02073419442556e-05,
|
| 2241 |
+
"loss": 0.0895,
|
| 2242 |
+
"step": 36300
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 11.41,
|
| 2246 |
+
"learning_rate": 6.99524133242692e-05,
|
| 2247 |
+
"loss": 0.0905,
|
| 2248 |
+
"step": 36400
|
| 2249 |
+
},
|
| 2250 |
+
{
|
| 2251 |
+
"epoch": 11.44,
|
| 2252 |
+
"learning_rate": 6.96974847042828e-05,
|
| 2253 |
+
"loss": 0.0891,
|
| 2254 |
+
"step": 36500
|
| 2255 |
+
},
|
| 2256 |
+
{
|
| 2257 |
+
"epoch": 11.47,
|
| 2258 |
+
"learning_rate": 6.944255608429638e-05,
|
| 2259 |
+
"loss": 0.0896,
|
| 2260 |
+
"step": 36600
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 11.5,
|
| 2264 |
+
"learning_rate": 6.918762746430999e-05,
|
| 2265 |
+
"loss": 0.0903,
|
| 2266 |
+
"step": 36700
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"epoch": 11.53,
|
| 2270 |
+
"learning_rate": 6.893269884432358e-05,
|
| 2271 |
+
"loss": 0.0901,
|
| 2272 |
+
"step": 36800
|
| 2273 |
+
},
|
| 2274 |
+
{
|
| 2275 |
+
"epoch": 11.56,
|
| 2276 |
+
"learning_rate": 6.867777022433718e-05,
|
| 2277 |
+
"loss": 0.0899,
|
| 2278 |
+
"step": 36900
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 11.59,
|
| 2282 |
+
"learning_rate": 6.842284160435077e-05,
|
| 2283 |
+
"loss": 0.0881,
|
| 2284 |
+
"step": 37000
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 11.62,
|
| 2288 |
+
"learning_rate": 6.816791298436437e-05,
|
| 2289 |
+
"loss": 0.0921,
|
| 2290 |
+
"step": 37100
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"epoch": 11.66,
|
| 2294 |
+
"learning_rate": 6.791298436437797e-05,
|
| 2295 |
+
"loss": 0.0884,
|
| 2296 |
+
"step": 37200
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"epoch": 11.69,
|
| 2300 |
+
"learning_rate": 6.765805574439157e-05,
|
| 2301 |
+
"loss": 0.0897,
|
| 2302 |
+
"step": 37300
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 11.72,
|
| 2306 |
+
"learning_rate": 6.740312712440516e-05,
|
| 2307 |
+
"loss": 0.0877,
|
| 2308 |
+
"step": 37400
|
| 2309 |
+
},
|
| 2310 |
+
{
|
| 2311 |
+
"epoch": 11.75,
|
| 2312 |
+
"learning_rate": 6.714819850441876e-05,
|
| 2313 |
+
"loss": 0.0894,
|
| 2314 |
+
"step": 37500
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"epoch": 11.78,
|
| 2318 |
+
"learning_rate": 6.689326988443235e-05,
|
| 2319 |
+
"loss": 0.0881,
|
| 2320 |
+
"step": 37600
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 11.81,
|
| 2324 |
+
"learning_rate": 6.663834126444595e-05,
|
| 2325 |
+
"loss": 0.0891,
|
| 2326 |
+
"step": 37700
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 11.84,
|
| 2330 |
+
"learning_rate": 6.638341264445955e-05,
|
| 2331 |
+
"loss": 0.0873,
|
| 2332 |
+
"step": 37800
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 11.88,
|
| 2336 |
+
"learning_rate": 6.612848402447315e-05,
|
| 2337 |
+
"loss": 0.0873,
|
| 2338 |
+
"step": 37900
|
| 2339 |
+
},
|
| 2340 |
+
{
|
| 2341 |
+
"epoch": 11.91,
|
| 2342 |
+
"learning_rate": 6.587355540448674e-05,
|
| 2343 |
+
"loss": 0.0887,
|
| 2344 |
+
"step": 38000
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 11.94,
|
| 2348 |
+
"learning_rate": 6.561862678450033e-05,
|
| 2349 |
+
"loss": 0.0867,
|
| 2350 |
+
"step": 38100
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 11.97,
|
| 2354 |
+
"learning_rate": 6.536369816451393e-05,
|
| 2355 |
+
"loss": 0.0906,
|
| 2356 |
+
"step": 38200
|
| 2357 |
+
},
|
| 2358 |
+
{
|
| 2359 |
+
"epoch": 12.0,
|
| 2360 |
+
"learning_rate": 6.510876954452752e-05,
|
| 2361 |
+
"loss": 0.0871,
|
| 2362 |
+
"step": 38300
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 12.03,
|
| 2366 |
+
"learning_rate": 6.485384092454113e-05,
|
| 2367 |
+
"loss": 0.0816,
|
| 2368 |
+
"step": 38400
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 12.06,
|
| 2372 |
+
"learning_rate": 6.459891230455472e-05,
|
| 2373 |
+
"loss": 0.0815,
|
| 2374 |
+
"step": 38500
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 12.09,
|
| 2378 |
+
"learning_rate": 6.434398368456832e-05,
|
| 2379 |
+
"loss": 0.0817,
|
| 2380 |
+
"step": 38600
|
| 2381 |
+
},
|
| 2382 |
+
{
|
| 2383 |
+
"epoch": 12.13,
|
| 2384 |
+
"learning_rate": 6.408905506458191e-05,
|
| 2385 |
+
"loss": 0.0835,
|
| 2386 |
+
"step": 38700
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 12.16,
|
| 2390 |
+
"learning_rate": 6.38341264445955e-05,
|
| 2391 |
+
"loss": 0.0833,
|
| 2392 |
+
"step": 38800
|
| 2393 |
+
},
|
| 2394 |
+
{
|
| 2395 |
+
"epoch": 12.19,
|
| 2396 |
+
"learning_rate": 6.35791978246091e-05,
|
| 2397 |
+
"loss": 0.081,
|
| 2398 |
+
"step": 38900
|
| 2399 |
+
},
|
| 2400 |
+
{
|
| 2401 |
+
"epoch": 12.22,
|
| 2402 |
+
"learning_rate": 6.33242692046227e-05,
|
| 2403 |
+
"loss": 0.0838,
|
| 2404 |
+
"step": 39000
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 12.25,
|
| 2408 |
+
"learning_rate": 6.30693405846363e-05,
|
| 2409 |
+
"loss": 0.0829,
|
| 2410 |
+
"step": 39100
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 12.28,
|
| 2414 |
+
"learning_rate": 6.281441196464989e-05,
|
| 2415 |
+
"loss": 0.082,
|
| 2416 |
+
"step": 39200
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 12.31,
|
| 2420 |
+
"learning_rate": 6.255948334466348e-05,
|
| 2421 |
+
"loss": 0.0826,
|
| 2422 |
+
"step": 39300
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 12.35,
|
| 2426 |
+
"learning_rate": 6.230455472467709e-05,
|
| 2427 |
+
"loss": 0.0799,
|
| 2428 |
+
"step": 39400
|
| 2429 |
+
},
|
| 2430 |
+
{
|
| 2431 |
+
"epoch": 12.38,
|
| 2432 |
+
"learning_rate": 6.204962610469068e-05,
|
| 2433 |
+
"loss": 0.0788,
|
| 2434 |
+
"step": 39500
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"epoch": 12.41,
|
| 2438 |
+
"learning_rate": 6.179469748470428e-05,
|
| 2439 |
+
"loss": 0.082,
|
| 2440 |
+
"step": 39600
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"epoch": 12.44,
|
| 2444 |
+
"learning_rate": 6.153976886471787e-05,
|
| 2445 |
+
"loss": 0.0823,
|
| 2446 |
+
"step": 39700
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 12.47,
|
| 2450 |
+
"learning_rate": 6.128484024473146e-05,
|
| 2451 |
+
"loss": 0.08,
|
| 2452 |
+
"step": 39800
