Commit
•
664d3c8
1
Parent(s):
78bbe81
End of training
Browse files- all_results.json +15 -0
- eval_results.json +10 -0
- runs/Aug13_07-03-56_22b062983b72/events.out.tfevents.1660407918.22b062983b72.1713.2 +3 -0
- train_results.json +8 -0
- trainer_state.json +1885 -0
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.8799521354933726,
|
| 4 |
+
"eval_loss": 0.5234681367874146,
|
| 5 |
+
"eval_runtime": 758.8099,
|
| 6 |
+
"eval_samples": 15520,
|
| 7 |
+
"eval_samples_per_second": 20.453,
|
| 8 |
+
"eval_steps_per_second": 20.453,
|
| 9 |
+
"perplexity": 1.687871279063992,
|
| 10 |
+
"train_loss": 1.3494602605485424,
|
| 11 |
+
"train_runtime": 31681.5151,
|
| 12 |
+
"train_samples": 15520,
|
| 13 |
+
"train_samples_per_second": 4.899,
|
| 14 |
+
"train_steps_per_second": 4.899
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.8799521354933726,
|
| 4 |
+
"eval_loss": 0.5234681367874146,
|
| 5 |
+
"eval_runtime": 758.8099,
|
| 6 |
+
"eval_samples": 15520,
|
| 7 |
+
"eval_samples_per_second": 20.453,
|
| 8 |
+
"eval_steps_per_second": 20.453,
|
| 9 |
+
"perplexity": 1.687871279063992
|
| 10 |
+
}
|
runs/Aug13_07-03-56_22b062983b72/events.out.tfevents.1660407918.22b062983b72.1713.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:372d6e0470de2ff64da107044af5d0bf155b3681b2026570422b3633bdb00e47
|
| 3 |
+
size 369
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"train_loss": 1.3494602605485424,
|
| 4 |
+
"train_runtime": 31681.5151,
|
| 5 |
+
"train_samples": 15520,
|
| 6 |
+
"train_samples_per_second": 4.899,
|
| 7 |
+
"train_steps_per_second": 4.899
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1885 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"global_step": 155200,
|
| 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": 4.98389175257732e-05,
|
| 13 |
+
"loss": 2.2272,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.06,
|
| 18 |
+
"learning_rate": 4.9677835051546396e-05,
|
| 19 |
+
"loss": 2.1881,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.1,
|
| 24 |
+
"learning_rate": 4.951675257731959e-05,
|
| 25 |
+
"loss": 2.1808,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.13,
|
| 30 |
+
"learning_rate": 4.935567010309279e-05,
|
| 31 |
+
"loss": 2.1891,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.16,
|
| 36 |
+
"learning_rate": 4.919458762886598e-05,
|
| 37 |
+
"loss": 2.2239,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.19,
|
| 42 |
+
"learning_rate": 4.903350515463918e-05,
|
| 43 |
+
"loss": 2.1983,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.23,
|
| 48 |
+
"learning_rate": 4.8872422680412375e-05,
|
| 49 |
+
"loss": 2.2063,
|
| 50 |
+
"step": 3500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.26,
|
| 54 |
+
"learning_rate": 4.871134020618557e-05,
|
| 55 |
+
"loss": 2.2125,
|
| 56 |
+
"step": 4000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.29,
|
| 60 |
+
"learning_rate": 4.855025773195876e-05,
|
| 61 |
+
"loss": 2.1918,
|
| 62 |
+
"step": 4500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.32,
|
| 66 |
+
"learning_rate": 4.838917525773196e-05,
|
| 67 |
+
"loss": 2.2064,
|
| 68 |
+
"step": 5000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.35,
|
| 72 |
+
"learning_rate": 4.8228092783505156e-05,
|
| 73 |
+
"loss": 2.1789,
|
| 74 |
+
"step": 5500
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.39,
|
| 78 |
+
"learning_rate": 4.806701030927835e-05,
|
| 79 |
+
"loss": 2.2123,
|
| 80 |
+
"step": 6000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.42,
|
| 84 |
+
"learning_rate": 4.790592783505155e-05,
|
| 85 |
+
"loss": 2.1775,
|
| 86 |
+
"step": 6500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.45,
|
| 90 |
+
"learning_rate": 4.774484536082475e-05,
|
| 91 |
+
"loss": 2.1837,
|
| 92 |
+
"step": 7000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.48,
|
| 96 |
+
"learning_rate": 4.758376288659794e-05,
|
| 97 |
+
"loss": 2.1812,
|
| 98 |
+
"step": 7500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.52,
|
| 102 |
+
"learning_rate": 4.7422680412371134e-05,
|
| 103 |
+
"loss": 2.1857,
|
| 104 |
+
"step": 8000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.55,
|
| 108 |
+
"learning_rate": 4.726159793814433e-05,
|
| 109 |
+
"loss": 2.1788,
|
| 110 |
+
"step": 8500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.58,
|
| 114 |
+
"learning_rate": 4.710051546391753e-05,
