End of training
Browse files- all_results.json +16 -0
- eval_results.json +11 -0
- runs/Feb15_09-10-04_906ba4838a3d/events.out.tfevents.1676496099.906ba4838a3d.452.2 +3 -0
- train_results.json +8 -0
- trainer_state.json +2593 -0
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.7188003869719445,
|
| 4 |
+
"eval_f1": 0.718989971086903,
|
| 5 |
+
"eval_loss": 0.8270788192749023,
|
| 6 |
+
"eval_precision": 0.7219837313936599,
|
| 7 |
+
"eval_recall": 0.7188003869719445,
|
| 8 |
+
"eval_runtime": 138.4655,
|
| 9 |
+
"eval_samples_per_second": 22.395,
|
| 10 |
+
"eval_steps_per_second": 0.701,
|
| 11 |
+
"total_flos": 3.843964371168166e+19,
|
| 12 |
+
"train_loss": 0.8177650722031741,
|
| 13 |
+
"train_runtime": 43638.4535,
|
| 14 |
+
"train_samples_per_second": 11.367,
|
| 15 |
+
"train_steps_per_second": 0.089
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.7188003869719445,
|
| 4 |
+
"eval_f1": 0.718989971086903,
|
| 5 |
+
"eval_loss": 0.8270788192749023,
|
| 6 |
+
"eval_precision": 0.7219837313936599,
|
| 7 |
+
"eval_recall": 0.7188003869719445,
|
| 8 |
+
"eval_runtime": 138.4655,
|
| 9 |
+
"eval_samples_per_second": 22.395,
|
| 10 |
+
"eval_steps_per_second": 0.701
|
| 11 |
+
}
|
runs/Feb15_09-10-04_906ba4838a3d/events.out.tfevents.1676496099.906ba4838a3d.452.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:915588e1ae24d5506015f30b6571d3b7b91868ef23548b7e50665377fa8beddb
|
| 3 |
+
size 512
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"total_flos": 3.843964371168166e+19,
|
| 4 |
+
"train_loss": 0.8177650722031741,
|
| 5 |
+
"train_runtime": 43638.4535,
|
| 6 |
+
"train_samples_per_second": 11.367,
|
| 7 |
+
"train_steps_per_second": 0.089
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2593 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.718989971086903,
|
| 3 |
+
"best_model_checkpoint": "20E-affecthq/checkpoint-2716",
|
| 4 |
+
"epoch": 20.0,
|
| 5 |
+
"global_step": 3880,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.05,
|
| 12 |
+
"learning_rate": 2.577319587628866e-07,
|
| 13 |
+
"loss": 2.0744,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.1,
|
| 18 |
+
"learning_rate": 5.154639175257732e-07,
|
| 19 |
+
"loss": 2.0746,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.15,
|
| 24 |
+
"learning_rate": 7.731958762886599e-07,
|
| 25 |
+
"loss": 2.0716,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.21,
|
| 30 |
+
"learning_rate": 1.0309278350515464e-06,
|
| 31 |
+
"loss": 2.0661,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.26,
|
| 36 |
+
"learning_rate": 1.288659793814433e-06,
|
| 37 |
+
"loss": 2.0648,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.31,
|
| 42 |
+
"learning_rate": 1.5463917525773197e-06,
|
| 43 |
+
"loss": 2.0612,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.36,
|
| 48 |
+
"learning_rate": 1.8041237113402063e-06,
|
| 49 |
+
"loss": 2.0582,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.41,
|
| 54 |
+
"learning_rate": 2.061855670103093e-06,
|
| 55 |
+
"loss": 2.0522,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.46,
|
| 60 |
+
"learning_rate": 2.3195876288659796e-06,
|
| 61 |
+
"loss": 2.0516,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.52,
|
| 66 |
+
"learning_rate": 2.577319587628866e-06,
|
| 67 |
+
"loss": 2.0432,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.57,
|
| 72 |
+
"learning_rate": 2.8350515463917527e-06,
|
| 73 |
+
"loss": 2.0425,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.62,
|
| 78 |
+
"learning_rate": 3.0927835051546395e-06,
|
| 79 |
+
"loss": 2.0368,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.67,
|
| 84 |
+
"learning_rate": 3.350515463917526e-06,
|
| 85 |
+
"loss": 2.0229,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.72,
|
| 90 |
+
"learning_rate": 3.6082474226804126e-06,
|
| 91 |
+
"loss": 2.0163,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.77,
|
| 96 |
+
"learning_rate": 3.865979381443299e-06,
|
| 97 |
+
"loss": 1.9999,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.82,
|
| 102 |
+
"learning_rate": 4.123711340206186e-06,
|
| 103 |
+
"loss": 1.9825,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.88,
|
| 108 |
+
"learning_rate": 4.381443298969073e-06,
|
| 109 |
+
"loss": 1.9702,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.93,
|
| 114 |
+
"learning_rate": 4.639175257731959e-06,
|
| 115 |
+
"loss": 1.9417,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.98,
|
| 120 |
+
"learning_rate": 4.8969072164948455e-06,
|
| 121 |
+
"loss": 1.9149,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 1.0,
|
| 126 |
+
"eval_accuracy": 0.37504030957755563,
|
| 127 |
+
"eval_f1": 0.30450075259542253,
|
| 128 |
+
"eval_loss": 1.8887449502944946,
|
| 129 |
+
"eval_precision": 0.34133971468533547,
|
| 130 |
+
"eval_recall": 0.37504030957755563,
|
| 131 |
+
"eval_runtime": 204.064,
|
| 132 |
+
"eval_samples_per_second": 15.196,
|
| 133 |
+
"eval_steps_per_second": 0.475,
|
| 134 |
+
"step": 194
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 1.03,
|
| 138 |
+
"learning_rate": 5.154639175257732e-06,
|
| 139 |
+
"loss": 1.8825,
|
| 140 |
+
"step": 200
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 1.08,
|
| 144 |
+
"learning_rate": 5.412371134020619e-06,
|
| 145 |
+
"loss": 1.8587,
|
| 146 |
+
"step": 210
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 1.13,
|
| 150 |
+
"learning_rate": 5.670103092783505e-06,
|
| 151 |
+
"loss": 1.7938,
|
| 152 |
+
"step": 220
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 1.19,
|
| 156 |
+
"learning_rate": 5.927835051546393e-06,
|
| 157 |
+
"loss": 1.7462,
|
| 158 |
+
"step": 230
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.24,
|
| 162 |
+
"learning_rate": 6.185567010309279e-06,
|
| 163 |
+
"loss": 1.6973,
|
| 164 |
+
"step": 240
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 1.29,
|
| 168 |
+
"learning_rate": 6.443298969072166e-06,
|
| 169 |
+
"loss": 1.6757,
|
| 170 |
+
"step": 250
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 1.34,
|
| 174 |
+
"learning_rate": 6.701030927835052e-06,
|
| 175 |
+
"loss": 1.6265,
|
| 176 |
+
"step": 260
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 1.39,
|
| 180 |
+
"learning_rate": 6.958762886597939e-06,
|
| 181 |
+
"loss": 1.5975,
|
| 182 |
+
"step": 270
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 1.44,
|
| 186 |
+
"learning_rate": 7.216494845360825e-06,
|
| 187 |
+
"loss": 1.5301,
|
| 188 |
+
"step": 280
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 1.49,
|
| 192 |
+
"learning_rate": 7.474226804123712e-06,
|
| 193 |
+
"loss": 1.5235,
|
| 194 |
+
"step": 290
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 1.55,
|
| 198 |
+
"learning_rate": 7.731958762886599e-06,
|
| 199 |
+
"loss": 1.4745,
|
| 200 |
+
"step": 300
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.6,
|
| 204 |
+
"learning_rate": 7.989690721649486e-06,
|
| 205 |
+
"loss": 1.4207,
|
| 206 |
+
"step": 310
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.65,
|
| 210 |
+
"learning_rate": 8.247422680412371e-06,
|
| 211 |
+
"loss": 1.4171,
|
| 212 |
+
"step": 320
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.7,
|
| 216 |
+
"learning_rate": 8.505154639175259e-06,
|
| 217 |
+
"loss": 1.419,
|
| 218 |
+
"step": 330
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 1.75,
|
| 222 |
+
"learning_rate": 8.762886597938146e-06,
|
| 223 |
+
"loss": 1.3424,
|
| 224 |
+
"step": 340
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 1.8,
|
| 228 |
+
"learning_rate": 9.020618556701031e-06,
|
| 229 |
+
"loss": 1.3403,
|
| 230 |
+
"step": 350
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 1.86,
|
| 234 |
+
"learning_rate": 9.278350515463918e-06,
|
| 235 |
+
"loss": 1.3536,
|
| 236 |
+
"step": 360
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 1.91,
|
| 240 |
+
"learning_rate": 9.536082474226806e-06,
|
| 241 |
+
"loss": 1.3036,
|
| 242 |
+
"step": 370
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.96,
|
| 246 |
+
"learning_rate": 9.793814432989691e-06,
|
| 247 |
+
"loss": 1.2903,
|
| 248 |
+
"step": 380
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 2.0,
|
| 252 |
+
"eval_accuracy": 0.5791680103192518,
|
| 253 |
+
"eval_f1": 0.5526365213062105,
|
| 254 |
+
"eval_loss": 1.2485474348068237,
|
| 255 |
+
"eval_precision": 0.5725810384525178,
|
| 256 |
+
"eval_recall": 0.5791680103192518,
|
| 257 |
+
"eval_runtime": 141.2499,
|
| 258 |
+
"eval_samples_per_second": 21.954,
|
| 259 |
+
"eval_steps_per_second": 0.687,
|
| 260 |
+
"step": 388
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 2.01,
|
| 264 |
+
"learning_rate": 9.994272623138604e-06,
|
| 265 |
+
"loss": 1.2612,
|
| 266 |
+
"step": 390
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 2.06,
|
| 270 |
+
"learning_rate": 9.965635738831616e-06,
|
| 271 |
+
"loss": 1.2295,
|
| 272 |
+
"step": 400
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.11,
|
| 276 |
+
"learning_rate": 9.93699885452463e-06,
|
| 277 |
+
"loss": 1.2367,
|
| 278 |
+
"step": 410
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 2.16,
|
| 282 |
+
"learning_rate": 9.908361970217641e-06,
|
| 283 |
+
"loss": 1.2015,
|
| 284 |
+
"step": 420
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 2.22,
|
| 288 |
+
"learning_rate": 9.879725085910653e-06,
|
| 289 |
+
"loss": 1.1918,
|
| 290 |
+
"step": 430
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 2.27,
|
| 294 |
+
"learning_rate": 9.851088201603667e-06,
|
| 295 |
+
"loss": 1.2132,