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 12.5,
|
| 2456 |
+
"learning_rate": 6.102991162474506e-05,
|
| 2457 |
+
"loss": 0.0802,
|
| 2458 |
+
"step": 39900
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 12.53,
|
| 2462 |
+
"learning_rate": 6.0774983004758665e-05,
|
| 2463 |
+
"loss": 0.0812,
|
| 2464 |
+
"step": 40000
|
| 2465 |
+
},
|
| 2466 |
+
{
|
| 2467 |
+
"epoch": 12.53,
|
| 2468 |
+
"eval_loss": 0.23678846657276154,
|
| 2469 |
+
"eval_runtime": 82.6634,
|
| 2470 |
+
"eval_samples_per_second": 108.162,
|
| 2471 |
+
"eval_steps_per_second": 1.694,
|
| 2472 |
+
"step": 40000
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 12.56,
|
| 2476 |
+
"learning_rate": 6.052005438477226e-05,
|
| 2477 |
+
"loss": 0.0826,
|
| 2478 |
+
"step": 40100
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 12.6,
|
| 2482 |
+
"learning_rate": 6.026512576478585e-05,
|
| 2483 |
+
"loss": 0.0814,
|
| 2484 |
+
"step": 40200
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 12.63,
|
| 2488 |
+
"learning_rate": 6.001019714479945e-05,
|
| 2489 |
+
"loss": 0.0797,
|
| 2490 |
+
"step": 40300
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 12.66,
|
| 2494 |
+
"learning_rate": 5.975526852481304e-05,
|
| 2495 |
+
"loss": 0.0805,
|
| 2496 |
+
"step": 40400
|
| 2497 |
+
},
|
| 2498 |
+
{
|
| 2499 |
+
"epoch": 12.69,
|
| 2500 |
+
"learning_rate": 5.9500339904826646e-05,
|
| 2501 |
+
"loss": 0.0796,
|
| 2502 |
+
"step": 40500
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 12.72,
|
| 2506 |
+
"learning_rate": 5.924541128484024e-05,
|
| 2507 |
+
"loss": 0.0789,
|
| 2508 |
+
"step": 40600
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 12.75,
|
| 2512 |
+
"learning_rate": 5.8990482664853834e-05,
|
| 2513 |
+
"loss": 0.0808,
|
| 2514 |
+
"step": 40700
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 12.78,
|
| 2518 |
+
"learning_rate": 5.873555404486743e-05,
|
| 2519 |
+
"loss": 0.0819,
|
| 2520 |
+
"step": 40800
|
| 2521 |
+
},
|
| 2522 |
+
{
|
| 2523 |
+
"epoch": 12.82,
|
| 2524 |
+
"learning_rate": 5.848062542488102e-05,
|
| 2525 |
+
"loss": 0.0814,
|
| 2526 |
+
"step": 40900
|
| 2527 |
+
},
|
| 2528 |
+
{
|
| 2529 |
+
"epoch": 12.85,
|
| 2530 |
+
"learning_rate": 5.8225696804894626e-05,
|
| 2531 |
+
"loss": 0.0788,
|
| 2532 |
+
"step": 41000
|
| 2533 |
+
},
|
| 2534 |
+
{
|
| 2535 |
+
"epoch": 12.88,
|
| 2536 |
+
"learning_rate": 5.7970768184908224e-05,
|
| 2537 |
+
"loss": 0.0799,
|
| 2538 |
+
"step": 41100
|
| 2539 |
+
},
|
| 2540 |
+
{
|
| 2541 |
+
"epoch": 12.91,
|
| 2542 |
+
"learning_rate": 5.7715839564921815e-05,
|
| 2543 |
+
"loss": 0.0798,
|
| 2544 |
+
"step": 41200
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 12.94,
|
| 2548 |
+
"learning_rate": 5.746091094493541e-05,
|
| 2549 |
+
"loss": 0.081,
|
| 2550 |
+
"step": 41300
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 12.97,
|
| 2554 |
+
"learning_rate": 5.720598232494901e-05,
|
| 2555 |
+
"loss": 0.079,
|
| 2556 |
+
"step": 41400
|
| 2557 |
+
},
|
| 2558 |
+
{
|
| 2559 |
+
"epoch": 13.0,
|
| 2560 |
+
"learning_rate": 5.69510537049626e-05,
|
| 2561 |
+
"loss": 0.0804,
|
| 2562 |
+
"step": 41500
|
| 2563 |
+
},
|
| 2564 |
+
{
|
| 2565 |
+
"epoch": 13.03,
|
| 2566 |
+
"learning_rate": 5.6696125084976205e-05,
|
| 2567 |
+
"loss": 0.0745,
|
| 2568 |
+
"step": 41600
|
| 2569 |
+
},
|
| 2570 |
+
{
|
| 2571 |
+
"epoch": 13.07,
|
| 2572 |
+
"learning_rate": 5.64411964649898e-05,
|
| 2573 |
+
"loss": 0.0754,
|
| 2574 |
+
"step": 41700
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 13.1,
|
| 2578 |
+
"learning_rate": 5.618626784500339e-05,
|
| 2579 |
+
"loss": 0.0735,
|
| 2580 |
+
"step": 41800
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 13.13,
|
| 2584 |
+
"learning_rate": 5.593133922501699e-05,
|
| 2585 |
+
"loss": 0.0739,
|
| 2586 |
+
"step": 41900
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 13.16,
|
| 2590 |
+
"learning_rate": 5.567641060503058e-05,
|
| 2591 |
+
"loss": 0.0734,
|
| 2592 |
+
"step": 42000
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 13.19,
|
| 2596 |
+
"learning_rate": 5.5421481985044185e-05,
|
| 2597 |
+
"loss": 0.0726,
|
| 2598 |
+
"step": 42100
|
| 2599 |
+
},
|
| 2600 |
+
{
|
| 2601 |
+
"epoch": 13.22,
|
| 2602 |
+
"learning_rate": 5.516655336505778e-05,
|
| 2603 |
+
"loss": 0.0753,
|
| 2604 |
+
"step": 42200
|
| 2605 |
+
},
|
| 2606 |
+
{
|
| 2607 |
+
"epoch": 13.25,
|
| 2608 |
+
"learning_rate": 5.491162474507137e-05,
|
| 2609 |
+
"loss": 0.0743,
|
| 2610 |
+
"step": 42300
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"epoch": 13.29,
|
| 2614 |
+
"learning_rate": 5.465669612508497e-05,
|
| 2615 |
+
"loss": 0.0755,
|
| 2616 |
+
"step": 42400
|
| 2617 |
+
},
|
| 2618 |
+
{
|
| 2619 |
+
"epoch": 13.32,
|
| 2620 |
+
"learning_rate": 5.440176750509856e-05,
|
| 2621 |
+
"loss": 0.0726,
|
| 2622 |
+
"step": 42500
|
| 2623 |
+
},
|
| 2624 |
+
{
|
| 2625 |
+
"epoch": 13.35,
|
| 2626 |
+
"learning_rate": 5.4146838885112166e-05,
|
| 2627 |
+
"loss": 0.077,
|
| 2628 |
+
"step": 42600
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 13.38,
|
| 2632 |
+
"learning_rate": 5.389191026512576e-05,
|
| 2633 |
+
"loss": 0.0753,
|
| 2634 |
+
"step": 42700
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 13.41,
|
| 2638 |
+
"learning_rate": 5.3636981645139354e-05,
|
| 2639 |
+
"loss": 0.0729,
|
| 2640 |
+
"step": 42800
|
| 2641 |
+
},
|
| 2642 |
+
{
|
| 2643 |
+
"epoch": 13.44,
|
| 2644 |
+
"learning_rate": 5.338205302515295e-05,
|
| 2645 |
+
"loss": 0.0731,
|
| 2646 |
+
"step": 42900
|
| 2647 |
+
},
|
| 2648 |
+
{
|
| 2649 |
+
"epoch": 13.47,
|
| 2650 |
+
"learning_rate": 5.312712440516655e-05,
|
| 2651 |
+
"loss": 0.0723,
|
| 2652 |
+
"step": 43000
|
| 2653 |
+
},
|
| 2654 |
+
{
|
| 2655 |
+
"epoch": 13.5,
|
| 2656 |
+
"learning_rate": 5.287219578518014e-05,
|
| 2657 |
+
"loss": 0.0741,
|
| 2658 |
+
"step": 43100
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 13.54,
|
| 2662 |
+
"learning_rate": 5.2617267165193744e-05,
|
| 2663 |
+
"loss": 0.0721,
|
| 2664 |
+
"step": 43200
|
| 2665 |
+
},
|
| 2666 |
+
{
|
| 2667 |
+
"epoch": 13.57,
|