|
| 115 |
+
"loss": 2.2204,
|
| 116 |
+
"step": 9000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.61,
|
| 120 |
+
"learning_rate": 4.6939432989690725e-05,
|
| 121 |
+
"loss": 2.1704,
|
| 122 |
+
"step": 9500
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.64,
|
| 126 |
+
"learning_rate": 4.677835051546392e-05,
|
| 127 |
+
"loss": 2.2155,
|
| 128 |
+
"step": 10000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.68,
|
| 132 |
+
"learning_rate": 4.661726804123712e-05,
|
| 133 |
+
"loss": 2.2153,
|
| 134 |
+
"step": 10500
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.71,
|
| 138 |
+
"learning_rate": 4.6456185567010316e-05,
|
| 139 |
+
"loss": 2.2051,
|
| 140 |
+
"step": 11000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.74,
|
| 144 |
+
"learning_rate": 4.6295103092783506e-05,
|
| 145 |
+
"loss": 2.2273,
|
| 146 |
+
"step": 11500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.77,
|
| 150 |
+
"learning_rate": 4.61340206185567e-05,
|
| 151 |
+
"loss": 2.1818,
|
| 152 |
+
"step": 12000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.81,
|
| 156 |
+
"learning_rate": 4.59729381443299e-05,
|
| 157 |
+
"loss": 2.1954,
|
| 158 |
+
"step": 12500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.84,
|
| 162 |
+
"learning_rate": 4.581185567010309e-05,
|
| 163 |
+
"loss": 2.1635,
|
| 164 |
+
"step": 13000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.87,
|
| 168 |
+
"learning_rate": 4.565077319587629e-05,
|
| 169 |
+
"loss": 2.1799,
|
| 170 |
+
"step": 13500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.9,
|
| 174 |
+
"learning_rate": 4.5489690721649484e-05,
|
| 175 |
+
"loss": 2.1742,
|
| 176 |
+
"step": 14000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.93,
|
| 180 |
+
"learning_rate": 4.532860824742268e-05,
|
| 181 |
+
"loss": 2.1402,
|
| 182 |
+
"step": 14500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.97,
|
| 186 |
+
"learning_rate": 4.516752577319588e-05,
|
| 187 |
+
"loss": 2.1322,
|
| 188 |
+
"step": 15000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 1.0,
|
| 192 |
+
"learning_rate": 4.5006443298969075e-05,
|
| 193 |
+
"loss": 2.1972,
|
| 194 |
+
"step": 15500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 1.03,
|
| 198 |
+
"learning_rate": 4.484536082474227e-05,
|
| 199 |
+
"loss": 1.8657,
|
| 200 |
+
"step": 16000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.06,
|
| 204 |
+
"learning_rate": 4.468427835051547e-05,
|
| 205 |
+
"loss": 1.8975,
|
| 206 |
+
"step": 16500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.1,
|
| 210 |
+
"learning_rate": 4.452319587628866e-05,
|
| 211 |
+
"loss": 1.8662,
|
| 212 |
+
"step": 17000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.13,
|
| 216 |
+
"learning_rate": 4.4362113402061856e-05,
|
| 217 |
+
"loss": 1.8394,
|
| 218 |
+
"step": 17500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 1.16,
|
| 222 |
+
"learning_rate": 4.4201030927835053e-05,
|
| 223 |
+
"loss": 1.8698,
|
| 224 |
+
"step": 18000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 1.19,
|
| 228 |
+
"learning_rate": 4.403994845360825e-05,
|
| 229 |
+
"loss": 1.8718,
|
| 230 |
+
"step": 18500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 1.22,
|
| 234 |
+
"learning_rate": 4.387886597938145e-05,
|
| 235 |
+
"loss": 1.8674,
|
| 236 |
+
"step": 19000
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 1.26,
|
| 240 |
+
"learning_rate": 4.3717783505154644e-05,
|
| 241 |
+
"loss": 1.9007,
|
| 242 |
+
"step": 19500
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.29,
|
| 246 |
+
"learning_rate": 4.3556701030927835e-05,
|
| 247 |
+
"loss": 1.8961,
|
| 248 |
+
"step": 20000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 1.32,
|
| 252 |
+
"learning_rate": 4.339561855670103e-05,
|
| 253 |
+
"loss": 1.8516,
|
| 254 |
+
"step": 20500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.35,
|
| 258 |
+
"learning_rate": 4.323453608247423e-05,
|
| 259 |
+
"loss": 1.9022,
|
| 260 |
+
"step": 21000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 1.39,
|
| 264 |
+
"learning_rate": 4.3073453608247426e-05,
|
| 265 |
+
"loss": 1.8913,
|
| 266 |
+
"step": 21500
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 1.42,
|
| 270 |
+
"learning_rate": 4.2912371134020616e-05,
|
| 271 |
+
"loss": 1.9092,
|
| 272 |
+
"step": 22000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 1.45,