|
| 296 |
+
"step": 440
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 2.32,
|
| 300 |
+
"learning_rate": 9.822451317296679e-06,
|
| 301 |
+
"loss": 1.1727,
|
| 302 |
+
"step": 450
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 2.37,
|
| 306 |
+
"learning_rate": 9.793814432989691e-06,
|
| 307 |
+
"loss": 1.1598,
|
| 308 |
+
"step": 460
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 2.42,
|
| 312 |
+
"learning_rate": 9.765177548682703e-06,
|
| 313 |
+
"loss": 1.137,
|
| 314 |
+
"step": 470
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.47,
|
| 318 |
+
"learning_rate": 9.736540664375717e-06,
|
| 319 |
+
"loss": 1.167,
|
| 320 |
+
"step": 480
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 2.53,
|
| 324 |
+
"learning_rate": 9.707903780068729e-06,
|
| 325 |
+
"loss": 1.1318,
|
| 326 |
+
"step": 490
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 2.58,
|
| 330 |
+
"learning_rate": 9.679266895761742e-06,
|
| 331 |
+
"loss": 1.1026,
|
| 332 |
+
"step": 500
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 2.63,
|
| 336 |
+
"learning_rate": 9.650630011454754e-06,
|
| 337 |
+
"loss": 1.1069,
|
| 338 |
+
"step": 510
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 2.68,
|
| 342 |
+
"learning_rate": 9.621993127147768e-06,
|
| 343 |
+
"loss": 1.0721,
|
| 344 |
+
"step": 520
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 2.73,
|
| 348 |
+
"learning_rate": 9.59335624284078e-06,
|
| 349 |
+
"loss": 1.0903,
|
| 350 |
+
"step": 530
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 2.78,
|
| 354 |
+
"learning_rate": 9.564719358533792e-06,
|
| 355 |
+
"loss": 1.0951,
|
| 356 |
+
"step": 540
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.84,
|
| 360 |
+
"learning_rate": 9.536082474226806e-06,
|
| 361 |
+
"loss": 1.1402,
|
| 362 |
+
"step": 550
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.89,
|
| 366 |
+
"learning_rate": 9.507445589919818e-06,
|
| 367 |
+
"loss": 1.1012,
|
| 368 |
+
"step": 560
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 2.94,
|
| 372 |
+
"learning_rate": 9.47880870561283e-06,
|
| 373 |
+
"loss": 1.0767,
|
| 374 |
+
"step": 570
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 2.99,
|
| 378 |
+
"learning_rate": 9.450171821305843e-06,
|
| 379 |
+
"loss": 1.071,
|
| 380 |
+
"step": 580
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 3.0,
|
| 384 |
+
"eval_accuracy": 0.6320541760722348,
|
| 385 |
+
"eval_f1": 0.6228282492696803,
|
| 386 |
+
"eval_loss": 1.0587016344070435,
|
| 387 |
+
"eval_precision": 0.6258393540507952,
|
| 388 |
+
"eval_recall": 0.6320541760722348,
|
| 389 |
+
"eval_runtime": 156.4705,
|
| 390 |
+
"eval_samples_per_second": 19.818,
|
| 391 |
+
"eval_steps_per_second": 0.62,
|
| 392 |
+
"step": 582
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 3.04,
|
| 396 |
+
"learning_rate": 9.421534936998855e-06,
|
| 397 |
+
"loss": 1.0438,
|
| 398 |
+
"step": 590
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 3.09,
|
| 402 |
+
"learning_rate": 9.392898052691867e-06,
|
| 403 |
+
"loss": 1.0081,
|
| 404 |
+
"step": 600
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 3.14,
|
| 408 |
+
"learning_rate": 9.36426116838488e-06,
|
| 409 |
+
"loss": 1.0547,
|
| 410 |
+
"step": 610
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 3.2,
|
| 414 |
+
"learning_rate": 9.335624284077893e-06,
|
| 415 |
+
"loss": 1.0588,
|
| 416 |
+
"step": 620
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 3.25,
|
| 420 |
+
"learning_rate": 9.306987399770906e-06,
|
| 421 |
+
"loss": 1.0339,
|
| 422 |
+
"step": 630
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 3.3,
|
| 426 |
+
"learning_rate": 9.278350515463918e-06,
|
| 427 |
+
"loss": 1.0542,
|
| 428 |
+
"step": 640
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 3.35,
|
| 432 |
+
"learning_rate": 9.249713631156932e-06,
|
| 433 |
+
"loss": 1.0186,
|
| 434 |
+
"step": 650
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 3.4,
|
| 438 |
+
"learning_rate": 9.221076746849944e-06,
|
| 439 |
+
"loss": 0.9947,
|
| 440 |
+
"step": 660
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 3.45,
|
| 444 |
+
"learning_rate": 9.192439862542956e-06,
|
| 445 |
+
"loss": 1.0304,
|
| 446 |
+
"step": 670
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 3.51,
|
| 450 |
+
"learning_rate": 9.163802978235968e-06,
|
| 451 |
+
"loss": 0.9644,
|
| 452 |
+
"step": 680
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 3.56,
|
| 456 |
+
"learning_rate": 9.135166093928982e-06,
|
| 457 |
+
"loss": 1.0023,
|
| 458 |
+
"step": 690
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 3.61,
|
| 462 |
+
"learning_rate": 9.106529209621994e-06,
|
| 463 |
+
"loss": 0.9797,
|
| 464 |
+
"step": 700
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 3.66,
|
| 468 |
+
"learning_rate": 9.077892325315006e-06,
|
| 469 |
+
"loss": 0.9779,
|
| 470 |
+
"step": 710
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 3.71,
|
| 474 |
+
"learning_rate": 9.04925544100802e-06,
|
| 475 |
+
"loss": 1.0429,
|
| 476 |
+
"step": 720
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 3.76,
|
| 480 |
+
"learning_rate": 9.020618556701031e-06,
|
| 481 |
+
"loss": 0.9848,
|
| 482 |
+
"step": 730
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 3.81,
|
| 486 |
+
"learning_rate": 8.991981672394045e-06,
|
| 487 |
+
"loss": 0.9981,
|
| 488 |
+
"step": 740
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 3.87,
|
| 492 |
+
"learning_rate": 8.963344788087057e-06,
|
| 493 |
+
"loss": 0.9571,
|
| 494 |
+
"step": 750
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 3.92,
|
| 498 |
+
"learning_rate": 8.93470790378007e-06,
|
| 499 |
+
"loss": 0.9786,
|
| 500 |
+
"step": 760
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 3.97,
|
| 504 |
+
"learning_rate": 8.90607101947308e-06,
|
| 505 |
+
"loss": 1.0185,
|
| 506 |
+
"step": 770
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 4.0,
|
| 510 |
+
"eval_accuracy": 0.6617220251531764,
|
| 511 |
+
"eval_f1": 0.6552882607911024,
|
| 512 |
+
"eval_loss": 0.9817498326301575,
|
| 513 |
+
"eval_precision": 0.6583817535052932,
|
| 514 |
+
"eval_recall": 0.6617220251531764,
|
| 515 |
+
"eval_runtime": 140.4425,
|
| 516 |
+
"eval_samples_per_second": 22.08,
|
| 517 |
+
"eval_steps_per_second": 0.691,
|
| 518 |
+
"step": 776
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 4.02,
|
| 522 |
+
"learning_rate": 8.877434135166094e-06,
|
| 523 |
+
"loss": 0.955,
|
| 524 |
+
"step": 780
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 4.07,
|
| 528 |
+
"learning_rate": 8.848797250859106e-06,
|
| 529 |
+
"loss": 0.9995,
|
| 530 |
+
"step": 790
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 4.12,
|
| 534 |
+
"learning_rate": 8.82016036655212e-06,
|
| 535 |
+
"loss": 0.9615,
|
| 536 |
+
"step": 800
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 4.18,
|
| 540 |
+
"learning_rate": 8.791523482245132e-06,
|
| 541 |
+
"loss": 0.8775,
|
| 542 |
+
"step": 810
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 4.23,
|
| 546 |
+
"learning_rate": 8.762886597938146e-06,
|
| 547 |
+
"loss": 0.948,
|
| 548 |
+
"step": 820
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 4.28,
|
| 552 |
+
"learning_rate": 8.734249713631158e-06,
|
| 553 |
+
"loss": 0.9255,
|
| 554 |
+
"step": 830
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 4.33,
|
| 558 |
+
"learning_rate": 8.70561282932417e-06,
|
| 559 |
+
"loss": 0.9125,
|
| 560 |
+
"step": 840
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 4.38,
|
| 564 |
+
"learning_rate": 8.676975945017183e-06,
|
| 565 |
+
"loss": 0.9153,
|
| 566 |
+
"step": 850
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 4.43,
|
| 570 |
+
"learning_rate": 8.648339060710195e-06,
|
| 571 |
+
"loss": 0.9227,
|
| 572 |
+
"step": 860
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 4.48,
|
| 576 |
+
"learning_rate": 8.619702176403209e-06,
|
| 577 |
+
"loss": 0.945,
|
| 578 |
+
"step": 870
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 4.54,
|
| 582 |
+
"learning_rate": 8.591065292096221e-06,
|
| 583 |
+
"loss": 0.9121,
|
| 584 |
+
"step": 880
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 4.59,
|
| 588 |
+
"learning_rate": 8.562428407789233e-06,
|
| 589 |
+
"loss": 0.9311,
|
| 590 |
+
"step": 890
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 4.64,
|
| 594 |
+
"learning_rate": 8.533791523482245e-06,
|
| 595 |
+
"loss": 0.8944,
|
| 596 |
+
"step": 900
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 4.69,
|
| 600 |
+
"learning_rate": 8.505154639175259e-06,
|
| 601 |
+
"loss": 0.9093,
|
| 602 |
+
"step": 910
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 4.74,
|
| 606 |
+
"learning_rate": 8.47651775486827e-06,
|
| 607 |
+
"loss": 0.8956,
|
| 608 |
+
"step": 920
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 4.79,
|
| 612 |
+
"learning_rate": 8.447880870561284e-06,
|
| 613 |
+
"loss": 0.963,
|
| 614 |
+
"step": 930
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 4.85,
|
| 618 |
+
"learning_rate": 8.419243986254296e-06,
|
| 619 |
+
"loss": 0.9149,
|
| 620 |
+
"step": 940
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 4.9,
|
| 624 |
+
"learning_rate": 8.39060710194731e-06,
|
| 625 |
+
"loss": 0.909,
|
| 626 |
+
"step": 950
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 4.95,
|
| 630 |
+
"learning_rate": 8.361970217640322e-06,
|
| 631 |
+
"loss": 0.8628,
|
| 632 |
+
"step": 960
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 5.0,