| 2668 |
+
"learning_rate": 5.236233854520734e-05,
|
| 2669 |
+
"loss": 0.0726,
|
| 2670 |
+
"step": 43300
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 13.6,
|
| 2674 |
+
"learning_rate": 5.210740992522093e-05,
|
| 2675 |
+
"loss": 0.0754,
|
| 2676 |
+
"step": 43400
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 13.63,
|
| 2680 |
+
"learning_rate": 5.185248130523453e-05,
|
| 2681 |
+
"loss": 0.0741,
|
| 2682 |
+
"step": 43500
|
| 2683 |
+
},
|
| 2684 |
+
{
|
| 2685 |
+
"epoch": 13.66,
|
| 2686 |
+
"learning_rate": 5.159755268524812e-05,
|
| 2687 |
+
"loss": 0.0736,
|
| 2688 |
+
"step": 43600
|
| 2689 |
+
},
|
| 2690 |
+
{
|
| 2691 |
+
"epoch": 13.69,
|
| 2692 |
+
"learning_rate": 5.1342624065261725e-05,
|
| 2693 |
+
"loss": 0.0724,
|
| 2694 |
+
"step": 43700
|
| 2695 |
+
},
|
| 2696 |
+
{
|
| 2697 |
+
"epoch": 13.72,
|
| 2698 |
+
"learning_rate": 5.108769544527532e-05,
|
| 2699 |
+
"loss": 0.0739,
|
| 2700 |
+
"step": 43800
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 13.76,
|
| 2704 |
+
"learning_rate": 5.083276682528891e-05,
|
| 2705 |
+
"loss": 0.072,
|
| 2706 |
+
"step": 43900
|
| 2707 |
+
},
|
| 2708 |
+
{
|
| 2709 |
+
"epoch": 13.79,
|
| 2710 |
+
"learning_rate": 5.057783820530251e-05,
|
| 2711 |
+
"loss": 0.0734,
|
| 2712 |
+
"step": 44000
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 13.82,
|
| 2716 |
+
"learning_rate": 5.03229095853161e-05,
|
| 2717 |
+
"loss": 0.0734,
|
| 2718 |
+
"step": 44100
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 13.85,
|
| 2722 |
+
"learning_rate": 5.0067980965329705e-05,
|
| 2723 |
+
"loss": 0.0718,
|
| 2724 |
+
"step": 44200
|
| 2725 |
+
},
|
| 2726 |
+
{
|
| 2727 |
+
"epoch": 13.88,
|
| 2728 |
+
"learning_rate": 4.98130523453433e-05,
|
| 2729 |
+
"loss": 0.0732,
|
| 2730 |
+
"step": 44300
|
| 2731 |
+
},
|
| 2732 |
+
{
|
| 2733 |
+
"epoch": 13.91,
|
| 2734 |
+
"learning_rate": 4.9558123725356893e-05,
|
| 2735 |
+
"loss": 0.0759,
|
| 2736 |
+
"step": 44400
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 13.94,
|
| 2740 |
+
"learning_rate": 4.930319510537049e-05,
|
| 2741 |
+
"loss": 0.0715,
|
| 2742 |
+
"step": 44500
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 13.97,
|
| 2746 |
+
"learning_rate": 4.904826648538409e-05,
|
| 2747 |
+
"loss": 0.0729,
|
| 2748 |
+
"step": 44600
|
| 2749 |
+
},
|
| 2750 |
+
{
|
| 2751 |
+
"epoch": 14.01,
|
| 2752 |
+
"learning_rate": 4.879333786539768e-05,
|
| 2753 |
+
"loss": 0.0712,
|
| 2754 |
+
"step": 44700
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 14.04,
|
| 2758 |
+
"learning_rate": 4.853840924541128e-05,
|
| 2759 |
+
"loss": 0.0668,
|
| 2760 |
+
"step": 44800
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 14.07,
|
| 2764 |
+
"learning_rate": 4.828348062542488e-05,
|
| 2765 |
+
"loss": 0.0677,
|
| 2766 |
+
"step": 44900
|
| 2767 |
+
},
|
| 2768 |
+
{
|
| 2769 |
+
"epoch": 14.1,
|
| 2770 |
+
"learning_rate": 4.802855200543847e-05,
|
| 2771 |
+
"loss": 0.066,
|
| 2772 |
+
"step": 45000
|
| 2773 |
+
},
|
| 2774 |
+
{
|
| 2775 |
+
"epoch": 14.1,
|
| 2776 |
+
"eval_loss": 0.22296179831027985,
|
| 2777 |
+
"eval_runtime": 91.9595,
|
| 2778 |
+
"eval_samples_per_second": 97.228,
|
| 2779 |
+
"eval_steps_per_second": 1.522,
|
| 2780 |
+
"step": 45000
|
| 2781 |
+
},
|
| 2782 |
+
{
|
| 2783 |
+
"epoch": 14.13,
|
| 2784 |
+
"learning_rate": 4.777362338545207e-05,
|
| 2785 |
+
"loss": 0.0669,
|
| 2786 |
+
"step": 45100
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 14.16,
|
| 2790 |
+
"learning_rate": 4.751869476546566e-05,
|
| 2791 |
+
"loss": 0.0668,
|
| 2792 |
+
"step": 45200
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 14.19,
|
| 2796 |
+
"learning_rate": 4.7263766145479264e-05,
|
| 2797 |
+
"loss": 0.0664,
|
| 2798 |
+
"step": 45300
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 14.23,
|
| 2802 |
+
"learning_rate": 4.700883752549286e-05,
|
| 2803 |
+
"loss": 0.0661,
|
| 2804 |
+
"step": 45400
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 14.26,
|
| 2808 |
+
"learning_rate": 4.675390890550645e-05,
|
| 2809 |
+
"loss": 0.0672,
|
| 2810 |
+
"step": 45500
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 14.29,
|
| 2814 |
+
"learning_rate": 4.649898028552005e-05,
|
| 2815 |
+
"loss": 0.0665,
|
| 2816 |
+
"step": 45600
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 14.32,
|
| 2820 |
+
"learning_rate": 4.624405166553364e-05,
|
| 2821 |
+
"loss": 0.0677,
|
| 2822 |
+
"step": 45700
|
| 2823 |
+
},
|
| 2824 |
+
{
|
| 2825 |
+
"epoch": 14.35,
|
| 2826 |
+
"learning_rate": 4.5989123045547245e-05,
|
| 2827 |
+
"loss": 0.0676,
|
| 2828 |
+
"step": 45800
|
| 2829 |
+
},
|
| 2830 |
+
{
|
| 2831 |
+
"epoch": 14.38,
|
| 2832 |
+
"learning_rate": 4.573419442556084e-05,
|
| 2833 |
+
"loss": 0.0665,
|
| 2834 |
+
"step": 45900
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 14.41,
|
| 2838 |
+
"learning_rate": 4.547926580557443e-05,
|
| 2839 |
+
"loss": 0.0681,
|
| 2840 |
+
"step": 46000
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 14.44,
|
| 2844 |
+
"learning_rate": 4.522433718558803e-05,
|
| 2845 |
+
"loss": 0.0673,
|
| 2846 |
+
"step": 46100
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 14.48,
|
| 2850 |
+
"learning_rate": 4.496940856560163e-05,
|
| 2851 |
+
"loss": 0.0669,
|
| 2852 |
+
"step": 46200
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 14.51,
|
| 2856 |
+
"learning_rate": 4.471447994561522e-05,
|
| 2857 |
+
"loss": 0.067,
|
| 2858 |
+
"step": 46300
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 14.54,
|
| 2862 |
+
"learning_rate": 4.445955132562882e-05,
|
| 2863 |
+
"loss": 0.0665,
|
| 2864 |
+
"step": 46400
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 14.57,
|
| 2868 |
+
"learning_rate": 4.420462270564242e-05,
|
| 2869 |
+
"loss": 0.0675,
|
| 2870 |
+
"step": 46500
|
| 2871 |
+
},
|
| 2872 |
+
{
|
| 2873 |
+
"epoch": 14.6,
|
| 2874 |
+
"learning_rate": 4.394969408565601e-05,
|
| 2875 |
+
"loss": 0.0658,
|
| 2876 |
+
"step": 46600
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 14.63,
|
| 2880 |
+
"learning_rate": 4.369476546566961e-05,