|
| 276 |
+
"learning_rate": 4.275128865979381e-05,
|
| 277 |
+
"loss": 1.8871,
|
| 278 |
+
"step": 22500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 1.48,
|
| 282 |
+
"learning_rate": 4.259020618556701e-05,
|
| 283 |
+
"loss": 1.9027,
|
| 284 |
+
"step": 23000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 1.51,
|
| 288 |
+
"learning_rate": 4.242912371134021e-05,
|
| 289 |
+
"loss": 1.9055,
|
| 290 |
+
"step": 23500
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.55,
|
| 294 |
+
"learning_rate": 4.2268041237113404e-05,
|
| 295 |
+
"loss": 1.8874,
|
| 296 |
+
"step": 24000
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 1.58,
|
| 300 |
+
"learning_rate": 4.21069587628866e-05,
|
| 301 |
+
"loss": 1.9049,
|
| 302 |
+
"step": 24500
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 1.61,
|
| 306 |
+
"learning_rate": 4.19458762886598e-05,
|
| 307 |
+
"loss": 1.9106,
|
| 308 |
+
"step": 25000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 1.64,
|
| 312 |
+
"learning_rate": 4.1784793814432995e-05,
|
| 313 |
+
"loss": 1.9016,
|
| 314 |
+
"step": 25500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 1.68,
|
| 318 |
+
"learning_rate": 4.162371134020619e-05,
|
| 319 |
+
"loss": 1.9368,
|
| 320 |
+
"step": 26000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 1.71,
|
| 324 |
+
"learning_rate": 4.146262886597939e-05,
|
| 325 |
+
"loss": 1.9027,
|
| 326 |
+
"step": 26500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 1.74,
|
| 330 |
+
"learning_rate": 4.130154639175258e-05,
|
| 331 |
+
"loss": 1.9143,
|
| 332 |
+
"step": 27000
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 1.77,
|
| 336 |
+
"learning_rate": 4.1140463917525776e-05,
|
| 337 |
+
"loss": 1.9305,
|
| 338 |
+
"step": 27500
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 1.8,
|
| 342 |
+
"learning_rate": 4.097938144329897e-05,
|
| 343 |
+
"loss": 1.9137,
|
| 344 |
+
"step": 28000
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 1.84,
|
| 348 |
+
"learning_rate": 4.081829896907216e-05,
|
| 349 |
+
"loss": 1.9027,
|
| 350 |
+
"step": 28500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 1.87,
|
| 354 |
+
"learning_rate": 4.065721649484536e-05,
|
| 355 |
+
"loss": 1.9199,
|
| 356 |
+
"step": 29000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 1.9,
|
| 360 |
+
"learning_rate": 4.049613402061856e-05,
|
| 361 |
+
"loss": 1.935,
|
| 362 |
+
"step": 29500
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 1.93,
|
| 366 |
+
"learning_rate": 4.0335051546391754e-05,
|
| 367 |
+
"loss": 1.9394,
|
| 368 |
+
"step": 30000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 1.97,
|
| 372 |
+
"learning_rate": 4.017396907216495e-05,
|
| 373 |
+
"loss": 1.9376,
|
| 374 |
+
"step": 30500
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 2.0,
|
| 378 |
+
"learning_rate": 4.001288659793815e-05,
|
| 379 |
+
"loss": 1.8943,
|
| 380 |
+
"step": 31000
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 2.03,
|
| 384 |
+
"learning_rate": 3.9851804123711345e-05,
|
| 385 |
+
"loss": 1.6319,
|
| 386 |
+
"step": 31500
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 2.06,
|
| 390 |
+
"learning_rate": 3.9690721649484535e-05,
|
| 391 |
+
"loss": 1.5991,
|
| 392 |
+
"step": 32000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 2.09,
|
| 396 |
+
"learning_rate": 3.952963917525773e-05,
|
| 397 |
+
"loss": 1.6505,
|
| 398 |
+
"step": 32500
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.13,
|
| 402 |
+
"learning_rate": 3.936855670103093e-05,
|
| 403 |
+
"loss": 1.6288,
|
| 404 |
+
"step": 33000
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 2.16,
|
| 408 |
+
"learning_rate": 3.9207474226804126e-05,
|
| 409 |
+
"loss": 1.6686,
|
| 410 |
+
"step": 33500
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 2.19,
|
| 414 |
+
"learning_rate": 3.904639175257732e-05,
|
| 415 |
+
"loss": 1.6538,
|
| 416 |
+
"step": 34000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 2.22,
|
| 420 |
+
"learning_rate": 3.888530927835052e-05,
|
| 421 |
+
"loss": 1.6228,
|
| 422 |
+
"step": 34500
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 2.26,
|
| 426 |
+
"learning_rate": 3.872422680412372e-05,
|
| 427 |
+
"loss": 1.6465,
|
| 428 |
+
"step": 35000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 2.29,
|
| 432 |
+
"learning_rate": 3.8563144329896914e-05,