|
| 636 |
+
"learning_rate": 8.333333333333334e-06,
|
| 637 |
+
"loss": 0.894,
|
| 638 |
+
"step": 970
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 5.0,
|
| 642 |
+
"eval_accuracy": 0.6868752015478877,
|
| 643 |
+
"eval_f1": 0.6820473818962892,
|
| 644 |
+
"eval_loss": 0.9292588829994202,
|
| 645 |
+
"eval_precision": 0.6872338048544774,
|
| 646 |
+
"eval_recall": 0.6868752015478877,
|
| 647 |
+
"eval_runtime": 140.2334,
|
| 648 |
+
"eval_samples_per_second": 22.113,
|
| 649 |
+
"eval_steps_per_second": 0.692,
|
| 650 |
+
"step": 970
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 5.05,
|
| 654 |
+
"learning_rate": 8.304696449026347e-06,
|
| 655 |
+
"loss": 0.8574,
|
| 656 |
+
"step": 980
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 5.1,
|
| 660 |
+
"learning_rate": 8.27605956471936e-06,
|
| 661 |
+
"loss": 0.8418,
|
| 662 |
+
"step": 990
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 5.15,
|
| 666 |
+
"learning_rate": 8.247422680412371e-06,
|
| 667 |
+
"loss": 0.8805,
|
| 668 |
+
"step": 1000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 5.21,
|
| 672 |
+
"learning_rate": 8.218785796105383e-06,
|
| 673 |
+
"loss": 0.8624,
|
| 674 |
+
"step": 1010
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 5.26,
|
| 678 |
+
"learning_rate": 8.190148911798397e-06,
|
| 679 |
+
"loss": 0.8462,
|
| 680 |
+
"step": 1020
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 5.31,
|
| 684 |
+
"learning_rate": 8.161512027491409e-06,
|
| 685 |
+
"loss": 0.8716,
|
| 686 |
+
"step": 1030
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 5.36,
|
| 690 |
+
"learning_rate": 8.132875143184423e-06,
|
| 691 |
+
"loss": 0.8289,
|
| 692 |
+
"step": 1040
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 5.41,
|
| 696 |
+
"learning_rate": 8.104238258877435e-06,
|
| 697 |
+
"loss": 0.8918,
|
| 698 |
+
"step": 1050
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 5.46,
|
| 702 |
+
"learning_rate": 8.075601374570448e-06,
|
| 703 |
+
"loss": 0.8383,
|
| 704 |
+
"step": 1060
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 5.52,
|
| 708 |
+
"learning_rate": 8.04696449026346e-06,
|
| 709 |
+
"loss": 0.8729,
|
| 710 |
+
"step": 1070
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 5.57,
|
| 714 |
+
"learning_rate": 8.018327605956472e-06,
|
| 715 |
+
"loss": 0.8548,
|
| 716 |
+
"step": 1080
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 5.62,
|
| 720 |
+
"learning_rate": 7.989690721649486e-06,
|
| 721 |
+
"loss": 0.8519,
|
| 722 |
+
"step": 1090
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 5.67,
|
| 726 |
+
"learning_rate": 7.961053837342498e-06,
|
| 727 |
+
"loss": 0.869,
|
| 728 |
+
"step": 1100
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 5.72,
|
| 732 |
+
"learning_rate": 7.93241695303551e-06,
|
| 733 |
+
"loss": 0.8671,
|
| 734 |
+
"step": 1110
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 5.77,
|
| 738 |
+
"learning_rate": 7.903780068728523e-06,
|
| 739 |
+
"loss": 0.8645,
|
| 740 |
+
"step": 1120
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 5.82,
|
| 744 |
+
"learning_rate": 7.875143184421535e-06,
|
| 745 |
+
"loss": 0.8585,
|
| 746 |
+
"step": 1130
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 5.88,
|
| 750 |
+
"learning_rate": 7.846506300114547e-06,
|
| 751 |
+
"loss": 0.8622,
|
| 752 |
+
"step": 1140
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 5.93,
|
| 756 |
+
"learning_rate": 7.817869415807561e-06,
|
| 757 |
+
"loss": 0.8516,
|
| 758 |
+
"step": 1150
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 5.98,
|
| 762 |
+
"learning_rate": 7.789232531500573e-06,
|
| 763 |
+
"loss": 0.8283,
|
| 764 |
+
"step": 1160
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 6.0,
|
| 768 |
+
"eval_accuracy": 0.6936472105772331,
|
| 769 |
+
"eval_f1": 0.6905055219394208,
|
| 770 |
+
"eval_loss": 0.8881023526191711,
|
| 771 |
+
"eval_precision": 0.6928623941074218,
|
| 772 |
+
"eval_recall": 0.6936472105772331,
|
| 773 |
+
"eval_runtime": 139.4491,
|
| 774 |
+
"eval_samples_per_second": 22.238,
|
| 775 |
+
"eval_steps_per_second": 0.696,
|
| 776 |
+
"step": 1164
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 6.03,
|
| 780 |
+
"learning_rate": 7.760595647193587e-06,
|
| 781 |
+
"loss": 0.8177,
|
| 782 |
+
"step": 1170
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 6.08,
|
| 786 |
+
"learning_rate": 7.731958762886599e-06,
|
| 787 |
+
"loss": 0.8277,
|
| 788 |
+
"step": 1180
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 6.13,
|
| 792 |
+
"learning_rate": 7.703321878579612e-06,
|
| 793 |
+
"loss": 0.7713,
|
| 794 |
+
"step": 1190
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 6.19,
|
| 798 |
+
"learning_rate": 7.674684994272624e-06,
|
| 799 |
+
"loss": 0.8008,
|
| 800 |
+
"step": 1200
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 6.24,
|
| 804 |
+
"learning_rate": 7.646048109965636e-06,
|
| 805 |
+
"loss": 0.823,
|
| 806 |
+
"step": 1210
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 6.29,
|
| 810 |
+
"learning_rate": 7.617411225658648e-06,
|
| 811 |
+
"loss": 0.8256,
|
| 812 |
+
"step": 1220
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 6.34,
|
| 816 |
+
"learning_rate": 7.588774341351661e-06,
|
| 817 |
+
"loss": 0.8161,
|
| 818 |
+
"step": 1230
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 6.39,
|
| 822 |
+
"learning_rate": 7.560137457044674e-06,
|
| 823 |
+
"loss": 0.7932,
|
| 824 |
+
"step": 1240
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 6.44,
|
| 828 |
+
"learning_rate": 7.531500572737687e-06,
|
| 829 |
+
"loss": 0.829,
|
| 830 |
+
"step": 1250
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 6.49,
|
| 834 |
+
"learning_rate": 7.5028636884306995e-06,
|
| 835 |
+
"loss": 0.853,
|
| 836 |
+
"step": 1260
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 6.55,
|
| 840 |
+
"learning_rate": 7.474226804123712e-06,
|
| 841 |
+
"loss": 0.8209,
|
| 842 |
+
"step": 1270
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 6.6,
|
| 846 |
+
"learning_rate": 7.445589919816725e-06,
|
| 847 |
+
"loss": 0.8186,
|
| 848 |
+
"step": 1280
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 6.65,
|
| 852 |
+
"learning_rate": 7.416953035509737e-06,
|
| 853 |
+
"loss": 0.7944,
|
| 854 |
+
"step": 1290
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 6.7,
|
| 858 |
+
"learning_rate": 7.38831615120275e-06,
|
| 859 |
+
"loss": 0.7968,
|
| 860 |
+
"step": 1300
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 6.75,
|
| 864 |
+
"learning_rate": 7.359679266895763e-06,
|
| 865 |
+
"loss": 0.7907,
|
| 866 |
+
"step": 1310
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 6.8,
|
| 870 |
+
"learning_rate": 7.331042382588775e-06,
|
| 871 |
+
"loss": 0.7876,
|
| 872 |
+
"step": 1320
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 6.86,
|
| 876 |
+
"learning_rate": 7.3024054982817875e-06,
|
| 877 |
+
"loss": 0.8202,
|
| 878 |
+
"step": 1330
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 6.91,
|
| 882 |
+
"learning_rate": 7.2737686139747995e-06,
|
| 883 |
+
"loss": 0.7491,
|
| 884 |
+
"step": 1340
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 6.96,
|
| 888 |
+
"learning_rate": 7.245131729667812e-06,
|
| 889 |
+
"loss": 0.8185,
|
| 890 |
+
"step": 1350
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 7.0,
|
| 894 |
+
"eval_accuracy": 0.6981618832634634,
|
| 895 |
+
"eval_f1": 0.6987690565427545,
|
| 896 |
+
"eval_loss": 0.8659195303916931,
|
| 897 |
+
"eval_precision": 0.7011005120637903,
|
| 898 |
+
"eval_recall": 0.6981618832634634,
|
| 899 |
+
"eval_runtime": 137.6906,
|
| 900 |
+
"eval_samples_per_second": 22.522,
|
| 901 |
+
"eval_steps_per_second": 0.704,
|
| 902 |
+
"step": 1358
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 7.01,
|
| 906 |
+
"learning_rate": 7.216494845360825e-06,
|
| 907 |
+
"loss": 0.783,
|
| 908 |
+
"step": 1360
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 7.06,
|
| 912 |
+
"learning_rate": 7.187857961053838e-06,
|
| 913 |
+
"loss": 0.7655,
|
| 914 |
+
"step": 1370
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 7.11,
|
| 918 |
+
"learning_rate": 7.159221076746851e-06,
|
| 919 |
+
"loss": 0.7166,
|
| 920 |
+
"step": 1380
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 7.16,
|
| 924 |
+
"learning_rate": 7.130584192439864e-06,
|
| 925 |
+
"loss": 0.7497,
|
| 926 |
+
"step": 1390
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 7.22,
|
| 930 |
+
"learning_rate": 7.101947308132876e-06,
|
| 931 |
+
"loss": 0.7692,
|
| 932 |
+
"step": 1400
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 7.27,
|
| 936 |
+
"learning_rate": 7.073310423825888e-06,
|
| 937 |
+
"loss": 0.7567,
|
| 938 |
+
"step": 1410
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 7.32,
|
| 942 |
+
"learning_rate": 7.044673539518901e-06,
|
| 943 |
+
"loss": 0.8078,
|
| 944 |
+
"step": 1420
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 7.37,
|
| 948 |
+
"learning_rate": 7.016036655211913e-06,
|
| 949 |
+
"loss": 0.7567,
|
| 950 |
+
"step": 1430
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 7.42,
|
| 954 |
+
"learning_rate": 6.987399770904926e-06,
|
| 955 |
+
"loss": 0.7553,
|
| 956 |
+
"step": 1440
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 7.47,
|
| 960 |
+
"learning_rate": 6.958762886597939e-06,
|
| 961 |
+
"loss": 0.7315,
|
| 962 |
+
"step": 1450
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 7.53,
|
| 966 |
+
"learning_rate": 6.930126002290951e-06,