|
| 2881 |
+
"loss": 0.064,
|
| 2882 |
+
"step": 46700
|
| 2883 |
+
},
|
| 2884 |
+
{
|
| 2885 |
+
"epoch": 14.66,
|
| 2886 |
+
"learning_rate": 4.34398368456832e-05,
|
| 2887 |
+
"loss": 0.0679,
|
| 2888 |
+
"step": 46800
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 14.7,
|
| 2892 |
+
"learning_rate": 4.31849082256968e-05,
|
| 2893 |
+
"loss": 0.0652,
|
| 2894 |
+
"step": 46900
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 14.73,
|
| 2898 |
+
"learning_rate": 4.29299796057104e-05,
|
| 2899 |
+
"loss": 0.0654,
|
| 2900 |
+
"step": 47000
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 14.76,
|
| 2904 |
+
"learning_rate": 4.267505098572399e-05,
|
| 2905 |
+
"loss": 0.0672,
|
| 2906 |
+
"step": 47100
|
| 2907 |
+
},
|
| 2908 |
+
{
|
| 2909 |
+
"epoch": 14.79,
|
| 2910 |
+
"learning_rate": 4.242012236573759e-05,
|
| 2911 |
+
"loss": 0.0656,
|
| 2912 |
+
"step": 47200
|
| 2913 |
+
},
|
| 2914 |
+
{
|
| 2915 |
+
"epoch": 14.82,
|
| 2916 |
+
"learning_rate": 4.216519374575118e-05,
|
| 2917 |
+
"loss": 0.066,
|
| 2918 |
+
"step": 47300
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 14.85,
|
| 2922 |
+
"learning_rate": 4.1910265125764784e-05,
|
| 2923 |
+
"loss": 0.065,
|
| 2924 |
+
"step": 47400
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 14.88,
|
| 2928 |
+
"learning_rate": 4.165533650577838e-05,
|
| 2929 |
+
"loss": 0.0649,
|
| 2930 |
+
"step": 47500
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 14.91,
|
| 2934 |
+
"learning_rate": 4.140040788579197e-05,
|
| 2935 |
+
"loss": 0.0665,
|
| 2936 |
+
"step": 47600
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 14.95,
|
| 2940 |
+
"learning_rate": 4.114547926580557e-05,
|
| 2941 |
+
"loss": 0.0631,
|
| 2942 |
+
"step": 47700
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 14.98,
|
| 2946 |
+
"learning_rate": 4.089055064581917e-05,
|
| 2947 |
+
"loss": 0.0658,
|
| 2948 |
+
"step": 47800
|
| 2949 |
+
},
|
| 2950 |
+
{
|
| 2951 |
+
"epoch": 15.01,
|
| 2952 |
+
"learning_rate": 4.063562202583276e-05,
|
| 2953 |
+
"loss": 0.0646,
|
| 2954 |
+
"step": 47900
|
| 2955 |
+
},
|
| 2956 |
+
{
|
| 2957 |
+
"epoch": 15.04,
|
| 2958 |
+
"learning_rate": 4.038069340584636e-05,
|
| 2959 |
+
"loss": 0.0594,
|
| 2960 |
+
"step": 48000
|
| 2961 |
+
},
|
| 2962 |
+
{
|
| 2963 |
+
"epoch": 15.07,
|
| 2964 |
+
"learning_rate": 4.012576478585996e-05,
|
| 2965 |
+
"loss": 0.0594,
|
| 2966 |
+
"step": 48100
|
| 2967 |
+
},
|
| 2968 |
+
{
|
| 2969 |
+
"epoch": 15.1,
|
| 2970 |
+
"learning_rate": 3.987083616587355e-05,
|
| 2971 |
+
"loss": 0.0598,
|
| 2972 |
+
"step": 48200
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 15.13,
|
| 2976 |
+
"learning_rate": 3.961590754588715e-05,
|
| 2977 |
+
"loss": 0.0603,
|
| 2978 |
+
"step": 48300
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 15.17,
|
| 2982 |
+
"learning_rate": 3.936097892590074e-05,
|
| 2983 |
+
"loss": 0.0616,
|
| 2984 |
+
"step": 48400
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 15.2,
|
| 2988 |
+
"learning_rate": 3.910605030591434e-05,
|
| 2989 |
+
"loss": 0.0595,
|
| 2990 |
+
"step": 48500
|
| 2991 |
+
},
|
| 2992 |
+
{
|
| 2993 |
+
"epoch": 15.23,
|
| 2994 |
+
"learning_rate": 3.885112168592794e-05,
|
| 2995 |
+
"loss": 0.0598,
|
| 2996 |
+
"step": 48600
|
| 2997 |
+
},
|
| 2998 |
+
{
|
| 2999 |
+
"epoch": 15.26,
|
| 3000 |
+
"learning_rate": 3.859619306594153e-05,
|
| 3001 |
+
"loss": 0.0614,
|
| 3002 |
+
"step": 48700
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 15.29,
|
| 3006 |
+
"learning_rate": 3.834126444595513e-05,
|
| 3007 |
+
"loss": 0.06,
|
| 3008 |
+
"step": 48800
|
| 3009 |
+
},
|
| 3010 |
+
{
|
| 3011 |
+
"epoch": 15.32,
|
| 3012 |
+
"learning_rate": 3.808633582596872e-05,
|
| 3013 |
+
"loss": 0.0611,
|
| 3014 |
+
"step": 48900
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 15.35,
|
| 3018 |
+
"learning_rate": 3.783140720598232e-05,
|
| 3019 |
+
"loss": 0.0598,
|
| 3020 |
+
"step": 49000
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 15.38,
|
| 3024 |
+
"learning_rate": 3.757647858599592e-05,
|
| 3025 |
+
"loss": 0.059,
|
| 3026 |
+
"step": 49100
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 15.42,
|
| 3030 |
+
"learning_rate": 3.732154996600951e-05,
|
| 3031 |
+
"loss": 0.0596,
|
| 3032 |
+
"step": 49200
|
| 3033 |
+
},
|
| 3034 |
+
{
|
| 3035 |
+
"epoch": 15.45,
|
| 3036 |
+
"learning_rate": 3.706662134602311e-05,
|
| 3037 |
+
"loss": 0.0589,
|
| 3038 |
+
"step": 49300
|
| 3039 |
+
},
|
| 3040 |
+
{
|
| 3041 |
+
"epoch": 15.48,
|
| 3042 |
+
"learning_rate": 3.6811692726036707e-05,
|
| 3043 |
+
"loss": 0.0606,
|
| 3044 |
+
"step": 49400
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 15.51,
|
| 3048 |
+
"learning_rate": 3.6556764106050304e-05,
|
| 3049 |
+
"loss": 0.0598,
|
| 3050 |
+
"step": 49500
|
| 3051 |
+
},
|
| 3052 |
+
{
|
| 3053 |
+
"epoch": 15.54,
|
| 3054 |
+
"learning_rate": 3.6301835486063895e-05,
|
| 3055 |
+
"loss": 0.0606,
|
| 3056 |
+
"step": 49600
|
| 3057 |
+
},
|
| 3058 |
+
{
|
| 3059 |
+
"epoch": 15.57,
|
| 3060 |
+
"learning_rate": 3.60469068660775e-05,
|
| 3061 |
+
"loss": 0.0588,
|
| 3062 |
+
"step": 49700
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 15.6,
|
| 3066 |
+
"learning_rate": 3.579197824609109e-05,
|
| 3067 |
+
"loss": 0.0575,
|
| 3068 |
+
"step": 49800
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 15.64,
|
| 3072 |
+
"learning_rate": 3.553704962610469e-05,
|
| 3073 |
+
"loss": 0.0596,
|
| 3074 |
+
"step": 49900
|
| 3075 |
+
},
|
| 3076 |
+
{
|
| 3077 |
+
"epoch": 15.67,
|
| 3078 |
+
"learning_rate": 3.5282121006118285e-05,
|
| 3079 |
+
"loss": 0.0574,
|
| 3080 |
+
"step": 50000
|
| 3081 |
+
},
|
| 3082 |
+
{
|
| 3083 |
+
"epoch": 15.67,
|
| 3084 |
+
"eval_loss": 0.2118692249059677,
|
| 3085 |
+
"eval_runtime": 95.2746,
|
| 3086 |
+
"eval_samples_per_second": 93.845,
|
| 3087 |
+
"eval_steps_per_second": 1.469,
|
| 3088 |
+
"step": 50000
|
| 3089 |
+
},
|
| 3090 |
+
{
|
| 3091 |
+
"epoch": 15.7,
|
| 3092 |
+
"learning_rate": 3.502719238613188e-05,
|
| 3093 |
+
"loss": 0.0592,
|
| 3094 |
+
"step": 50100