|
| 433 |
+
"loss": 1.67,
|
| 434 |
+
"step": 35500
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 2.32,
|
| 438 |
+
"learning_rate": 3.8402061855670104e-05,
|
| 439 |
+
"loss": 1.6598,
|
| 440 |
+
"step": 36000
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 2.35,
|
| 444 |
+
"learning_rate": 3.82409793814433e-05,
|
| 445 |
+
"loss": 1.6909,
|
| 446 |
+
"step": 36500
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 2.38,
|
| 450 |
+
"learning_rate": 3.807989690721649e-05,
|
| 451 |
+
"loss": 1.6686,
|
| 452 |
+
"step": 37000
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 2.42,
|
| 456 |
+
"learning_rate": 3.791881443298969e-05,
|
| 457 |
+
"loss": 1.6587,
|
| 458 |
+
"step": 37500
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 2.45,
|
| 462 |
+
"learning_rate": 3.7757731958762886e-05,
|
| 463 |
+
"loss": 1.6826,
|
| 464 |
+
"step": 38000
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 2.48,
|
| 468 |
+
"learning_rate": 3.759664948453608e-05,
|
| 469 |
+
"loss": 1.6711,
|
| 470 |
+
"step": 38500
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 2.51,
|
| 474 |
+
"learning_rate": 3.743556701030928e-05,
|
| 475 |
+
"loss": 1.6633,
|
| 476 |
+
"step": 39000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 2.55,
|
| 480 |
+
"learning_rate": 3.7274484536082477e-05,
|
| 481 |
+
"loss": 1.6806,
|
| 482 |
+
"step": 39500
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 2.58,
|
| 486 |
+
"learning_rate": 3.7113402061855674e-05,
|
| 487 |
+
"loss": 1.6888,
|
| 488 |
+
"step": 40000
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 2.61,
|
| 492 |
+
"learning_rate": 3.695231958762887e-05,
|
| 493 |
+
"loss": 1.6788,
|
| 494 |
+
"step": 40500
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 2.64,
|
| 498 |
+
"learning_rate": 3.679123711340207e-05,
|
| 499 |
+
"loss": 1.6988,
|
| 500 |
+
"step": 41000
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 2.67,
|
| 504 |
+
"learning_rate": 3.663015463917526e-05,
|
| 505 |
+
"loss": 1.6428,
|
| 506 |
+
"step": 41500
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 2.71,
|
| 510 |
+
"learning_rate": 3.6469072164948455e-05,
|
| 511 |
+
"loss": 1.6797,
|
| 512 |
+
"step": 42000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 2.74,
|
| 516 |
+
"learning_rate": 3.630798969072165e-05,
|
| 517 |
+
"loss": 1.6623,
|
| 518 |
+
"step": 42500
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 2.77,
|
| 522 |
+
"learning_rate": 3.614690721649485e-05,
|
| 523 |
+
"loss": 1.6899,
|
| 524 |
+
"step": 43000
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 2.8,
|
| 528 |
+
"learning_rate": 3.5985824742268046e-05,
|
| 529 |
+
"loss": 1.706,
|
| 530 |
+
"step": 43500
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 2.84,
|
| 534 |
+
"learning_rate": 3.5824742268041236e-05,
|
| 535 |
+
"loss": 1.6955,
|
| 536 |
+
"step": 44000
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 2.87,
|
| 540 |
+
"learning_rate": 3.566365979381443e-05,
|
| 541 |
+
"loss": 1.7074,
|
| 542 |
+
"step": 44500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 2.9,
|
| 546 |
+
"learning_rate": 3.550257731958763e-05,
|
| 547 |
+
"loss": 1.7156,
|
| 548 |
+
"step": 45000
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 2.93,
|
| 552 |
+
"learning_rate": 3.534149484536083e-05,
|
| 553 |
+
"loss": 1.6655,
|
| 554 |
+
"step": 45500
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 2.96,
|
| 558 |
+
"learning_rate": 3.5180412371134024e-05,
|
| 559 |
+
"loss": 1.6848,
|
| 560 |
+
"step": 46000
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 3.0,
|
| 564 |
+
"learning_rate": 3.5019329896907214e-05,
|
| 565 |
+
"loss": 1.6926,
|
| 566 |
+
"step": 46500
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 3.03,
|
| 570 |
+
"learning_rate": 3.485824742268041e-05,
|
| 571 |
+
"loss": 1.4406,
|
| 572 |
+
"step": 47000
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 3.06,
|
| 576 |
+
"learning_rate": 3.469716494845361e-05,
|
| 577 |
+
"loss": 1.4185,
|
| 578 |
+
"step": 47500
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 3.09,
|
| 582 |
+
"learning_rate": 3.4536082474226805e-05,
|
| 583 |
+
"loss": 1.4487,
|
| 584 |
+
"step": 48000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 3.12,
|
| 588 |
+
"learning_rate": 3.4375e-05,
|
| 589 |
+
"loss": 1.4217,
|
| 590 |
+
"step": 48500