|
| 967 |
+
"loss": 0.7585,
|
| 968 |
+
"step": 1460
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 7.58,
|
| 972 |
+
"learning_rate": 6.901489117983964e-06,
|
| 973 |
+
"loss": 0.7772,
|
| 974 |
+
"step": 1470
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 7.63,
|
| 978 |
+
"learning_rate": 6.872852233676976e-06,
|
| 979 |
+
"loss": 0.7388,
|
| 980 |
+
"step": 1480
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 7.68,
|
| 984 |
+
"learning_rate": 6.844215349369989e-06,
|
| 985 |
+
"loss": 0.7477,
|
| 986 |
+
"step": 1490
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 7.73,
|
| 990 |
+
"learning_rate": 6.815578465063002e-06,
|
| 991 |
+
"loss": 0.7693,
|
| 992 |
+
"step": 1500
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 7.78,
|
| 996 |
+
"learning_rate": 6.786941580756015e-06,
|
| 997 |
+
"loss": 0.7444,
|
| 998 |
+
"step": 1510
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 7.84,
|
| 1002 |
+
"learning_rate": 6.758304696449028e-06,
|
| 1003 |
+
"loss": 0.8008,
|
| 1004 |
+
"step": 1520
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 7.89,
|
| 1008 |
+
"learning_rate": 6.72966781214204e-06,
|
| 1009 |
+
"loss": 0.7655,
|
| 1010 |
+
"step": 1530
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 7.94,
|
| 1014 |
+
"learning_rate": 6.701030927835052e-06,
|
| 1015 |
+
"loss": 0.8314,
|
| 1016 |
+
"step": 1540
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 7.99,
|
| 1020 |
+
"learning_rate": 6.672394043528064e-06,
|
| 1021 |
+
"loss": 0.7499,
|
| 1022 |
+
"step": 1550
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 8.0,
|
| 1026 |
+
"eval_accuracy": 0.7046114156723637,
|
| 1027 |
+
"eval_f1": 0.7021460447498534,
|
| 1028 |
+
"eval_loss": 0.8557855486869812,
|
| 1029 |
+
"eval_precision": 0.7050032399049077,
|
| 1030 |
+
"eval_recall": 0.7046114156723637,
|
| 1031 |
+
"eval_runtime": 138.2055,
|
| 1032 |
+
"eval_samples_per_second": 22.438,
|
| 1033 |
+
"eval_steps_per_second": 0.702,
|
| 1034 |
+
"step": 1552
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 8.04,
|
| 1038 |
+
"learning_rate": 6.643757159221077e-06,
|
| 1039 |
+
"loss": 0.7519,
|
| 1040 |
+
"step": 1560
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 8.09,
|
| 1044 |
+
"learning_rate": 6.61512027491409e-06,
|
| 1045 |
+
"loss": 0.719,
|
| 1046 |
+
"step": 1570
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 8.14,
|
| 1050 |
+
"learning_rate": 6.586483390607102e-06,
|
| 1051 |
+
"loss": 0.7143,
|
| 1052 |
+
"step": 1580
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 8.2,
|
| 1056 |
+
"learning_rate": 6.557846506300115e-06,
|
| 1057 |
+
"loss": 0.7502,
|
| 1058 |
+
"step": 1590
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 8.25,
|
| 1062 |
+
"learning_rate": 6.529209621993128e-06,
|
| 1063 |
+
"loss": 0.753,
|
| 1064 |
+
"step": 1600
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 8.3,
|
| 1068 |
+
"learning_rate": 6.5005727376861405e-06,
|
| 1069 |
+
"loss": 0.7136,
|
| 1070 |
+
"step": 1610
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 8.35,
|
| 1074 |
+
"learning_rate": 6.471935853379153e-06,
|
| 1075 |
+
"loss": 0.7153,
|
| 1076 |
+
"step": 1620
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 8.4,
|
| 1080 |
+
"learning_rate": 6.443298969072166e-06,
|
| 1081 |
+
"loss": 0.7353,
|
| 1082 |
+
"step": 1630
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 8.45,
|
| 1086 |
+
"learning_rate": 6.414662084765179e-06,
|
| 1087 |
+
"loss": 0.7612,
|
| 1088 |
+
"step": 1640
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 8.51,
|
| 1092 |
+
"learning_rate": 6.38602520045819e-06,
|
| 1093 |
+
"loss": 0.7023,
|
| 1094 |
+
"step": 1650
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 8.56,
|
| 1098 |
+
"learning_rate": 6.357388316151203e-06,
|
| 1099 |
+
"loss": 0.7415,
|
| 1100 |
+
"step": 1660
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 8.61,
|
| 1104 |
+
"learning_rate": 6.328751431844216e-06,
|
| 1105 |
+
"loss": 0.7073,
|
| 1106 |
+
"step": 1670
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 8.66,
|
| 1110 |
+
"learning_rate": 6.3001145475372285e-06,
|
| 1111 |
+
"loss": 0.7085,
|
| 1112 |
+
"step": 1680
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 8.71,
|
| 1116 |
+
"learning_rate": 6.271477663230241e-06,
|
| 1117 |
+
"loss": 0.7469,
|
| 1118 |
+
"step": 1690
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 8.76,
|
| 1122 |
+
"learning_rate": 6.242840778923253e-06,
|
| 1123 |
+
"loss": 0.7371,
|
| 1124 |
+
"step": 1700
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 8.81,
|
| 1128 |
+
"learning_rate": 6.214203894616266e-06,
|
| 1129 |
+
"loss": 0.7228,
|
| 1130 |
+
"step": 1710
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 8.87,
|
| 1134 |
+
"learning_rate": 6.185567010309279e-06,
|
| 1135 |
+
"loss": 0.7249,
|
| 1136 |
+
"step": 1720
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 8.92,
|
| 1140 |
+
"learning_rate": 6.156930126002292e-06,
|
| 1141 |
+
"loss": 0.7462,
|
| 1142 |
+
"step": 1730
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 8.97,
|
| 1146 |
+
"learning_rate": 6.1282932416953046e-06,
|
| 1147 |
+
"loss": 0.7219,
|
| 1148 |
+
"step": 1740
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 9.0,
|
| 1152 |
+
"eval_accuracy": 0.7123508545630441,
|
| 1153 |
+
"eval_f1": 0.7127491151695062,
|
| 1154 |
+
"eval_loss": 0.8398727774620056,
|
| 1155 |
+
"eval_precision": 0.7164935273335432,
|
| 1156 |
+
"eval_recall": 0.7123508545630441,
|
| 1157 |
+
"eval_runtime": 138.5399,
|
| 1158 |
+
"eval_samples_per_second": 22.383,
|
| 1159 |
+
"eval_steps_per_second": 0.7,
|
| 1160 |
+
"step": 1746
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 9.02,
|
| 1164 |
+
"learning_rate": 6.099656357388317e-06,
|
| 1165 |
+
"loss": 0.7051,
|
| 1166 |
+
"step": 1750
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 9.07,
|
| 1170 |
+
"learning_rate": 6.0710194730813285e-06,
|
| 1171 |
+
"loss": 0.7038,
|
| 1172 |
+
"step": 1760
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 9.12,
|
| 1176 |
+
"learning_rate": 6.042382588774341e-06,
|
| 1177 |
+
"loss": 0.6517,
|
| 1178 |
+
"step": 1770
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 9.18,
|
| 1182 |
+
"learning_rate": 6.013745704467354e-06,
|
| 1183 |
+
"loss": 0.7214,
|
| 1184 |
+
"step": 1780
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 9.23,
|
| 1188 |
+
"learning_rate": 5.985108820160367e-06,
|
| 1189 |
+
"loss": 0.698,
|
| 1190 |
+
"step": 1790
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 9.28,
|
| 1194 |
+
"learning_rate": 5.95647193585338e-06,
|
| 1195 |
+
"loss": 0.6854,
|
| 1196 |
+
"step": 1800
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 9.33,
|
| 1200 |
+
"learning_rate": 5.927835051546393e-06,
|
| 1201 |
+
"loss": 0.7,
|
| 1202 |
+
"step": 1810
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 9.38,
|
| 1206 |
+
"learning_rate": 5.8991981672394046e-06,
|
| 1207 |
+
"loss": 0.684,
|
| 1208 |
+
"step": 1820
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 9.43,
|
| 1212 |
+
"learning_rate": 5.870561282932417e-06,
|
| 1213 |
+
"loss": 0.693,
|
| 1214 |
+
"step": 1830
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 9.48,
|
| 1218 |
+
"learning_rate": 5.84192439862543e-06,
|
| 1219 |
+
"loss": 0.6796,
|
| 1220 |
+
"step": 1840
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 9.54,
|
| 1224 |
+
"learning_rate": 5.813287514318443e-06,
|
| 1225 |
+
"loss": 0.7016,
|
| 1226 |
+
"step": 1850
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 9.59,
|
| 1230 |
+
"learning_rate": 5.784650630011456e-06,
|
| 1231 |
+
"loss": 0.7032,
|
| 1232 |
+
"step": 1860
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 9.64,
|
| 1236 |
+
"learning_rate": 5.756013745704467e-06,
|
| 1237 |
+
"loss": 0.7215,
|
| 1238 |
+
"step": 1870
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 9.69,
|
| 1242 |
+
"learning_rate": 5.72737686139748e-06,
|
| 1243 |
+
"loss": 0.7012,
|
| 1244 |
+
"step": 1880
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 9.74,
|
| 1248 |
+
"learning_rate": 5.698739977090493e-06,
|
| 1249 |
+
"loss": 0.68,
|
| 1250 |
+
"step": 1890
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 9.79,
|
| 1254 |
+
"learning_rate": 5.670103092783505e-06,
|
| 1255 |
+
"loss": 0.6938,
|
| 1256 |
+
"step": 1900
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 9.85,
|
| 1260 |
+
"learning_rate": 5.641466208476518e-06,
|
| 1261 |
+
"loss": 0.6876,
|
| 1262 |
+
"step": 1910
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 9.9,
|
| 1266 |
+
"learning_rate": 5.612829324169531e-06,
|
| 1267 |
+
"loss": 0.7367,
|
| 1268 |
+
"step": 1920
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 9.95,
|
| 1272 |
+
"learning_rate": 5.584192439862544e-06,
|
| 1273 |
+
"loss": 0.6787,
|
| 1274 |
+
"step": 1930
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 10.0,
|
| 1278 |
+
"learning_rate": 5.555555555555557e-06,
|
| 1279 |
+
"loss": 0.7382,
|
| 1280 |
+
"step": 1940
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 10.0,
|
| 1284 |
+
"eval_accuracy": 0.7158980973879394,
|
| 1285 |
+
"eval_f1": 0.7145145595526569,
|
| 1286 |
+
"eval_loss": 0.8300238251686096,
|
| 1287 |
+
"eval_precision": 0.7183831182263194,
|
| 1288 |
+
"eval_recall": 0.7158980973879394,
|
| 1289 |
+
"eval_runtime": 142.0327,
|
| 1290 |
+
"eval_samples_per_second": 21.833,
|
| 1291 |
+
"eval_steps_per_second": 0.683,
|
| 1292 |
+
"step": 1940