|
| 3095 |
+
},
|
| 3096 |
+
{
|
| 3097 |
+
"epoch": 15.73,
|
| 3098 |
+
"learning_rate": 3.477226376614548e-05,
|
| 3099 |
+
"loss": 0.0585,
|
| 3100 |
+
"step": 50200
|
| 3101 |
+
},
|
| 3102 |
+
{
|
| 3103 |
+
"epoch": 15.76,
|
| 3104 |
+
"learning_rate": 3.451733514615907e-05,
|
| 3105 |
+
"loss": 0.0572,
|
| 3106 |
+
"step": 50300
|
| 3107 |
+
},
|
| 3108 |
+
{
|
| 3109 |
+
"epoch": 15.79,
|
| 3110 |
+
"learning_rate": 3.426240652617267e-05,
|
| 3111 |
+
"loss": 0.0603,
|
| 3112 |
+
"step": 50400
|
| 3113 |
+
},
|
| 3114 |
+
{
|
| 3115 |
+
"epoch": 15.82,
|
| 3116 |
+
"learning_rate": 3.4007477906186265e-05,
|
| 3117 |
+
"loss": 0.0589,
|
| 3118 |
+
"step": 50500
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 15.85,
|
| 3122 |
+
"learning_rate": 3.375254928619986e-05,
|
| 3123 |
+
"loss": 0.0573,
|
| 3124 |
+
"step": 50600
|
| 3125 |
+
},
|
| 3126 |
+
{
|
| 3127 |
+
"epoch": 15.89,
|
| 3128 |
+
"learning_rate": 3.3497620666213453e-05,
|
| 3129 |
+
"loss": 0.0582,
|
| 3130 |
+
"step": 50700
|
| 3131 |
+
},
|
| 3132 |
+
{
|
| 3133 |
+
"epoch": 15.92,
|
| 3134 |
+
"learning_rate": 3.324269204622705e-05,
|
| 3135 |
+
"loss": 0.0582,
|
| 3136 |
+
"step": 50800
|
| 3137 |
+
},
|
| 3138 |
+
{
|
| 3139 |
+
"epoch": 15.95,
|
| 3140 |
+
"learning_rate": 3.298776342624065e-05,
|
| 3141 |
+
"loss": 0.0598,
|
| 3142 |
+
"step": 50900
|
| 3143 |
+
},
|
| 3144 |
+
{
|
| 3145 |
+
"epoch": 15.98,
|
| 3146 |
+
"learning_rate": 3.2732834806254246e-05,
|
| 3147 |
+
"loss": 0.059,
|
| 3148 |
+
"step": 51000
|
| 3149 |
+
},
|
| 3150 |
+
{
|
| 3151 |
+
"epoch": 16.01,
|
| 3152 |
+
"learning_rate": 3.2477906186267843e-05,
|
| 3153 |
+
"loss": 0.0567,
|
| 3154 |
+
"step": 51100
|
| 3155 |
+
},
|
| 3156 |
+
{
|
| 3157 |
+
"epoch": 16.04,
|
| 3158 |
+
"learning_rate": 3.2222977566281434e-05,
|
| 3159 |
+
"loss": 0.0555,
|
| 3160 |
+
"step": 51200
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 16.07,
|
| 3164 |
+
"learning_rate": 3.196804894629504e-05,
|
| 3165 |
+
"loss": 0.0545,
|
| 3166 |
+
"step": 51300
|
| 3167 |
+
},
|
| 3168 |
+
{
|
| 3169 |
+
"epoch": 16.11,
|
| 3170 |
+
"learning_rate": 3.171312032630863e-05,
|
| 3171 |
+
"loss": 0.0536,
|
| 3172 |
+
"step": 51400
|
| 3173 |
+
},
|
| 3174 |
+
{
|
| 3175 |
+
"epoch": 16.14,
|
| 3176 |
+
"learning_rate": 3.1458191706322227e-05,
|
| 3177 |
+
"loss": 0.0542,
|
| 3178 |
+
"step": 51500
|
| 3179 |
+
},
|
| 3180 |
+
{
|
| 3181 |
+
"epoch": 16.17,
|
| 3182 |
+
"learning_rate": 3.1203263086335824e-05,
|
| 3183 |
+
"loss": 0.0536,
|
| 3184 |
+
"step": 51600
|
| 3185 |
+
},
|
| 3186 |
+
{
|
| 3187 |
+
"epoch": 16.2,
|
| 3188 |
+
"learning_rate": 3.094833446634942e-05,
|
| 3189 |
+
"loss": 0.0529,
|
| 3190 |
+
"step": 51700
|
| 3191 |
+
},
|
| 3192 |
+
{
|
| 3193 |
+
"epoch": 16.23,
|
| 3194 |
+
"learning_rate": 3.069340584636302e-05,
|
| 3195 |
+
"loss": 0.0529,
|
| 3196 |
+
"step": 51800
|
| 3197 |
+
},
|
| 3198 |
+
{
|
| 3199 |
+
"epoch": 16.26,
|
| 3200 |
+
"learning_rate": 3.0438477226376613e-05,
|
| 3201 |
+
"loss": 0.055,
|
| 3202 |
+
"step": 51900
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 16.29,
|
| 3206 |
+
"learning_rate": 3.0183548606390207e-05,
|
| 3207 |
+
"loss": 0.0529,
|
| 3208 |
+
"step": 52000
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 16.32,
|
| 3212 |
+
"learning_rate": 2.9928619986403805e-05,
|
| 3213 |
+
"loss": 0.0519,
|
| 3214 |
+
"step": 52100
|
| 3215 |
+
},
|
| 3216 |
+
{
|
| 3217 |
+
"epoch": 16.36,
|
| 3218 |
+
"learning_rate": 2.96736913664174e-05,
|
| 3219 |
+
"loss": 0.0545,
|
| 3220 |
+
"step": 52200
|
| 3221 |
+
},
|
| 3222 |
+
{
|
| 3223 |
+
"epoch": 16.39,
|
| 3224 |
+
"learning_rate": 2.9418762746430996e-05,
|
| 3225 |
+
"loss": 0.0538,
|
| 3226 |
+
"step": 52300
|
| 3227 |
+
},
|
| 3228 |
+
{
|
| 3229 |
+
"epoch": 16.42,
|
| 3230 |
+
"learning_rate": 2.9163834126444594e-05,
|
| 3231 |
+
"loss": 0.0534,
|
| 3232 |
+
"step": 52400
|
| 3233 |
+
},
|
| 3234 |
+
{
|
| 3235 |
+
"epoch": 16.45,
|
| 3236 |
+
"learning_rate": 2.8908905506458188e-05,
|
| 3237 |
+
"loss": 0.0521,
|
| 3238 |
+
"step": 52500
|
| 3239 |
+
},
|
| 3240 |
+
{
|
| 3241 |
+
"epoch": 16.48,
|
| 3242 |
+
"learning_rate": 2.8653976886471785e-05,
|
| 3243 |
+
"loss": 0.0532,
|
| 3244 |
+
"step": 52600
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 16.51,
|
| 3248 |
+
"learning_rate": 2.8399048266485383e-05,
|
| 3249 |
+
"loss": 0.0547,
|
| 3250 |
+
"step": 52700
|
| 3251 |
+
},
|
| 3252 |
+
{
|
| 3253 |
+
"epoch": 16.54,
|
| 3254 |
+
"learning_rate": 2.8144119646498977e-05,
|
| 3255 |
+
"loss": 0.0545,
|
| 3256 |
+
"step": 52800
|
| 3257 |
+
},
|
| 3258 |
+
{
|
| 3259 |
+
"epoch": 16.58,
|
| 3260 |
+
"learning_rate": 2.7889191026512574e-05,
|
| 3261 |
+
"loss": 0.0519,
|
| 3262 |
+
"step": 52900
|
| 3263 |
+
},
|
| 3264 |
+
{
|
| 3265 |
+
"epoch": 16.61,
|
| 3266 |
+
"learning_rate": 2.763426240652617e-05,
|
| 3267 |
+
"loss": 0.054,
|
| 3268 |
+
"step": 53000
|
| 3269 |
+
},
|
| 3270 |
+
{
|
| 3271 |
+
"epoch": 16.64,
|
| 3272 |
+
"learning_rate": 2.7379333786539766e-05,
|
| 3273 |
+
"loss": 0.0532,
|
| 3274 |
+
"step": 53100
|
| 3275 |
+
},
|
| 3276 |
+
{
|
| 3277 |
+
"epoch": 16.67,
|
| 3278 |
+
"learning_rate": 2.7124405166553363e-05,
|
| 3279 |
+
"loss": 0.053,
|
| 3280 |
+
"step": 53200
|
| 3281 |
+
},
|
| 3282 |
+
{
|
| 3283 |
+
"epoch": 16.7,
|
| 3284 |
+
"learning_rate": 2.6869476546566958e-05,
|
| 3285 |
+
"loss": 0.0524,
|
| 3286 |
+
"step": 53300
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 16.73,
|
| 3290 |
+
"learning_rate": 2.6614547926580555e-05,
|
| 3291 |
+
"loss": 0.0531,
|
| 3292 |
+
"step": 53400
|
| 3293 |
+
},
|
| 3294 |
+
{
|
| 3295 |
+
"epoch": 16.76,
|
| 3296 |
+
"learning_rate": 2.6359619306594152e-05,
|
| 3297 |
+
"loss": 0.0519,
|
| 3298 |
+
"step": 53500
|
| 3299 |
+
},
|
| 3300 |
+
{
|
| 3301 |
+
"epoch": 16.79,
|
| 3302 |
+
"learning_rate": 2.6104690686607747e-05,
|
| 3303 |
+
"loss": 0.0526,
|
| 3304 |
+
"step": 53600
|
| 3305 |
+
},
|
| 3306 |
+
{
|
| 3307 |
+
"epoch": 16.83,
|
| 3308 |
+
"learning_rate": 2.5849762066621344e-05,
|
| 3309 |
+
"loss": 0.0513,