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 3.16,
|
| 594 |
+
"learning_rate": 3.42139175257732e-05,
|
| 595 |
+
"loss": 1.469,
|
| 596 |
+
"step": 49000
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 3.19,
|
| 600 |
+
"learning_rate": 3.4052835051546396e-05,
|
| 601 |
+
"loss": 1.4631,
|
| 602 |
+
"step": 49500
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 3.22,
|
| 606 |
+
"learning_rate": 3.389175257731959e-05,
|
| 607 |
+
"loss": 1.4367,
|
| 608 |
+
"step": 50000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 3.25,
|
| 612 |
+
"learning_rate": 3.373067010309279e-05,
|
| 613 |
+
"loss": 1.4733,
|
| 614 |
+
"step": 50500
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 3.29,
|
| 618 |
+
"learning_rate": 3.356958762886598e-05,
|
| 619 |
+
"loss": 1.455,
|
| 620 |
+
"step": 51000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 3.32,
|
| 624 |
+
"learning_rate": 3.340850515463918e-05,
|
| 625 |
+
"loss": 1.4458,
|
| 626 |
+
"step": 51500
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 3.35,
|
| 630 |
+
"learning_rate": 3.3247422680412374e-05,
|
| 631 |
+
"loss": 1.4681,
|
| 632 |
+
"step": 52000
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 3.38,
|
| 636 |
+
"learning_rate": 3.3086340206185564e-05,
|
| 637 |
+
"loss": 1.4796,
|
| 638 |
+
"step": 52500
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 3.41,
|
| 642 |
+
"learning_rate": 3.292525773195876e-05,
|
| 643 |
+
"loss": 1.4969,
|
| 644 |
+
"step": 53000
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 3.45,
|
| 648 |
+
"learning_rate": 3.276417525773196e-05,
|
| 649 |
+
"loss": 1.4861,
|
| 650 |
+
"step": 53500
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 3.48,
|
| 654 |
+
"learning_rate": 3.2603092783505155e-05,
|
| 655 |
+
"loss": 1.5032,
|
| 656 |
+
"step": 54000
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 3.51,
|
| 660 |
+
"learning_rate": 3.244201030927835e-05,
|
| 661 |
+
"loss": 1.463,
|
| 662 |
+
"step": 54500
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 3.54,
|
| 666 |
+
"learning_rate": 3.228092783505155e-05,
|
| 667 |
+
"loss": 1.4772,
|
| 668 |
+
"step": 55000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 3.58,
|
| 672 |
+
"learning_rate": 3.2119845360824746e-05,
|
| 673 |
+
"loss": 1.4969,
|
| 674 |
+
"step": 55500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 3.61,
|
| 678 |
+
"learning_rate": 3.1958762886597937e-05,
|
| 679 |
+
"loss": 1.5088,
|
| 680 |
+
"step": 56000
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 3.64,
|
| 684 |
+
"learning_rate": 3.1797680412371134e-05,
|
| 685 |
+
"loss": 1.487,
|
| 686 |
+
"step": 56500
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 3.67,
|
| 690 |
+
"learning_rate": 3.163659793814433e-05,
|
| 691 |
+
"loss": 1.477,
|
| 692 |
+
"step": 57000
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 3.7,
|
| 696 |
+
"learning_rate": 3.147551546391753e-05,
|
| 697 |
+
"loss": 1.486,
|
| 698 |
+
"step": 57500
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 3.74,
|
| 702 |
+
"learning_rate": 3.1314432989690725e-05,
|
| 703 |
+
"loss": 1.4838,
|
| 704 |
+
"step": 58000
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 3.77,
|
| 708 |
+
"learning_rate": 3.115335051546392e-05,
|
| 709 |
+
"loss": 1.4992,
|
| 710 |
+
"step": 58500
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 3.8,
|
| 714 |
+
"learning_rate": 3.099226804123712e-05,
|
| 715 |
+
"loss": 1.4949,
|
| 716 |
+
"step": 59000
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 3.83,
|
| 720 |
+
"learning_rate": 3.0831185567010316e-05,
|
| 721 |
+
"loss": 1.5133,
|
| 722 |
+
"step": 59500
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 3.87,
|
| 726 |
+
"learning_rate": 3.0670103092783506e-05,
|
| 727 |
+
"loss": 1.5086,
|
| 728 |
+
"step": 60000
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 3.9,
|
| 732 |
+
"learning_rate": 3.0509020618556706e-05,
|
| 733 |
+
"loss": 1.4887,
|
| 734 |
+
"step": 60500
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 3.93,
|
| 738 |
+
"learning_rate": 3.0347938144329896e-05,
|
| 739 |
+
"loss": 1.5304,
|
| 740 |
+
"step": 61000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 3.96,
|
| 744 |
+
"learning_rate": 3.0186855670103093e-05,
|
| 745 |
+
"loss": 1.5248,
|
| 746 |
+
"step": 61500
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 3.99,