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 10.05,
|
| 1296 |
+
"learning_rate": 5.526918671248569e-06,
|
| 1297 |
+
"loss": 0.6854,
|
| 1298 |
+
"step": 1950
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 10.1,
|
| 1302 |
+
"learning_rate": 5.4982817869415815e-06,
|
| 1303 |
+
"loss": 0.7064,
|
| 1304 |
+
"step": 1960
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 10.15,
|
| 1308 |
+
"learning_rate": 5.469644902634594e-06,
|
| 1309 |
+
"loss": 0.6548,
|
| 1310 |
+
"step": 1970
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 10.21,
|
| 1314 |
+
"learning_rate": 5.441008018327606e-06,
|
| 1315 |
+
"loss": 0.6742,
|
| 1316 |
+
"step": 1980
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 10.26,
|
| 1320 |
+
"learning_rate": 5.412371134020619e-06,
|
| 1321 |
+
"loss": 0.6549,
|
| 1322 |
+
"step": 1990
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 10.31,
|
| 1326 |
+
"learning_rate": 5.383734249713631e-06,
|
| 1327 |
+
"loss": 0.6867,
|
| 1328 |
+
"step": 2000
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 10.36,
|
| 1332 |
+
"learning_rate": 5.355097365406644e-06,
|
| 1333 |
+
"loss": 0.6595,
|
| 1334 |
+
"step": 2010
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 10.41,
|
| 1338 |
+
"learning_rate": 5.326460481099657e-06,
|
| 1339 |
+
"loss": 0.6721,
|
| 1340 |
+
"step": 2020
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 10.46,
|
| 1344 |
+
"learning_rate": 5.2978235967926695e-06,
|
| 1345 |
+
"loss": 0.6652,
|
| 1346 |
+
"step": 2030
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 10.52,
|
| 1350 |
+
"learning_rate": 5.269186712485682e-06,
|
| 1351 |
+
"loss": 0.6689,
|
| 1352 |
+
"step": 2040
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 10.57,
|
| 1356 |
+
"learning_rate": 5.240549828178695e-06,
|
| 1357 |
+
"loss": 0.6372,
|
| 1358 |
+
"step": 2050
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 10.62,
|
| 1362 |
+
"learning_rate": 5.211912943871708e-06,
|
| 1363 |
+
"loss": 0.7036,
|
| 1364 |
+
"step": 2060
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 10.67,
|
| 1368 |
+
"learning_rate": 5.18327605956472e-06,
|
| 1369 |
+
"loss": 0.6809,
|
| 1370 |
+
"step": 2070
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 10.72,
|
| 1374 |
+
"learning_rate": 5.154639175257732e-06,
|
| 1375 |
+
"loss": 0.7015,
|
| 1376 |
+
"step": 2080
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 10.77,
|
| 1380 |
+
"learning_rate": 5.126002290950745e-06,
|
| 1381 |
+
"loss": 0.6779,
|
| 1382 |
+
"step": 2090
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 10.82,
|
| 1386 |
+
"learning_rate": 5.0973654066437575e-06,
|
| 1387 |
+
"loss": 0.6516,
|
| 1388 |
+
"step": 2100
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 10.88,
|
| 1392 |
+
"learning_rate": 5.06872852233677e-06,
|
| 1393 |
+
"loss": 0.6445,
|
| 1394 |
+
"step": 2110
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 10.93,
|
| 1398 |
+
"learning_rate": 5.040091638029782e-06,
|
| 1399 |
+
"loss": 0.6939,
|
| 1400 |
+
"step": 2120
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 10.98,
|
| 1404 |
+
"learning_rate": 5.011454753722795e-06,
|
| 1405 |
+
"loss": 0.6392,
|
| 1406 |
+
"step": 2130
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 11.0,
|
| 1410 |
+
"eval_accuracy": 0.708803611738149,
|
| 1411 |
+
"eval_f1": 0.7094972389297926,
|
| 1412 |
+
"eval_loss": 0.8328733444213867,
|
| 1413 |
+
"eval_precision": 0.7134851917714098,
|
| 1414 |
+
"eval_recall": 0.708803611738149,
|
| 1415 |
+
"eval_runtime": 136.2862,
|
| 1416 |
+
"eval_samples_per_second": 22.754,
|
| 1417 |
+
"eval_steps_per_second": 0.712,
|
| 1418 |
+
"step": 2134
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 11.03,
|
| 1422 |
+
"learning_rate": 4.982817869415808e-06,
|
| 1423 |
+
"loss": 0.6682,
|
| 1424 |
+
"step": 2140
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 11.08,
|
| 1428 |
+
"learning_rate": 4.954180985108821e-06,
|
| 1429 |
+
"loss": 0.6482,
|
| 1430 |
+
"step": 2150
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 11.13,
|
| 1434 |
+
"learning_rate": 4.9255441008018336e-06,
|
| 1435 |
+
"loss": 0.6229,
|
| 1436 |
+
"step": 2160
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 11.19,
|
| 1440 |
+
"learning_rate": 4.8969072164948455e-06,
|
| 1441 |
+
"loss": 0.6611,
|
| 1442 |
+
"step": 2170
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 11.24,
|
| 1446 |
+
"learning_rate": 4.868270332187858e-06,
|
| 1447 |
+
"loss": 0.6164,
|
| 1448 |
+
"step": 2180
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 11.29,
|
| 1452 |
+
"learning_rate": 4.839633447880871e-06,
|
| 1453 |
+
"loss": 0.6623,
|
| 1454 |
+
"step": 2190
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 11.34,
|
| 1458 |
+
"learning_rate": 4.810996563573884e-06,
|
| 1459 |
+
"loss": 0.6415,
|
| 1460 |
+
"step": 2200
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 11.39,
|
| 1464 |
+
"learning_rate": 4.782359679266896e-06,
|
| 1465 |
+
"loss": 0.6349,
|
| 1466 |
+
"step": 2210
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 11.44,
|
| 1470 |
+
"learning_rate": 4.753722794959909e-06,
|
| 1471 |
+
"loss": 0.6506,
|
| 1472 |
+
"step": 2220
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 11.49,
|
| 1476 |
+
"learning_rate": 4.725085910652922e-06,
|
| 1477 |
+
"loss": 0.6366,
|
| 1478 |
+
"step": 2230
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 11.55,
|
| 1482 |
+
"learning_rate": 4.6964490263459336e-06,
|
| 1483 |
+
"loss": 0.6533,
|
| 1484 |
+
"step": 2240
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 11.6,
|
| 1488 |
+
"learning_rate": 4.667812142038946e-06,
|
| 1489 |
+
"loss": 0.6704,
|
| 1490 |
+
"step": 2250
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 11.65,
|
| 1494 |
+
"learning_rate": 4.639175257731959e-06,
|
| 1495 |
+
"loss": 0.6482,
|
| 1496 |
+
"step": 2260
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 11.7,
|
| 1500 |
+
"learning_rate": 4.610538373424972e-06,
|
| 1501 |
+
"loss": 0.5899,
|
| 1502 |
+
"step": 2270
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 11.75,
|
| 1506 |
+
"learning_rate": 4.581901489117984e-06,
|
| 1507 |
+
"loss": 0.6463,
|
| 1508 |
+
"step": 2280
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 11.8,
|
| 1512 |
+
"learning_rate": 4.553264604810997e-06,
|
| 1513 |
+
"loss": 0.6575,
|
| 1514 |
+
"step": 2290
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 11.86,
|
| 1518 |
+
"learning_rate": 4.52462772050401e-06,
|
| 1519 |
+
"loss": 0.6422,
|
| 1520 |
+
"step": 2300
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 11.91,
|
| 1524 |
+
"learning_rate": 4.4959908361970224e-06,
|
| 1525 |
+
"loss": 0.6736,
|
| 1526 |
+
"step": 2310
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 11.96,
|
| 1530 |
+
"learning_rate": 4.467353951890035e-06,
|
| 1531 |
+
"loss": 0.6549,
|
| 1532 |
+
"step": 2320
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 12.0,
|
| 1536 |
+
"eval_accuracy": 0.7133182844243793,
|
| 1537 |
+
"eval_f1": 0.711971866511754,
|
| 1538 |
+
"eval_loss": 0.8296553492546082,
|
| 1539 |
+
"eval_precision": 0.7134785190322149,
|
| 1540 |
+
"eval_recall": 0.7133182844243793,
|
| 1541 |
+
"eval_runtime": 142.0833,
|
| 1542 |
+
"eval_samples_per_second": 21.825,
|
| 1543 |
+
"eval_steps_per_second": 0.683,
|
| 1544 |
+
"step": 2328
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 12.01,
|
| 1548 |
+
"learning_rate": 4.438717067583047e-06,
|
| 1549 |
+
"loss": 0.5999,
|
| 1550 |
+
"step": 2330
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 12.06,
|
| 1554 |
+
"learning_rate": 4.41008018327606e-06,
|
| 1555 |
+
"loss": 0.6216,
|
| 1556 |
+
"step": 2340
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 12.11,
|
| 1560 |
+
"learning_rate": 4.381443298969073e-06,
|
| 1561 |
+
"loss": 0.6263,
|
| 1562 |
+
"step": 2350
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 12.16,
|
| 1566 |
+
"learning_rate": 4.352806414662085e-06,
|
| 1567 |
+
"loss": 0.6119,
|
| 1568 |
+
"step": 2360
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 12.22,
|
| 1572 |
+
"learning_rate": 4.324169530355098e-06,
|
| 1573 |
+
"loss": 0.6132,
|
| 1574 |
+
"step": 2370
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 12.27,
|
| 1578 |
+
"learning_rate": 4.2955326460481105e-06,
|
| 1579 |
+
"loss": 0.6269,
|
| 1580 |
+
"step": 2380
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 12.32,
|
| 1584 |
+
"learning_rate": 4.2668957617411224e-06,
|
| 1585 |
+
"loss": 0.6457,
|
| 1586 |
+
"step": 2390
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 12.37,
|
| 1590 |
+
"learning_rate": 4.238258877434135e-06,
|
| 1591 |
+
"loss": 0.6278,
|
| 1592 |
+
"step": 2400
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 12.42,
|
| 1596 |
+
"learning_rate": 4.209621993127148e-06,
|
| 1597 |
+
"loss": 0.6119,
|
| 1598 |
+
"step": 2410
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 12.47,
|
| 1602 |
+
"learning_rate": 4.180985108820161e-06,
|
| 1603 |
+
"loss": 0.6198,
|
| 1604 |
+
"step": 2420
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 12.53,
|
| 1608 |
+
"learning_rate": 4.152348224513174e-06,
|
| 1609 |
+
"loss": 0.6232,
|
| 1610 |
+
"step": 2430
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 12.58,
|
| 1614 |
+
"learning_rate": 4.123711340206186e-06,
|
| 1615 |
+
"loss": 0.6215,
|
| 1616 |
+
"step": 2440
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 12.63,
|
| 1620 |
+
"learning_rate": 4.0950744558991985e-06,
|