|
| 3310 |
+
"step": 53700
|
| 3311 |
+
},
|
| 3312 |
+
{
|
| 3313 |
+
"epoch": 16.86,
|
| 3314 |
+
"learning_rate": 2.5594833446634938e-05,
|
| 3315 |
+
"loss": 0.0514,
|
| 3316 |
+
"step": 53800
|
| 3317 |
+
},
|
| 3318 |
+
{
|
| 3319 |
+
"epoch": 16.89,
|
| 3320 |
+
"learning_rate": 2.5339904826648536e-05,
|
| 3321 |
+
"loss": 0.0525,
|
| 3322 |
+
"step": 53900
|
| 3323 |
+
},
|
| 3324 |
+
{
|
| 3325 |
+
"epoch": 16.92,
|
| 3326 |
+
"learning_rate": 2.5084976206662133e-05,
|
| 3327 |
+
"loss": 0.0514,
|
| 3328 |
+
"step": 54000
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 16.95,
|
| 3332 |
+
"learning_rate": 2.4830047586675727e-05,
|
| 3333 |
+
"loss": 0.0521,
|
| 3334 |
+
"step": 54100
|
| 3335 |
+
},
|
| 3336 |
+
{
|
| 3337 |
+
"epoch": 16.98,
|
| 3338 |
+
"learning_rate": 2.4575118966689328e-05,
|
| 3339 |
+
"loss": 0.0538,
|
| 3340 |
+
"step": 54200
|
| 3341 |
+
},
|
| 3342 |
+
{
|
| 3343 |
+
"epoch": 17.01,
|
| 3344 |
+
"learning_rate": 2.4320190346702922e-05,
|
| 3345 |
+
"loss": 0.0512,
|
| 3346 |
+
"step": 54300
|
| 3347 |
+
},
|
| 3348 |
+
{
|
| 3349 |
+
"epoch": 17.05,
|
| 3350 |
+
"learning_rate": 2.4065261726716516e-05,
|
| 3351 |
+
"loss": 0.0507,
|
| 3352 |
+
"step": 54400
|
| 3353 |
+
},
|
| 3354 |
+
{
|
| 3355 |
+
"epoch": 17.08,
|
| 3356 |
+
"learning_rate": 2.3810333106730114e-05,
|
| 3357 |
+
"loss": 0.0478,
|
| 3358 |
+
"step": 54500
|
| 3359 |
+
},
|
| 3360 |
+
{
|
| 3361 |
+
"epoch": 17.11,
|
| 3362 |
+
"learning_rate": 2.3555404486743708e-05,
|
| 3363 |
+
"loss": 0.0478,
|
| 3364 |
+
"step": 54600
|
| 3365 |
+
},
|
| 3366 |
+
{
|
| 3367 |
+
"epoch": 17.14,
|
| 3368 |
+
"learning_rate": 2.330047586675731e-05,
|
| 3369 |
+
"loss": 0.0485,
|
| 3370 |
+
"step": 54700
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 17.17,
|
| 3374 |
+
"learning_rate": 2.3045547246770903e-05,
|
| 3375 |
+
"loss": 0.047,
|
| 3376 |
+
"step": 54800
|
| 3377 |
+
},
|
| 3378 |
+
{
|
| 3379 |
+
"epoch": 17.2,
|
| 3380 |
+
"learning_rate": 2.2790618626784497e-05,
|
| 3381 |
+
"loss": 0.048,
|
| 3382 |
+
"step": 54900
|
| 3383 |
+
},
|
| 3384 |
+
{
|
| 3385 |
+
"epoch": 17.23,
|
| 3386 |
+
"learning_rate": 2.2535690006798098e-05,
|
| 3387 |
+
"loss": 0.0463,
|
| 3388 |
+
"step": 55000
|
| 3389 |
+
},
|
| 3390 |
+
{
|
| 3391 |
+
"epoch": 17.23,
|
| 3392 |
+
"eval_loss": 0.20283755660057068,
|
| 3393 |
+
"eval_runtime": 69.2702,
|
| 3394 |
+
"eval_samples_per_second": 129.074,
|
| 3395 |
+
"eval_steps_per_second": 2.021,
|
| 3396 |
+
"step": 55000
|
| 3397 |
+
},
|
| 3398 |
+
{
|
| 3399 |
+
"epoch": 17.26,
|
| 3400 |
+
"learning_rate": 2.2280761386811692e-05,
|
| 3401 |
+
"loss": 0.0484,
|
| 3402 |
+
"step": 55100
|
| 3403 |
+
},
|
| 3404 |
+
{
|
| 3405 |
+
"epoch": 17.3,
|
| 3406 |
+
"learning_rate": 2.2025832766825286e-05,
|
| 3407 |
+
"loss": 0.0484,
|
| 3408 |
+
"step": 55200
|
| 3409 |
+
},
|
| 3410 |
+
{
|
| 3411 |
+
"epoch": 17.33,
|
| 3412 |
+
"learning_rate": 2.1770904146838883e-05,
|
| 3413 |
+
"loss": 0.0494,
|
| 3414 |
+
"step": 55300
|
| 3415 |
+
},
|
| 3416 |
+
{
|
| 3417 |
+
"epoch": 17.36,
|
| 3418 |
+
"learning_rate": 2.1515975526852478e-05,
|
| 3419 |
+
"loss": 0.0482,
|
| 3420 |
+
"step": 55400
|
| 3421 |
+
},
|
| 3422 |
+
{
|
| 3423 |
+
"epoch": 17.39,
|
| 3424 |
+
"learning_rate": 2.126104690686608e-05,
|
| 3425 |
+
"loss": 0.0474,
|
| 3426 |
+
"step": 55500
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 17.42,
|
| 3430 |
+
"learning_rate": 2.1006118286879672e-05,
|
| 3431 |
+
"loss": 0.0457,
|
| 3432 |
+
"step": 55600
|
| 3433 |
+
},
|
| 3434 |
+
{
|
| 3435 |
+
"epoch": 17.45,
|
| 3436 |
+
"learning_rate": 2.0751189666893267e-05,
|
| 3437 |
+
"loss": 0.0497,
|
| 3438 |
+
"step": 55700
|
| 3439 |
+
},
|
| 3440 |
+
{
|
| 3441 |
+
"epoch": 17.48,
|
| 3442 |
+
"learning_rate": 2.0496261046906867e-05,
|
| 3443 |
+
"loss": 0.0466,
|
| 3444 |
+
"step": 55800
|
| 3445 |
+
},
|
| 3446 |
+
{
|
| 3447 |
+
"epoch": 17.52,
|
| 3448 |
+
"learning_rate": 2.024133242692046e-05,
|
| 3449 |
+
"loss": 0.0468,
|
| 3450 |
+
"step": 55900
|
| 3451 |
+
},
|
| 3452 |
+
{
|
| 3453 |
+
"epoch": 17.55,
|
| 3454 |
+
"learning_rate": 1.9986403806934056e-05,
|
| 3455 |
+
"loss": 0.0477,
|
| 3456 |
+
"step": 56000
|
| 3457 |
+
},
|
| 3458 |
+
{
|
| 3459 |
+
"epoch": 17.58,
|
| 3460 |
+
"learning_rate": 1.9731475186947653e-05,
|
| 3461 |
+
"loss": 0.0468,
|
| 3462 |
+
"step": 56100
|
| 3463 |
+
},
|
| 3464 |
+
{
|
| 3465 |
+
"epoch": 17.61,
|
| 3466 |
+
"learning_rate": 1.9476546566961247e-05,
|
| 3467 |
+
"loss": 0.0466,
|
| 3468 |
+
"step": 56200
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 17.64,
|
| 3472 |
+
"learning_rate": 1.9221617946974848e-05,
|
| 3473 |
+
"loss": 0.0484,
|
| 3474 |
+
"step": 56300
|
| 3475 |
+
},
|
| 3476 |
+
{
|
| 3477 |
+
"epoch": 17.67,
|
| 3478 |
+
"learning_rate": 1.8966689326988442e-05,
|
| 3479 |
+
"loss": 0.0441,
|
| 3480 |
+
"step": 56400
|
| 3481 |
+
},
|
| 3482 |
+
{
|
| 3483 |
+
"epoch": 17.7,
|
| 3484 |
+
"learning_rate": 1.871176070700204e-05,
|
| 3485 |
+
"loss": 0.0486,
|
| 3486 |
+
"step": 56500
|
| 3487 |
+
},
|
| 3488 |
+
{
|
| 3489 |
+
"epoch": 17.73,
|
| 3490 |
+
"learning_rate": 1.8456832087015634e-05,
|
| 3491 |
+
"loss": 0.0477,
|
| 3492 |
+
"step": 56600
|
| 3493 |
+
},
|
| 3494 |
+
{
|
| 3495 |
+
"epoch": 17.77,
|
| 3496 |
+
"learning_rate": 1.820190346702923e-05,
|
| 3497 |
+
"loss": 0.0471,
|
| 3498 |
+
"step": 56700
|
| 3499 |
+
},
|
| 3500 |
+
{
|
| 3501 |
+
"epoch": 17.8,
|
| 3502 |
+
"learning_rate": 1.7946974847042825e-05,
|
| 3503 |
+
"loss": 0.0454,
|
| 3504 |
+
"step": 56800
|
| 3505 |
+
},
|
| 3506 |
+
{
|
| 3507 |
+
"epoch": 17.83,
|
| 3508 |
+
"learning_rate": 1.7692046227056423e-05,
|
| 3509 |
+
"loss": 0.0456,
|
| 3510 |
+
"step": 56900
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 17.86,
|
| 3514 |
+
"learning_rate": 1.7437117607070017e-05,
|
| 3515 |
+
"loss": 0.0467,
|
| 3516 |
+
"step": 57000
|
| 3517 |
+
},
|
| 3518 |
+
{
|
| 3519 |
+
"epoch": 17.89,
|
| 3520 |
+
"learning_rate": 1.7182188987083614e-05,
|
| 3521 |
+
"loss": 0.0466,
|
| 3522 |
+
"step": 57100
|