|
| 750 |
+
"learning_rate": 3.002577319587629e-05,
|
| 751 |
+
"loss": 1.4788,
|
| 752 |
+
"step": 62000
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 4.03,
|
| 756 |
+
"learning_rate": 2.9864690721649484e-05,
|
| 757 |
+
"loss": 1.3236,
|
| 758 |
+
"step": 62500
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 4.06,
|
| 762 |
+
"learning_rate": 2.970360824742268e-05,
|
| 763 |
+
"loss": 1.242,
|
| 764 |
+
"step": 63000
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 4.09,
|
| 768 |
+
"learning_rate": 2.9542525773195878e-05,
|
| 769 |
+
"loss": 1.2471,
|
| 770 |
+
"step": 63500
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 4.12,
|
| 774 |
+
"learning_rate": 2.9381443298969075e-05,
|
| 775 |
+
"loss": 1.2918,
|
| 776 |
+
"step": 64000
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 4.16,
|
| 780 |
+
"learning_rate": 2.9220360824742272e-05,
|
| 781 |
+
"loss": 1.2944,
|
| 782 |
+
"step": 64500
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 4.19,
|
| 786 |
+
"learning_rate": 2.905927835051547e-05,
|
| 787 |
+
"loss": 1.2813,
|
| 788 |
+
"step": 65000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 4.22,
|
| 792 |
+
"learning_rate": 2.889819587628866e-05,
|
| 793 |
+
"loss": 1.2879,
|
| 794 |
+
"step": 65500
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 4.25,
|
| 798 |
+
"learning_rate": 2.8737113402061856e-05,
|
| 799 |
+
"loss": 1.28,
|
| 800 |
+
"step": 66000
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 4.28,
|
| 804 |
+
"learning_rate": 2.857603092783505e-05,
|
| 805 |
+
"loss": 1.2657,
|
| 806 |
+
"step": 66500
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 4.32,
|
| 810 |
+
"learning_rate": 2.8414948453608247e-05,
|
| 811 |
+
"loss": 1.3145,
|
| 812 |
+
"step": 67000
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 4.35,
|
| 816 |
+
"learning_rate": 2.8253865979381444e-05,
|
| 817 |
+
"loss": 1.3195,
|
| 818 |
+
"step": 67500
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 4.38,
|
| 822 |
+
"learning_rate": 2.809278350515464e-05,
|
| 823 |
+
"loss": 1.2844,
|
| 824 |
+
"step": 68000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 4.41,
|
| 828 |
+
"learning_rate": 2.7931701030927838e-05,
|
| 829 |
+
"loss": 1.3339,
|
| 830 |
+
"step": 68500
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 4.45,
|
| 834 |
+
"learning_rate": 2.7770618556701035e-05,
|
| 835 |
+
"loss": 1.3075,
|
| 836 |
+
"step": 69000
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 4.48,
|
| 840 |
+
"learning_rate": 2.760953608247423e-05,
|
| 841 |
+
"loss": 1.314,
|
| 842 |
+
"step": 69500
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 4.51,
|
| 846 |
+
"learning_rate": 2.7448453608247425e-05,
|
| 847 |
+
"loss": 1.308,
|
| 848 |
+
"step": 70000
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 4.54,
|
| 852 |
+
"learning_rate": 2.728737113402062e-05,
|
| 853 |
+
"loss": 1.3234,
|
| 854 |
+
"step": 70500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 4.57,
|
| 858 |
+
"learning_rate": 2.7126288659793812e-05,
|
| 859 |
+
"loss": 1.3307,
|
| 860 |
+
"step": 71000
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 4.61,
|
| 864 |
+
"learning_rate": 2.696520618556701e-05,
|
| 865 |
+
"loss": 1.3426,
|
| 866 |
+
"step": 71500
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 4.64,
|
| 870 |
+
"learning_rate": 2.6804123711340206e-05,
|
| 871 |
+
"loss": 1.3332,
|
| 872 |
+
"step": 72000
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 4.67,
|
| 876 |
+
"learning_rate": 2.6643041237113403e-05,
|
| 877 |
+
"loss": 1.3369,
|
| 878 |
+
"step": 72500
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 4.7,
|
| 882 |
+
"learning_rate": 2.64819587628866e-05,
|
| 883 |
+
"loss": 1.3446,
|
| 884 |
+
"step": 73000
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 4.74,
|
| 888 |
+
"learning_rate": 2.6320876288659797e-05,
|
| 889 |
+
"loss": 1.3266,
|
| 890 |
+
"step": 73500
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 4.77,
|
| 894 |
+
"learning_rate": 2.615979381443299e-05,
|
| 895 |
+
"loss": 1.3303,
|
| 896 |
+
"step": 74000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 4.8,
|
| 900 |
+
"learning_rate": 2.5998711340206188e-05,
|
| 901 |
+
"loss": 1.332,
|
| 902 |
+
"step": 74500
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 4.83,
|
| 906 |
+
"learning_rate": 2.5837628865979385e-05,