| 1621 |
+
"loss": 0.644,
|
| 1622 |
+
"step": 2450
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 12.68,
|
| 1626 |
+
"learning_rate": 4.066437571592211e-06,
|
| 1627 |
+
"loss": 0.6071,
|
| 1628 |
+
"step": 2460
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 12.73,
|
| 1632 |
+
"learning_rate": 4.037800687285224e-06,
|
| 1633 |
+
"loss": 0.6035,
|
| 1634 |
+
"step": 2470
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 12.78,
|
| 1638 |
+
"learning_rate": 4.009163802978236e-06,
|
| 1639 |
+
"loss": 0.6482,
|
| 1640 |
+
"step": 2480
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 12.84,
|
| 1644 |
+
"learning_rate": 3.980526918671249e-06,
|
| 1645 |
+
"loss": 0.6152,
|
| 1646 |
+
"step": 2490
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 12.89,
|
| 1650 |
+
"learning_rate": 3.951890034364262e-06,
|
| 1651 |
+
"loss": 0.6212,
|
| 1652 |
+
"step": 2500
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 12.94,
|
| 1656 |
+
"learning_rate": 3.923253150057274e-06,
|
| 1657 |
+
"loss": 0.6033,
|
| 1658 |
+
"step": 2510
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 12.99,
|
| 1662 |
+
"learning_rate": 3.8946162657502865e-06,
|
| 1663 |
+
"loss": 0.6762,
|
| 1664 |
+
"step": 2520
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 13.0,
|
| 1668 |
+
"eval_accuracy": 0.7155756207674944,
|
| 1669 |
+
"eval_f1": 0.7152962310825538,
|
| 1670 |
+
"eval_loss": 0.8179510235786438,
|
| 1671 |
+
"eval_precision": 0.7161790866733511,
|
| 1672 |
+
"eval_recall": 0.7155756207674944,
|
| 1673 |
+
"eval_runtime": 138.6525,
|
| 1674 |
+
"eval_samples_per_second": 22.365,
|
| 1675 |
+
"eval_steps_per_second": 0.7,
|
| 1676 |
+
"step": 2522
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 13.04,
|
| 1680 |
+
"learning_rate": 3.865979381443299e-06,
|
| 1681 |
+
"loss": 0.5963,
|
| 1682 |
+
"step": 2530
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 13.09,
|
| 1686 |
+
"learning_rate": 3.837342497136312e-06,
|
| 1687 |
+
"loss": 0.6163,
|
| 1688 |
+
"step": 2540
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 13.14,
|
| 1692 |
+
"learning_rate": 3.808705612829324e-06,
|
| 1693 |
+
"loss": 0.6262,
|
| 1694 |
+
"step": 2550
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 13.2,
|
| 1698 |
+
"learning_rate": 3.780068728522337e-06,
|
| 1699 |
+
"loss": 0.6166,
|
| 1700 |
+
"step": 2560
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 13.25,
|
| 1704 |
+
"learning_rate": 3.7514318442153498e-06,
|
| 1705 |
+
"loss": 0.62,
|
| 1706 |
+
"step": 2570
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 13.3,
|
| 1710 |
+
"learning_rate": 3.7227949599083626e-06,
|
| 1711 |
+
"loss": 0.5676,
|
| 1712 |
+
"step": 2580
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 13.35,
|
| 1716 |
+
"learning_rate": 3.694158075601375e-06,
|
| 1717 |
+
"loss": 0.6093,
|
| 1718 |
+
"step": 2590
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 13.4,
|
| 1722 |
+
"learning_rate": 3.6655211912943874e-06,
|
| 1723 |
+
"loss": 0.5997,
|
| 1724 |
+
"step": 2600
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 13.45,
|
| 1728 |
+
"learning_rate": 3.6368843069873997e-06,
|
| 1729 |
+
"loss": 0.619,
|
| 1730 |
+
"step": 2610
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 13.51,
|
| 1734 |
+
"learning_rate": 3.6082474226804126e-06,
|
| 1735 |
+
"loss": 0.6019,
|
| 1736 |
+
"step": 2620
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 13.56,
|
| 1740 |
+
"learning_rate": 3.5796105383734254e-06,
|
| 1741 |
+
"loss": 0.6117,
|
| 1742 |
+
"step": 2630
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 13.61,
|
| 1746 |
+
"learning_rate": 3.550973654066438e-06,
|
| 1747 |
+
"loss": 0.5879,
|
| 1748 |
+
"step": 2640
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 13.66,
|
| 1752 |
+
"learning_rate": 3.5223367697594506e-06,
|
| 1753 |
+
"loss": 0.581,
|
| 1754 |
+
"step": 2650
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 13.71,
|
| 1758 |
+
"learning_rate": 3.493699885452463e-06,
|
| 1759 |
+
"loss": 0.633,
|
| 1760 |
+
"step": 2660
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 13.76,
|
| 1764 |
+
"learning_rate": 3.4650630011454754e-06,
|
| 1765 |
+
"loss": 0.6022,
|
| 1766 |
+
"step": 2670
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 13.81,
|
| 1770 |
+
"learning_rate": 3.436426116838488e-06,
|
| 1771 |
+
"loss": 0.602,
|
| 1772 |
+
"step": 2680
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 13.87,
|
| 1776 |
+
"learning_rate": 3.407789232531501e-06,
|
| 1777 |
+
"loss": 0.5738,
|
| 1778 |
+
"step": 2690
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 13.92,
|
| 1782 |
+
"learning_rate": 3.379152348224514e-06,
|
| 1783 |
+
"loss": 0.6074,
|
| 1784 |
+
"step": 2700
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 13.97,
|
| 1788 |
+
"learning_rate": 3.350515463917526e-06,
|
| 1789 |
+
"loss": 0.5937,
|
| 1790 |
+
"step": 2710
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 14.0,
|
| 1794 |
+
"eval_accuracy": 0.7188003869719445,
|
| 1795 |
+
"eval_f1": 0.718989971086903,
|
| 1796 |
+
"eval_loss": 0.8270788192749023,
|
| 1797 |
+
"eval_precision": 0.7219837313936599,
|
| 1798 |
+
"eval_recall": 0.7188003869719445,
|
| 1799 |
+
"eval_runtime": 137.3251,
|
| 1800 |
+
"eval_samples_per_second": 22.581,
|
| 1801 |
+
"eval_steps_per_second": 0.706,
|
| 1802 |
+
"step": 2716
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 14.02,
|
| 1806 |
+
"learning_rate": 3.3218785796105386e-06,
|
| 1807 |
+
"loss": 0.5785,
|
| 1808 |
+
"step": 2720
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 14.07,
|
| 1812 |
+
"learning_rate": 3.293241695303551e-06,
|
| 1813 |
+
"loss": 0.5945,
|
| 1814 |
+
"step": 2730
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 14.12,
|
| 1818 |
+
"learning_rate": 3.264604810996564e-06,
|
| 1819 |
+
"loss": 0.5927,
|
| 1820 |
+
"step": 2740
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 14.18,
|
| 1824 |
+
"learning_rate": 3.2359679266895766e-06,
|
| 1825 |
+
"loss": 0.6038,
|
| 1826 |
+
"step": 2750
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 14.23,
|
| 1830 |
+
"learning_rate": 3.2073310423825895e-06,
|
| 1831 |
+
"loss": 0.5904,
|
| 1832 |
+
"step": 2760
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 14.28,
|
| 1836 |
+
"learning_rate": 3.1786941580756014e-06,
|
| 1837 |
+
"loss": 0.5885,
|
| 1838 |
+
"step": 2770
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 14.33,
|
| 1842 |
+
"learning_rate": 3.1500572737686143e-06,
|
| 1843 |
+
"loss": 0.5421,
|
| 1844 |
+
"step": 2780
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 14.38,
|
| 1848 |
+
"learning_rate": 3.1214203894616266e-06,
|
| 1849 |
+
"loss": 0.5882,
|
| 1850 |
+
"step": 2790
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 14.43,
|
| 1854 |
+
"learning_rate": 3.0927835051546395e-06,
|
| 1855 |
+
"loss": 0.6144,
|
| 1856 |
+
"step": 2800
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 14.48,
|
| 1860 |
+
"learning_rate": 3.0641466208476523e-06,
|
| 1861 |
+
"loss": 0.6108,
|
| 1862 |
+
"step": 2810
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 14.54,
|
| 1866 |
+
"learning_rate": 3.0355097365406643e-06,
|
| 1867 |
+
"loss": 0.5852,
|
| 1868 |
+
"step": 2820
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 14.59,
|
| 1872 |
+
"learning_rate": 3.006872852233677e-06,
|
| 1873 |
+
"loss": 0.5534,
|
| 1874 |
+
"step": 2830
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 14.64,
|
| 1878 |
+
"learning_rate": 2.97823596792669e-06,
|
| 1879 |
+
"loss": 0.583,
|
| 1880 |
+
"step": 2840
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 14.69,
|
| 1884 |
+
"learning_rate": 2.9495990836197023e-06,
|
| 1885 |
+
"loss": 0.5909,
|
| 1886 |
+
"step": 2850
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 14.74,
|
| 1890 |
+
"learning_rate": 2.920962199312715e-06,
|
| 1891 |
+
"loss": 0.5972,
|
| 1892 |
+
"step": 2860
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 14.79,
|
| 1896 |
+
"learning_rate": 2.892325315005728e-06,
|
| 1897 |
+
"loss": 0.5697,
|
| 1898 |
+
"step": 2870
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 14.85,
|
| 1902 |
+
"learning_rate": 2.86368843069874e-06,
|
| 1903 |
+
"loss": 0.5914,
|
| 1904 |
+
"step": 2880
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 14.9,
|
| 1908 |
+
"learning_rate": 2.8350515463917527e-06,
|
| 1909 |
+
"loss": 0.6051,
|
| 1910 |
+
"step": 2890
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 14.95,
|
| 1914 |
+
"learning_rate": 2.8064146620847655e-06,
|
| 1915 |
+
"loss": 0.566,
|
| 1916 |
+
"step": 2900
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 15.0,
|
| 1920 |
+
"learning_rate": 2.7777777777777783e-06,
|
| 1921 |
+
"loss": 0.569,
|
| 1922 |
+
"step": 2910
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 15.0,
|
| 1926 |
+
"eval_accuracy": 0.7178329571106095,
|
| 1927 |
+
"eval_f1": 0.7165091349937845,
|
| 1928 |
+
"eval_loss": 0.8245362043380737,
|
| 1929 |
+
"eval_precision": 0.7174763240731499,
|
| 1930 |
+
"eval_recall": 0.7178329571106095,
|
| 1931 |
+
"eval_runtime": 137.9864,
|
| 1932 |
+
"eval_samples_per_second": 22.473,
|
| 1933 |
+
"eval_steps_per_second": 0.703,
|
| 1934 |
+
"step": 2910
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 15.05,
|
| 1938 |
+
"learning_rate": 2.7491408934707907e-06,
|
| 1939 |
+
"loss": 0.5732,
|
| 1940 |
+
"step": 2920
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 15.1,
|
| 1944 |
+
"learning_rate": 2.720504009163803e-06,