| 3523 |
+
},
|
| 3524 |
+
{
|
| 3525 |
+
"epoch": 17.92,
|
| 3526 |
+
"learning_rate": 1.6927260367097212e-05,
|
| 3527 |
+
"loss": 0.046,
|
| 3528 |
+
"step": 57200
|
| 3529 |
+
},
|
| 3530 |
+
{
|
| 3531 |
+
"epoch": 17.95,
|
| 3532 |
+
"learning_rate": 1.667233174711081e-05,
|
| 3533 |
+
"loss": 0.0475,
|
| 3534 |
+
"step": 57300
|
| 3535 |
+
},
|
| 3536 |
+
{
|
| 3537 |
+
"epoch": 17.99,
|
| 3538 |
+
"learning_rate": 1.6417403127124403e-05,
|
| 3539 |
+
"loss": 0.0455,
|
| 3540 |
+
"step": 57400
|
| 3541 |
+
},
|
| 3542 |
+
{
|
| 3543 |
+
"epoch": 18.02,
|
| 3544 |
+
"learning_rate": 1.6162474507138e-05,
|
| 3545 |
+
"loss": 0.0447,
|
| 3546 |
+
"step": 57500
|
| 3547 |
+
},
|
| 3548 |
+
{
|
| 3549 |
+
"epoch": 18.05,
|
| 3550 |
+
"learning_rate": 1.5907545887151595e-05,
|
| 3551 |
+
"loss": 0.0422,
|
| 3552 |
+
"step": 57600
|
| 3553 |
+
},
|
| 3554 |
+
{
|
| 3555 |
+
"epoch": 18.08,
|
| 3556 |
+
"learning_rate": 1.5652617267165193e-05,
|
| 3557 |
+
"loss": 0.0432,
|
| 3558 |
+
"step": 57700
|
| 3559 |
+
},
|
| 3560 |
+
{
|
| 3561 |
+
"epoch": 18.11,
|
| 3562 |
+
"learning_rate": 1.5397688647178787e-05,
|
| 3563 |
+
"loss": 0.0423,
|
| 3564 |
+
"step": 57800
|
| 3565 |
+
},
|
| 3566 |
+
{
|
| 3567 |
+
"epoch": 18.14,
|
| 3568 |
+
"learning_rate": 1.5142760027192384e-05,
|
| 3569 |
+
"loss": 0.0427,
|
| 3570 |
+
"step": 57900
|
| 3571 |
+
},
|
| 3572 |
+
{
|
| 3573 |
+
"epoch": 18.17,
|
| 3574 |
+
"learning_rate": 1.4887831407205982e-05,
|
| 3575 |
+
"loss": 0.0445,
|
| 3576 |
+
"step": 58000
|
| 3577 |
+
},
|
| 3578 |
+
{
|
| 3579 |
+
"epoch": 18.2,
|
| 3580 |
+
"learning_rate": 1.4632902787219579e-05,
|
| 3581 |
+
"loss": 0.0437,
|
| 3582 |
+
"step": 58100
|
| 3583 |
+
},
|
| 3584 |
+
{
|
| 3585 |
+
"epoch": 18.24,
|
| 3586 |
+
"learning_rate": 1.4377974167233173e-05,
|
| 3587 |
+
"loss": 0.0433,
|
| 3588 |
+
"step": 58200
|
| 3589 |
+
},
|
| 3590 |
+
{
|
| 3591 |
+
"epoch": 18.27,
|
| 3592 |
+
"learning_rate": 1.4123045547246769e-05,
|
| 3593 |
+
"loss": 0.0447,
|
| 3594 |
+
"step": 58300
|
| 3595 |
+
},
|
| 3596 |
+
{
|
| 3597 |
+
"epoch": 18.3,
|
| 3598 |
+
"learning_rate": 1.3868116927260366e-05,
|
| 3599 |
+
"loss": 0.0417,
|
| 3600 |
+
"step": 58400
|
| 3601 |
+
},
|
| 3602 |
+
{
|
| 3603 |
+
"epoch": 18.33,
|
| 3604 |
+
"learning_rate": 1.3613188307273964e-05,
|
| 3605 |
+
"loss": 0.0435,
|
| 3606 |
+
"step": 58500
|
| 3607 |
+
},
|
| 3608 |
+
{
|
| 3609 |
+
"epoch": 18.36,
|
| 3610 |
+
"learning_rate": 1.3358259687287558e-05,
|
| 3611 |
+
"loss": 0.0423,
|
| 3612 |
+
"step": 58600
|
| 3613 |
+
},
|
| 3614 |
+
{
|
| 3615 |
+
"epoch": 18.39,
|
| 3616 |
+
"learning_rate": 1.3103331067301154e-05,
|
| 3617 |
+
"loss": 0.0429,
|
| 3618 |
+
"step": 58700
|
| 3619 |
+
},
|
| 3620 |
+
{
|
| 3621 |
+
"epoch": 18.42,
|
| 3622 |
+
"learning_rate": 1.2848402447314751e-05,
|
| 3623 |
+
"loss": 0.0417,
|
| 3624 |
+
"step": 58800
|
| 3625 |
+
},
|
| 3626 |
+
{
|
| 3627 |
+
"epoch": 18.46,
|
| 3628 |
+
"learning_rate": 1.2593473827328349e-05,
|
| 3629 |
+
"loss": 0.0422,
|
| 3630 |
+
"step": 58900
|
| 3631 |
+
},
|
| 3632 |
+
{
|
| 3633 |
+
"epoch": 18.49,
|
| 3634 |
+
"learning_rate": 1.2338545207341943e-05,
|
| 3635 |
+
"loss": 0.0433,
|
| 3636 |
+
"step": 59000
|
| 3637 |
+
},
|
| 3638 |
+
{
|
| 3639 |
+
"epoch": 18.52,
|
| 3640 |
+
"learning_rate": 1.2083616587355539e-05,
|
| 3641 |
+
"loss": 0.0407,
|
| 3642 |
+
"step": 59100
|
| 3643 |
+
},
|
| 3644 |
+
{
|
| 3645 |
+
"epoch": 18.55,
|
| 3646 |
+
"learning_rate": 1.1828687967369136e-05,
|
| 3647 |
+
"loss": 0.0432,
|
| 3648 |
+
"step": 59200
|
| 3649 |
+
},
|
| 3650 |
+
{
|
| 3651 |
+
"epoch": 18.58,
|
| 3652 |
+
"learning_rate": 1.1573759347382734e-05,
|
| 3653 |
+
"loss": 0.0417,
|
| 3654 |
+
"step": 59300
|
| 3655 |
+
},
|
| 3656 |
+
{
|
| 3657 |
+
"epoch": 18.61,
|
| 3658 |
+
"learning_rate": 1.1318830727396328e-05,
|
| 3659 |
+
"loss": 0.0419,
|
| 3660 |
+
"step": 59400
|
| 3661 |
+
},
|
| 3662 |
+
{
|
| 3663 |
+
"epoch": 18.64,
|
| 3664 |
+
"learning_rate": 1.1063902107409923e-05,
|
| 3665 |
+
"loss": 0.0415,
|
| 3666 |
+
"step": 59500
|
| 3667 |
+
},
|
| 3668 |
+
{
|
| 3669 |
+
"epoch": 18.67,
|
| 3670 |
+
"learning_rate": 1.0808973487423521e-05,
|
| 3671 |
+
"loss": 0.0407,
|
| 3672 |
+
"step": 59600
|
| 3673 |
+
},
|
| 3674 |
+
{
|
| 3675 |
+
"epoch": 18.71,
|
| 3676 |
+
"learning_rate": 1.0554044867437118e-05,
|
| 3677 |
+
"loss": 0.0409,
|
| 3678 |
+
"step": 59700
|
| 3679 |
+
},
|
| 3680 |
+
{
|
| 3681 |
+
"epoch": 18.74,
|
| 3682 |
+
"learning_rate": 1.0299116247450713e-05,
|
| 3683 |
+
"loss": 0.0424,
|
| 3684 |
+
"step": 59800
|
| 3685 |
+
},
|
| 3686 |
+
{
|
| 3687 |
+
"epoch": 18.77,
|
| 3688 |
+
"learning_rate": 1.0044187627464308e-05,
|
| 3689 |
+
"loss": 0.0419,
|
| 3690 |
+
"step": 59900
|
| 3691 |
+
},
|
| 3692 |
+
{
|
| 3693 |
+
"epoch": 18.8,
|
| 3694 |
+
"learning_rate": 9.789259007477906e-06,
|
| 3695 |
+
"loss": 0.04,
|
| 3696 |
+
"step": 60000
|
| 3697 |
+
},
|
| 3698 |
+
{
|
| 3699 |
+
"epoch": 18.8,
|
| 3700 |
+
"eval_loss": 0.1957029402256012,
|
| 3701 |
+
"eval_runtime": 84.9296,
|
| 3702 |
+
"eval_samples_per_second": 105.275,
|
| 3703 |
+
"eval_steps_per_second": 1.648,
|
| 3704 |
+
"step": 60000
|
| 3705 |
+
},
|
| 3706 |
+
{
|
| 3707 |
+
"epoch": 18.83,
|
| 3708 |
+
"learning_rate": 9.534330387491503e-06,
|
| 3709 |
+
"loss": 0.0405,
|
| 3710 |
+
"step": 60100
|
| 3711 |
+
},
|
| 3712 |
+
{
|
| 3713 |
+
"epoch": 18.86,
|
| 3714 |
+
"learning_rate": 9.279401767505097e-06,
|
| 3715 |
+
"loss": 0.0421,
|
| 3716 |
+
"step": 60200
|
| 3717 |
+
},
|
| 3718 |
+
{
|
| 3719 |
+
"epoch": 18.89,
|
| 3720 |
+
"learning_rate": 9.024473147518693e-06,
|
| 3721 |
+
"loss": 0.041,
|
| 3722 |
+
"step": 60300
|
| 3723 |
+
},
|
| 3724 |
+
{
|
| 3725 |
+
"epoch": 18.93,
|
| 3726 |
+
"learning_rate": 8.76954452753229e-06,
|
| 3727 |
+
"loss": 0.0406,
|
| 3728 |
+
"step": 60400
|
| 3729 |
+
},
|
| 3730 |
+
{
|
| 3731 |
+
"epoch": 18.96,
|
| 3732 |
+
"learning_rate": 8.514615907545886e-06,
|
| 3733 |
+
"loss": 0.0407,
|
| 3734 |
+
"step": 60500
|
| 3735 |
+
},
|
| 3736 |
+
{
|
| 3737 |
+
"epoch": 18.99,