|
| 907 |
+
"loss": 1.3235,
|
| 908 |
+
"step": 75000
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 4.86,
|
| 912 |
+
"learning_rate": 2.5676546391752575e-05,
|
| 913 |
+
"loss": 1.3405,
|
| 914 |
+
"step": 75500
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 4.9,
|
| 918 |
+
"learning_rate": 2.5515463917525772e-05,
|
| 919 |
+
"loss": 1.3568,
|
| 920 |
+
"step": 76000
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 4.93,
|
| 924 |
+
"learning_rate": 2.535438144329897e-05,
|
| 925 |
+
"loss": 1.3569,
|
| 926 |
+
"step": 76500
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 4.96,
|
| 930 |
+
"learning_rate": 2.5193298969072166e-05,
|
| 931 |
+
"loss": 1.3393,
|
| 932 |
+
"step": 77000
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 4.99,
|
| 936 |
+
"learning_rate": 2.5032216494845363e-05,
|
| 937 |
+
"loss": 1.3391,
|
| 938 |
+
"step": 77500
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 5.03,
|
| 942 |
+
"learning_rate": 2.4871134020618557e-05,
|
| 943 |
+
"loss": 1.1727,
|
| 944 |
+
"step": 78000
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 5.06,
|
| 948 |
+
"learning_rate": 2.4710051546391754e-05,
|
| 949 |
+
"loss": 1.1299,
|
| 950 |
+
"step": 78500
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 5.09,
|
| 954 |
+
"learning_rate": 2.4548969072164947e-05,
|
| 955 |
+
"loss": 1.1557,
|
| 956 |
+
"step": 79000
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 5.12,
|
| 960 |
+
"learning_rate": 2.4387886597938144e-05,
|
| 961 |
+
"loss": 1.1172,
|
| 962 |
+
"step": 79500
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 5.15,
|
| 966 |
+
"learning_rate": 2.422680412371134e-05,
|
| 967 |
+
"loss": 1.1455,
|
| 968 |
+
"step": 80000
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 5.19,
|
| 972 |
+
"learning_rate": 2.406572164948454e-05,
|
| 973 |
+
"loss": 1.1802,
|
| 974 |
+
"step": 80500
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 5.22,
|
| 978 |
+
"learning_rate": 2.3904639175257735e-05,
|
| 979 |
+
"loss": 1.156,
|
| 980 |
+
"step": 81000
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 5.25,
|
| 984 |
+
"learning_rate": 2.374355670103093e-05,
|
| 985 |
+
"loss": 1.1724,
|
| 986 |
+
"step": 81500
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 5.28,
|
| 990 |
+
"learning_rate": 2.3582474226804126e-05,
|
| 991 |
+
"loss": 1.1663,
|
| 992 |
+
"step": 82000
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 5.32,
|
| 996 |
+
"learning_rate": 2.342139175257732e-05,
|
| 997 |
+
"loss": 1.1727,
|
| 998 |
+
"step": 82500
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 5.35,
|
| 1002 |
+
"learning_rate": 2.3260309278350517e-05,
|
| 1003 |
+
"loss": 1.1767,
|
| 1004 |
+
"step": 83000
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 5.38,
|
| 1008 |
+
"learning_rate": 2.3099226804123713e-05,
|
| 1009 |
+
"loss": 1.1817,
|
| 1010 |
+
"step": 83500
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 5.41,
|
| 1014 |
+
"learning_rate": 2.2938144329896907e-05,
|
| 1015 |
+
"loss": 1.1878,
|
| 1016 |
+
"step": 84000
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 5.44,
|
| 1020 |
+
"learning_rate": 2.2777061855670104e-05,
|
| 1021 |
+
"loss": 1.1664,
|
| 1022 |
+
"step": 84500
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 5.48,
|
| 1026 |
+
"learning_rate": 2.26159793814433e-05,
|
| 1027 |
+
"loss": 1.1513,
|
| 1028 |
+
"step": 85000
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 5.51,
|
| 1032 |
+
"learning_rate": 2.2454896907216498e-05,
|
| 1033 |
+
"loss": 1.1762,
|
| 1034 |
+
"step": 85500
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 5.54,
|
| 1038 |
+
"learning_rate": 2.229381443298969e-05,
|
| 1039 |
+
"loss": 1.1857,
|
| 1040 |
+
"step": 86000
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 5.57,
|
| 1044 |
+
"learning_rate": 2.2132731958762885e-05,
|
| 1045 |
+
"loss": 1.1986,
|
| 1046 |
+
"step": 86500
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 5.61,
|
| 1050 |
+
"learning_rate": 2.1971649484536082e-05,
|
| 1051 |
+
"loss": 1.1738,
|
| 1052 |
+
"step": 87000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 5.64,
|
| 1056 |
+
"learning_rate": 2.181056701030928e-05,
|
| 1057 |
+
"loss": 1.2002,
|
| 1058 |
+
"step": 87500
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 5.67,
|
| 1062 |
+
"learning_rate": 2.1649484536082476e-05,