|
| 1945 |
+
"loss": 0.6155,
|
| 1946 |
+
"step": 2930
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 15.15,
|
| 1950 |
+
"learning_rate": 2.6918671248568155e-06,
|
| 1951 |
+
"loss": 0.5725,
|
| 1952 |
+
"step": 2940
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 15.21,
|
| 1956 |
+
"learning_rate": 2.6632302405498283e-06,
|
| 1957 |
+
"loss": 0.57,
|
| 1958 |
+
"step": 2950
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 15.26,
|
| 1962 |
+
"learning_rate": 2.634593356242841e-06,
|
| 1963 |
+
"loss": 0.5978,
|
| 1964 |
+
"step": 2960
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 15.31,
|
| 1968 |
+
"learning_rate": 2.605956471935854e-06,
|
| 1969 |
+
"loss": 0.5634,
|
| 1970 |
+
"step": 2970
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 15.36,
|
| 1974 |
+
"learning_rate": 2.577319587628866e-06,
|
| 1975 |
+
"loss": 0.5842,
|
| 1976 |
+
"step": 2980
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 15.41,
|
| 1980 |
+
"learning_rate": 2.5486827033218788e-06,
|
| 1981 |
+
"loss": 0.5699,
|
| 1982 |
+
"step": 2990
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 15.46,
|
| 1986 |
+
"learning_rate": 2.520045819014891e-06,
|
| 1987 |
+
"loss": 0.576,
|
| 1988 |
+
"step": 3000
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 15.52,
|
| 1992 |
+
"learning_rate": 2.491408934707904e-06,
|
| 1993 |
+
"loss": 0.5578,
|
| 1994 |
+
"step": 3010
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 15.57,
|
| 1998 |
+
"learning_rate": 2.4627720504009168e-06,
|
| 1999 |
+
"loss": 0.5726,
|
| 2000 |
+
"step": 3020
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 15.62,
|
| 2004 |
+
"learning_rate": 2.434135166093929e-06,
|
| 2005 |
+
"loss": 0.5669,
|
| 2006 |
+
"step": 3030
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 15.67,
|
| 2010 |
+
"learning_rate": 2.405498281786942e-06,
|
| 2011 |
+
"loss": 0.5945,
|
| 2012 |
+
"step": 3040
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 15.72,
|
| 2016 |
+
"learning_rate": 2.3768613974799544e-06,
|
| 2017 |
+
"loss": 0.5548,
|
| 2018 |
+
"step": 3050
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 15.77,
|
| 2022 |
+
"learning_rate": 2.3482245131729668e-06,
|
| 2023 |
+
"loss": 0.5442,
|
| 2024 |
+
"step": 3060
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 15.82,
|
| 2028 |
+
"learning_rate": 2.3195876288659796e-06,
|
| 2029 |
+
"loss": 0.5575,
|
| 2030 |
+
"step": 3070
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 15.88,
|
| 2034 |
+
"learning_rate": 2.290950744558992e-06,
|
| 2035 |
+
"loss": 0.5429,
|
| 2036 |
+
"step": 3080
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 15.93,
|
| 2040 |
+
"learning_rate": 2.262313860252005e-06,
|
| 2041 |
+
"loss": 0.5476,
|
| 2042 |
+
"step": 3090
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 15.98,
|
| 2046 |
+
"learning_rate": 2.2336769759450176e-06,
|
| 2047 |
+
"loss": 0.5623,
|
| 2048 |
+
"step": 3100
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 16.0,
|
| 2052 |
+
"eval_accuracy": 0.7165430506288294,
|
| 2053 |
+
"eval_f1": 0.7156652378724117,
|
| 2054 |
+
"eval_loss": 0.8228491544723511,
|
| 2055 |
+
"eval_precision": 0.7153021143244204,
|
| 2056 |
+
"eval_recall": 0.7165430506288294,
|
| 2057 |
+
"eval_runtime": 137.8917,
|
| 2058 |
+
"eval_samples_per_second": 22.489,
|
| 2059 |
+
"eval_steps_per_second": 0.703,
|
| 2060 |
+
"step": 3104
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 16.03,
|
| 2064 |
+
"learning_rate": 2.20504009163803e-06,
|
| 2065 |
+
"loss": 0.5951,
|
| 2066 |
+
"step": 3110
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 16.08,
|
| 2070 |
+
"learning_rate": 2.1764032073310424e-06,
|
| 2071 |
+
"loss": 0.5594,
|
| 2072 |
+
"step": 3120
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 16.13,
|
| 2076 |
+
"learning_rate": 2.1477663230240552e-06,
|
| 2077 |
+
"loss": 0.6058,
|
| 2078 |
+
"step": 3130
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 16.19,
|
| 2082 |
+
"learning_rate": 2.1191294387170676e-06,
|
| 2083 |
+
"loss": 0.5372,
|
| 2084 |
+
"step": 3140
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 16.24,
|
| 2088 |
+
"learning_rate": 2.0904925544100804e-06,
|
| 2089 |
+
"loss": 0.566,
|
| 2090 |
+
"step": 3150
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 16.29,
|
| 2094 |
+
"learning_rate": 2.061855670103093e-06,
|
| 2095 |
+
"loss": 0.5524,
|
| 2096 |
+
"step": 3160
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 16.34,
|
| 2100 |
+
"learning_rate": 2.0332187857961057e-06,
|
| 2101 |
+
"loss": 0.527,
|
| 2102 |
+
"step": 3170
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 16.39,
|
| 2106 |
+
"learning_rate": 2.004581901489118e-06,
|
| 2107 |
+
"loss": 0.5995,
|
| 2108 |
+
"step": 3180
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 16.44,
|
| 2112 |
+
"learning_rate": 1.975945017182131e-06,
|
| 2113 |
+
"loss": 0.5271,
|
| 2114 |
+
"step": 3190
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 16.49,
|
| 2118 |
+
"learning_rate": 1.9473081328751433e-06,
|
| 2119 |
+
"loss": 0.5892,
|
| 2120 |
+
"step": 3200
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 16.55,
|
| 2124 |
+
"learning_rate": 1.918671248568156e-06,
|
| 2125 |
+
"loss": 0.5539,
|
| 2126 |
+
"step": 3210
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 16.6,
|
| 2130 |
+
"learning_rate": 1.8900343642611685e-06,
|
| 2131 |
+
"loss": 0.551,
|
| 2132 |
+
"step": 3220
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 16.65,
|
| 2136 |
+
"learning_rate": 1.8613974799541813e-06,
|
| 2137 |
+
"loss": 0.5638,
|
| 2138 |
+
"step": 3230
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 16.7,
|
| 2142 |
+
"learning_rate": 1.8327605956471937e-06,
|
| 2143 |
+
"loss": 0.5533,
|
| 2144 |
+
"step": 3240
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 16.75,
|
| 2148 |
+
"learning_rate": 1.8041237113402063e-06,
|
| 2149 |
+
"loss": 0.5886,
|
| 2150 |
+
"step": 3250
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 16.8,
|
| 2154 |
+
"learning_rate": 1.775486827033219e-06,
|
| 2155 |
+
"loss": 0.5325,
|
| 2156 |
+
"step": 3260
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 16.86,
|
| 2160 |
+
"learning_rate": 1.7468499427262315e-06,
|
| 2161 |
+
"loss": 0.5441,
|
| 2162 |
+
"step": 3270
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 16.91,
|
| 2166 |
+
"learning_rate": 1.718213058419244e-06,
|
| 2167 |
+
"loss": 0.5422,
|
| 2168 |
+
"step": 3280
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 16.96,
|
| 2172 |
+
"learning_rate": 1.689576174112257e-06,
|
| 2173 |
+
"loss": 0.5291,
|
| 2174 |
+
"step": 3290
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 17.0,
|
| 2178 |
+
"eval_accuracy": 0.7162205740083843,
|
| 2179 |
+
"eval_f1": 0.7155771742857344,
|
| 2180 |
+
"eval_loss": 0.8238459825515747,
|
| 2181 |
+
"eval_precision": 0.7165407692005947,
|
| 2182 |
+
"eval_recall": 0.7162205740083843,
|
| 2183 |
+
"eval_runtime": 138.7789,
|
| 2184 |
+
"eval_samples_per_second": 22.345,
|
| 2185 |
+
"eval_steps_per_second": 0.699,
|
| 2186 |
+
"step": 3298
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 17.01,
|
| 2190 |
+
"learning_rate": 1.6609392898052693e-06,
|
| 2191 |
+
"loss": 0.544,
|
| 2192 |
+
"step": 3300
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 17.06,
|
| 2196 |
+
"learning_rate": 1.632302405498282e-06,
|
| 2197 |
+
"loss": 0.5365,
|
| 2198 |
+
"step": 3310
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 17.11,
|
| 2202 |
+
"learning_rate": 1.6036655211912947e-06,
|
| 2203 |
+
"loss": 0.5507,
|
| 2204 |
+
"step": 3320
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 17.16,
|
| 2208 |
+
"learning_rate": 1.5750286368843071e-06,
|
| 2209 |
+
"loss": 0.5537,
|
| 2210 |
+
"step": 3330
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 17.22,
|
| 2214 |
+
"learning_rate": 1.5463917525773197e-06,
|
| 2215 |
+
"loss": 0.5241,
|
| 2216 |
+
"step": 3340
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 17.27,
|
| 2220 |
+
"learning_rate": 1.5177548682703321e-06,
|
| 2221 |
+
"loss": 0.5106,
|
| 2222 |
+
"step": 3350
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 17.32,
|
| 2226 |
+
"learning_rate": 1.489117983963345e-06,
|
| 2227 |
+
"loss": 0.5773,
|
| 2228 |
+
"step": 3360
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 17.37,
|
| 2232 |
+
"learning_rate": 1.4604810996563575e-06,
|
| 2233 |
+
"loss": 0.5466,
|
| 2234 |
+
"step": 3370
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 17.42,
|
| 2238 |
+
"learning_rate": 1.43184421534937e-06,
|
| 2239 |
+
"loss": 0.538,
|
| 2240 |
+
"step": 3380
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 17.47,
|
| 2244 |
+
"learning_rate": 1.4032073310423828e-06,
|
| 2245 |
+
"loss": 0.5373,
|
| 2246 |
+
"step": 3390
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 17.53,
|
| 2250 |
+
"learning_rate": 1.3745704467353954e-06,
|
| 2251 |
+
"loss": 0.5416,
|
| 2252 |
+
"step": 3400
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 17.58,
|
| 2256 |
+
"learning_rate": 1.3459335624284078e-06,
|
| 2257 |
+
"loss": 0.5536,
|
| 2258 |
+
"step": 3410
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 17.63,
|
| 2262 |
+
"learning_rate": 1.3172966781214206e-06,
|
| 2263 |
+
"loss": 0.5853,
|
| 2264 |
+
"step": 3420
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 17.68,
|
| 2268 |
+
"learning_rate": 1.288659793814433e-06,
|
| 2269 |
+
"loss": 0.5522,
|
| 2270 |
+
"step": 3430