|
| 3738 |
+
"learning_rate": 8.259687287559482e-06,
|
| 3739 |
+
"loss": 0.0403,
|
| 3740 |
+
"step": 60600
|
| 3741 |
+
},
|
| 3742 |
+
{
|
| 3743 |
+
"epoch": 19.02,
|
| 3744 |
+
"learning_rate": 8.004758667573078e-06,
|
| 3745 |
+
"loss": 0.0394,
|
| 3746 |
+
"step": 60700
|
| 3747 |
+
},
|
| 3748 |
+
{
|
| 3749 |
+
"epoch": 19.05,
|
| 3750 |
+
"learning_rate": 7.749830047586676e-06,
|
| 3751 |
+
"loss": 0.0392,
|
| 3752 |
+
"step": 60800
|
| 3753 |
+
},
|
| 3754 |
+
{
|
| 3755 |
+
"epoch": 19.08,
|
| 3756 |
+
"learning_rate": 7.494901427600271e-06,
|
| 3757 |
+
"loss": 0.0384,
|
| 3758 |
+
"step": 60900
|
| 3759 |
+
},
|
| 3760 |
+
{
|
| 3761 |
+
"epoch": 19.11,
|
| 3762 |
+
"learning_rate": 7.239972807613868e-06,
|
| 3763 |
+
"loss": 0.0382,
|
| 3764 |
+
"step": 61000
|
| 3765 |
+
},
|
| 3766 |
+
{
|
| 3767 |
+
"epoch": 19.14,
|
| 3768 |
+
"learning_rate": 6.985044187627464e-06,
|
| 3769 |
+
"loss": 0.0387,
|
| 3770 |
+
"step": 61100
|
| 3771 |
+
},
|
| 3772 |
+
{
|
| 3773 |
+
"epoch": 19.18,
|
| 3774 |
+
"learning_rate": 6.73011556764106e-06,
|
| 3775 |
+
"loss": 0.0396,
|
| 3776 |
+
"step": 61200
|
| 3777 |
+
},
|
| 3778 |
+
{
|
| 3779 |
+
"epoch": 19.21,
|
| 3780 |
+
"learning_rate": 6.475186947654656e-06,
|
| 3781 |
+
"loss": 0.0373,
|
| 3782 |
+
"step": 61300
|
| 3783 |
+
},
|
| 3784 |
+
{
|
| 3785 |
+
"epoch": 19.24,
|
| 3786 |
+
"learning_rate": 6.220258327668253e-06,
|
| 3787 |
+
"loss": 0.0367,
|
| 3788 |
+
"step": 61400
|
| 3789 |
+
},
|
| 3790 |
+
{
|
| 3791 |
+
"epoch": 19.27,
|
| 3792 |
+
"learning_rate": 5.9653297076818486e-06,
|
| 3793 |
+
"loss": 0.0403,
|
| 3794 |
+
"step": 61500
|
| 3795 |
+
},
|
| 3796 |
+
{
|
| 3797 |
+
"epoch": 19.3,
|
| 3798 |
+
"learning_rate": 5.710401087695445e-06,
|
| 3799 |
+
"loss": 0.0378,
|
| 3800 |
+
"step": 61600
|
| 3801 |
+
},
|
| 3802 |
+
{
|
| 3803 |
+
"epoch": 19.33,
|
| 3804 |
+
"learning_rate": 5.455472467709041e-06,
|
| 3805 |
+
"loss": 0.0383,
|
| 3806 |
+
"step": 61700
|
| 3807 |
+
},
|
| 3808 |
+
{
|
| 3809 |
+
"epoch": 19.36,
|
| 3810 |
+
"learning_rate": 5.200543847722638e-06,
|
| 3811 |
+
"loss": 0.0382,
|
| 3812 |
+
"step": 61800
|
| 3813 |
+
},
|
| 3814 |
+
{
|
| 3815 |
+
"epoch": 19.4,
|
| 3816 |
+
"learning_rate": 4.945615227736233e-06,
|
| 3817 |
+
"loss": 0.0398,
|
| 3818 |
+
"step": 61900
|
| 3819 |
+
},
|
| 3820 |
+
{
|
| 3821 |
+
"epoch": 19.43,
|
| 3822 |
+
"learning_rate": 4.69068660774983e-06,
|
| 3823 |
+
"loss": 0.0374,
|
| 3824 |
+
"step": 62000
|
| 3825 |
+
},
|
| 3826 |
+
{
|
| 3827 |
+
"epoch": 19.46,
|
| 3828 |
+
"learning_rate": 4.435757987763426e-06,
|
| 3829 |
+
"loss": 0.0383,
|
| 3830 |
+
"step": 62100
|
| 3831 |
+
},
|
| 3832 |
+
{
|
| 3833 |
+
"epoch": 19.49,
|
| 3834 |
+
"learning_rate": 4.180829367777022e-06,
|
| 3835 |
+
"loss": 0.0377,
|
| 3836 |
+
"step": 62200
|
| 3837 |
+
},
|
| 3838 |
+
{
|
| 3839 |
+
"epoch": 19.52,
|
| 3840 |
+
"learning_rate": 3.925900747790618e-06,
|
| 3841 |
+
"loss": 0.0377,
|
| 3842 |
+
"step": 62300
|
| 3843 |
+
},
|
| 3844 |
+
{
|
| 3845 |
+
"epoch": 19.55,
|
| 3846 |
+
"learning_rate": 3.6709721278042145e-06,
|
| 3847 |
+
"loss": 0.04,
|
| 3848 |
+
"step": 62400
|
| 3849 |
+
},
|
| 3850 |
+
{
|
| 3851 |
+
"epoch": 19.58,
|
| 3852 |
+
"learning_rate": 3.4160435078178107e-06,
|
| 3853 |
+
"loss": 0.0379,
|
| 3854 |
+
"step": 62500
|
| 3855 |
+
},
|
| 3856 |
+
{
|
| 3857 |
+
"epoch": 19.61,
|
| 3858 |
+
"learning_rate": 3.161114887831407e-06,
|
| 3859 |
+
"loss": 0.0371,
|
| 3860 |
+
"step": 62600
|
| 3861 |
+
},
|
| 3862 |
+
{
|
| 3863 |
+
"epoch": 19.65,
|
| 3864 |
+
"learning_rate": 2.906186267845003e-06,
|
| 3865 |
+
"loss": 0.0367,
|
| 3866 |
+
"step": 62700
|
| 3867 |
+
},
|
| 3868 |
+
{
|
| 3869 |
+
"epoch": 19.68,
|
| 3870 |
+
"learning_rate": 2.6512576478585993e-06,
|
| 3871 |
+
"loss": 0.0381,
|
| 3872 |
+
"step": 62800
|
| 3873 |
+
},
|
| 3874 |
+
{
|
| 3875 |
+
"epoch": 19.71,
|
| 3876 |
+
"learning_rate": 2.3963290278721955e-06,
|
| 3877 |
+
"loss": 0.0372,
|
| 3878 |
+
"step": 62900
|
| 3879 |
+
},
|
| 3880 |
+
{
|
| 3881 |
+
"epoch": 19.74,
|
| 3882 |
+
"learning_rate": 2.1414004078857917e-06,
|
| 3883 |
+
"loss": 0.0385,
|
| 3884 |
+
"step": 63000
|
| 3885 |
+
},
|
| 3886 |
+
{
|
| 3887 |
+
"epoch": 19.77,
|
| 3888 |
+
"learning_rate": 1.886471787899388e-06,
|
| 3889 |
+
"loss": 0.0377,
|
| 3890 |
+
"step": 63100
|
| 3891 |
+
},
|
| 3892 |
+
{
|
| 3893 |
+
"epoch": 19.8,
|
| 3894 |
+
"learning_rate": 1.6315431679129842e-06,
|
| 3895 |
+
"loss": 0.0377,
|
| 3896 |
+
"step": 63200
|
| 3897 |
+
},
|
| 3898 |
+
{
|
| 3899 |
+
"epoch": 19.83,
|
| 3900 |
+
"learning_rate": 1.3766145479265804e-06,
|
| 3901 |
+
"loss": 0.0366,
|
| 3902 |
+
"step": 63300
|
| 3903 |
+
},
|
| 3904 |
+
{
|
| 3905 |
+
"epoch": 19.87,
|
| 3906 |
+
"learning_rate": 1.1216859279401766e-06,
|
| 3907 |
+
"loss": 0.0382,
|
| 3908 |
+
"step": 63400
|
| 3909 |
+
},
|
| 3910 |
+
{
|
| 3911 |
+
"epoch": 19.9,
|
| 3912 |
+
"learning_rate": 8.667573079537729e-07,
|
| 3913 |
+
"loss": 0.0371,
|
| 3914 |
+
"step": 63500
|
| 3915 |
+
},
|
| 3916 |
+
{
|
| 3917 |
+
"epoch": 19.93,
|
| 3918 |
+
"learning_rate": 6.118286879673691e-07,
|
| 3919 |
+
"loss": 0.037,
|
| 3920 |
+
"step": 63600
|
| 3921 |
+
},
|
| 3922 |
+
{
|
| 3923 |
+
"epoch": 19.96,
|
| 3924 |
+
"learning_rate": 3.569000679809653e-07,
|
| 3925 |
+
"loss": 0.0367,
|
| 3926 |
+
"step": 63700
|
| 3927 |
+
},
|
| 3928 |
+
{
|
| 3929 |
+
"epoch": 19.99,
|
| 3930 |
+
"learning_rate": 1.0197144799456151e-07,
|
| 3931 |
+
"loss": 0.0366,
|
| 3932 |
+
"step": 63800
|
| 3933 |
+
},
|
| 3934 |
+
{
|
| 3935 |
+
"epoch": 20.0,
|
| 3936 |
+
"step": 63840,
|
| 3937 |
+
"total_flos": 1.325011889553408e+20,
|
| 3938 |
+
"train_loss": 0.13081293349950235,
|
| 3939 |
+
"train_runtime": 214172.6098,
|
| 3940 |
+
"train_samples_per_second": 76.308,
|
| 3941 |
+
"train_steps_per_second": 0.298
|
| 3942 |
+
}
|
| 3943 |
+
],
|
| 3944 |
+
"max_steps": 63840,
|
| 3945 |
+
"num_train_epochs": 21,
|
| 3946 |
+
"total_flos": 1.325011889553408e+20,
|
| 3947 |
+
"trial_name": null,
|
| 3948 |
+
"trial_params": null
|
| 3949 |
+
}
|