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 17.73,
|
| 2274 |
+
"learning_rate": 1.2600229095074456e-06,
|
| 2275 |
+
"loss": 0.5318,
|
| 2276 |
+
"step": 3440
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 17.78,
|
| 2280 |
+
"learning_rate": 1.2313860252004584e-06,
|
| 2281 |
+
"loss": 0.553,
|
| 2282 |
+
"step": 3450
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 17.84,
|
| 2286 |
+
"learning_rate": 1.202749140893471e-06,
|
| 2287 |
+
"loss": 0.5386,
|
| 2288 |
+
"step": 3460
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 17.89,
|
| 2292 |
+
"learning_rate": 1.1741122565864834e-06,
|
| 2293 |
+
"loss": 0.5333,
|
| 2294 |
+
"step": 3470
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 17.94,
|
| 2298 |
+
"learning_rate": 1.145475372279496e-06,
|
| 2299 |
+
"loss": 0.568,
|
| 2300 |
+
"step": 3480
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 17.99,
|
| 2304 |
+
"learning_rate": 1.1168384879725088e-06,
|
| 2305 |
+
"loss": 0.5775,
|
| 2306 |
+
"step": 3490
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 18.0,
|
| 2310 |
+
"eval_accuracy": 0.7152531441470493,
|
| 2311 |
+
"eval_f1": 0.7151086078956284,
|
| 2312 |
+
"eval_loss": 0.8245630860328674,
|
| 2313 |
+
"eval_precision": 0.7161566827277972,
|
| 2314 |
+
"eval_recall": 0.7152531441470493,
|
| 2315 |
+
"eval_runtime": 137.143,
|
| 2316 |
+
"eval_samples_per_second": 22.611,
|
| 2317 |
+
"eval_steps_per_second": 0.707,
|
| 2318 |
+
"step": 3492
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 18.04,
|
| 2322 |
+
"learning_rate": 1.0882016036655212e-06,
|
| 2323 |
+
"loss": 0.5582,
|
| 2324 |
+
"step": 3500
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 18.09,
|
| 2328 |
+
"learning_rate": 1.0595647193585338e-06,
|
| 2329 |
+
"loss": 0.5404,
|
| 2330 |
+
"step": 3510
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 18.14,
|
| 2334 |
+
"learning_rate": 1.0309278350515464e-06,
|
| 2335 |
+
"loss": 0.57,
|
| 2336 |
+
"step": 3520
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 18.2,
|
| 2340 |
+
"learning_rate": 1.002290950744559e-06,
|
| 2341 |
+
"loss": 0.5514,
|
| 2342 |
+
"step": 3530
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 18.25,
|
| 2346 |
+
"learning_rate": 9.736540664375716e-07,
|
| 2347 |
+
"loss": 0.48,
|
| 2348 |
+
"step": 3540
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 18.3,
|
| 2352 |
+
"learning_rate": 9.450171821305842e-07,
|
| 2353 |
+
"loss": 0.4917,
|
| 2354 |
+
"step": 3550
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 18.35,
|
| 2358 |
+
"learning_rate": 9.163802978235968e-07,
|
| 2359 |
+
"loss": 0.5253,
|
| 2360 |
+
"step": 3560
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 18.4,
|
| 2364 |
+
"learning_rate": 8.877434135166096e-07,
|
| 2365 |
+
"loss": 0.5138,
|
| 2366 |
+
"step": 3570
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 18.45,
|
| 2370 |
+
"learning_rate": 8.59106529209622e-07,
|
| 2371 |
+
"loss": 0.5686,
|
| 2372 |
+
"step": 3580
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 18.51,
|
| 2376 |
+
"learning_rate": 8.304696449026347e-07,
|
| 2377 |
+
"loss": 0.5669,
|
| 2378 |
+
"step": 3590
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 18.56,
|
| 2382 |
+
"learning_rate": 8.018327605956474e-07,
|
| 2383 |
+
"loss": 0.5286,
|
| 2384 |
+
"step": 3600
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 18.61,
|
| 2388 |
+
"learning_rate": 7.731958762886599e-07,
|
| 2389 |
+
"loss": 0.5736,
|
| 2390 |
+
"step": 3610
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 18.66,
|
| 2394 |
+
"learning_rate": 7.445589919816725e-07,
|
| 2395 |
+
"loss": 0.5202,
|
| 2396 |
+
"step": 3620
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 18.71,
|
| 2400 |
+
"learning_rate": 7.15922107674685e-07,
|
| 2401 |
+
"loss": 0.5417,
|
| 2402 |
+
"step": 3630
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 18.76,
|
| 2406 |
+
"learning_rate": 6.872852233676977e-07,
|
| 2407 |
+
"loss": 0.5409,
|
| 2408 |
+
"step": 3640
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 18.81,
|
| 2412 |
+
"learning_rate": 6.586483390607103e-07,
|
| 2413 |
+
"loss": 0.5737,
|
| 2414 |
+
"step": 3650
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 18.87,
|
| 2418 |
+
"learning_rate": 6.300114547537228e-07,
|
| 2419 |
+
"loss": 0.5383,
|
| 2420 |
+
"step": 3660
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 18.92,
|
| 2424 |
+
"learning_rate": 6.013745704467355e-07,
|
| 2425 |
+
"loss": 0.5322,
|
| 2426 |
+
"step": 3670
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 18.97,
|
| 2430 |
+
"learning_rate": 5.72737686139748e-07,
|
| 2431 |
+
"loss": 0.545,
|
| 2432 |
+
"step": 3680
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 19.0,
|
| 2436 |
+
"eval_accuracy": 0.7178329571106095,
|
| 2437 |
+
"eval_f1": 0.7173917212076267,
|
| 2438 |
+
"eval_loss": 0.825716495513916,
|
| 2439 |
+
"eval_precision": 0.7191676309063934,
|
| 2440 |
+
"eval_recall": 0.7178329571106095,
|
| 2441 |
+
"eval_runtime": 137.8289,
|
| 2442 |
+
"eval_samples_per_second": 22.499,
|
| 2443 |
+
"eval_steps_per_second": 0.704,
|
| 2444 |
+
"step": 3686
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 19.02,
|
| 2448 |
+
"learning_rate": 5.441008018327606e-07,
|
| 2449 |
+
"loss": 0.567,
|
| 2450 |
+
"step": 3690
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 19.07,
|
| 2454 |
+
"learning_rate": 5.154639175257732e-07,
|
| 2455 |
+
"loss": 0.5309,
|
| 2456 |
+
"step": 3700
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 19.12,
|
| 2460 |
+
"learning_rate": 4.868270332187858e-07,
|
| 2461 |
+
"loss": 0.548,
|
| 2462 |
+
"step": 3710
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 19.18,
|
| 2466 |
+
"learning_rate": 4.581901489117984e-07,
|
| 2467 |
+
"loss": 0.5147,
|
| 2468 |
+
"step": 3720
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 19.23,
|
| 2472 |
+
"learning_rate": 4.29553264604811e-07,
|
| 2473 |
+
"loss": 0.4884,
|
| 2474 |
+
"step": 3730
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 19.28,
|
| 2478 |
+
"learning_rate": 4.009163802978237e-07,
|
| 2479 |
+
"loss": 0.5347,
|
| 2480 |
+
"step": 3740
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 19.33,
|
| 2484 |
+
"learning_rate": 3.7227949599083624e-07,
|
| 2485 |
+
"loss": 0.5193,
|
| 2486 |
+
"step": 3750
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 19.38,
|
| 2490 |
+
"learning_rate": 3.4364261168384884e-07,
|
| 2491 |
+
"loss": 0.5337,
|
| 2492 |
+
"step": 3760
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 19.43,
|
| 2496 |
+
"learning_rate": 3.150057273768614e-07,
|
| 2497 |
+
"loss": 0.5594,
|
| 2498 |
+
"step": 3770
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 19.48,
|
| 2502 |
+
"learning_rate": 2.86368843069874e-07,
|
| 2503 |
+
"loss": 0.531,
|
| 2504 |
+
"step": 3780
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 19.54,
|
| 2508 |
+
"learning_rate": 2.577319587628866e-07,
|
| 2509 |
+
"loss": 0.5651,
|
| 2510 |
+
"step": 3790
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 19.59,
|
| 2514 |
+
"learning_rate": 2.290950744558992e-07,
|
| 2515 |
+
"loss": 0.4949,
|
| 2516 |
+
"step": 3800
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 19.64,
|
| 2520 |
+
"learning_rate": 2.0045819014891184e-07,
|
| 2521 |
+
"loss": 0.5284,
|
| 2522 |
+
"step": 3810
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 19.69,
|
| 2526 |
+
"learning_rate": 1.7182130584192442e-07,
|
| 2527 |
+
"loss": 0.5183,
|
| 2528 |
+
"step": 3820
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 19.74,
|
| 2532 |
+
"learning_rate": 1.43184421534937e-07,
|
| 2533 |
+
"loss": 0.5412,
|
| 2534 |
+
"step": 3830
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 19.79,
|
| 2538 |
+
"learning_rate": 1.145475372279496e-07,
|
| 2539 |
+
"loss": 0.547,
|
| 2540 |
+
"step": 3840
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 19.85,
|
| 2544 |
+
"learning_rate": 8.591065292096221e-08,
|
| 2545 |
+
"loss": 0.5675,
|
| 2546 |
+
"step": 3850
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 19.9,
|
| 2550 |
+
"learning_rate": 5.72737686139748e-08,
|
| 2551 |
+
"loss": 0.5541,
|
| 2552 |
+
"step": 3860
|
| 2553 |
+
},
|
| 2554 |
+
{
|
| 2555 |
+
"epoch": 19.95,
|
| 2556 |
+
"learning_rate": 2.86368843069874e-08,
|
| 2557 |
+
"loss": 0.5541,
|
| 2558 |
+
"step": 3870
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 20.0,
|
| 2562 |
+
"learning_rate": 0.0,
|
| 2563 |
+
"loss": 0.5409,
|
| 2564 |
+
"step": 3880
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 20.0,
|
| 2568 |
+
"eval_accuracy": 0.7178329571106095,
|
| 2569 |
+
"eval_f1": 0.717743945710896,
|
| 2570 |
+
"eval_loss": 0.8245128989219666,
|
| 2571 |
+
"eval_precision": 0.7187025517730355,
|
| 2572 |
+
"eval_recall": 0.7178329571106095,
|
| 2573 |
+
"eval_runtime": 138.2737,
|
| 2574 |
+
"eval_samples_per_second": 22.427,
|
| 2575 |
+
"eval_steps_per_second": 0.702,
|
| 2576 |
+
"step": 3880
|
| 2577 |
+
},
|
| 2578 |
+
{
|
| 2579 |
+
"epoch": 20.0,
|
| 2580 |
+
"step": 3880,
|
| 2581 |
+
"total_flos": 3.843964371168166e+19,
|
| 2582 |
+
"train_loss": 0.8177650722031741,
|
| 2583 |
+
"train_runtime": 43638.4535,
|
| 2584 |
+
"train_samples_per_second": 11.367,
|
| 2585 |
+
"train_steps_per_second": 0.089
|
| 2586 |
+
}
|
| 2587 |
+
],
|
| 2588 |
+
"max_steps": 3880,
|
| 2589 |
+
"num_train_epochs": 20,
|
| 2590 |
+
"total_flos": 3.843964371168166e+19,
|
| 2591 |
+
"trial_name": null,
|
| 2592 |
+
"trial_params": null
|
| 2593 |
+
}
|