|
| 1063 |
+
"loss": 1.1515,
|
| 1064 |
+
"step": 88000
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 5.7,
|
| 1068 |
+
"learning_rate": 2.1488402061855673e-05,
|
| 1069 |
+
"loss": 1.2072,
|
| 1070 |
+
"step": 88500
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 5.73,
|
| 1074 |
+
"learning_rate": 2.1327319587628867e-05,
|
| 1075 |
+
"loss": 1.1773,
|
| 1076 |
+
"step": 89000
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 5.77,
|
| 1080 |
+
"learning_rate": 2.1166237113402064e-05,
|
| 1081 |
+
"loss": 1.1906,
|
| 1082 |
+
"step": 89500
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 5.8,
|
| 1086 |
+
"learning_rate": 2.1005154639175257e-05,
|
| 1087 |
+
"loss": 1.2048,
|
| 1088 |
+
"step": 90000
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 5.83,
|
| 1092 |
+
"learning_rate": 2.0844072164948454e-05,
|
| 1093 |
+
"loss": 1.2176,
|
| 1094 |
+
"step": 90500
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 5.86,
|
| 1098 |
+
"learning_rate": 2.0682989690721648e-05,
|
| 1099 |
+
"loss": 1.1869,
|
| 1100 |
+
"step": 91000
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 5.9,
|
| 1104 |
+
"learning_rate": 2.0521907216494845e-05,
|
| 1105 |
+
"loss": 1.1964,
|
| 1106 |
+
"step": 91500
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 5.93,
|
| 1110 |
+
"learning_rate": 2.0360824742268042e-05,
|
| 1111 |
+
"loss": 1.176,
|
| 1112 |
+
"step": 92000
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 5.96,
|
| 1116 |
+
"learning_rate": 2.019974226804124e-05,
|
| 1117 |
+
"loss": 1.2437,
|
| 1118 |
+
"step": 92500
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 5.99,
|
| 1122 |
+
"learning_rate": 2.0038659793814436e-05,
|
| 1123 |
+
"loss": 1.2032,
|
| 1124 |
+
"step": 93000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 6.02,
|
| 1128 |
+
"learning_rate": 1.987757731958763e-05,
|
| 1129 |
+
"loss": 1.0818,
|
| 1130 |
+
"step": 93500
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 6.06,
|
| 1134 |
+
"learning_rate": 1.9716494845360827e-05,
|
| 1135 |
+
"loss": 1.0236,
|
| 1136 |
+
"step": 94000
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 6.09,
|
| 1140 |
+
"learning_rate": 1.955541237113402e-05,
|
| 1141 |
+
"loss": 1.0517,
|
| 1142 |
+
"step": 94500
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 6.12,
|
| 1146 |
+
"learning_rate": 1.9394329896907217e-05,
|
| 1147 |
+
"loss": 1.0446,
|
| 1148 |
+
"step": 95000
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 6.15,
|
| 1152 |
+
"learning_rate": 1.9233247422680414e-05,
|
| 1153 |
+
"loss": 1.0508,
|
| 1154 |
+
"step": 95500
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 6.19,
|
| 1158 |
+
"learning_rate": 1.9072164948453608e-05,
|
| 1159 |
+
"loss": 1.0296,
|
| 1160 |
+
"step": 96000
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 6.22,
|
| 1164 |
+
"learning_rate": 1.8911082474226805e-05,
|
| 1165 |
+
"loss": 1.0353,
|
| 1166 |
+
"step": 96500
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 6.25,
|
| 1170 |
+
"learning_rate": 1.8750000000000002e-05,
|
| 1171 |
+
"loss": 1.0498,
|
| 1172 |
+
"step": 97000
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 6.28,
|
| 1176 |
+
"learning_rate": 1.85889175257732e-05,
|
| 1177 |
+
"loss": 1.0701,
|
| 1178 |
+
"step": 97500
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 6.31,
|
| 1182 |
+
"learning_rate": 1.8427835051546392e-05,
|
| 1183 |
+
"loss": 1.0654,
|
| 1184 |
+
"step": 98000
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 6.35,
|
| 1188 |
+
"learning_rate": 1.8266752577319586e-05,
|
| 1189 |
+
"loss": 1.0504,
|
| 1190 |
+
"step": 98500
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 6.38,
|
| 1194 |
+
"learning_rate": 1.8105670103092783e-05,
|
| 1195 |
+
"loss": 1.0527,
|
| 1196 |
+
"step": 99000
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 6.41,
|
| 1200 |
+
"learning_rate": 1.794458762886598e-05,
|
| 1201 |
+
"loss": 1.0575,
|
| 1202 |
+
"step": 99500
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 6.44,
|
| 1206 |
+
"learning_rate": 1.7783505154639177e-05,
|
| 1207 |
+
"loss": 1.0683,
|
| 1208 |
+
"step": 100000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 6.48,
|
| 1212 |
+
"learning_rate": 1.7622422680412374e-05,
|
| 1213 |
+
"loss": 1.0589,
|
| 1214 |
+
"step": 100500
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 6.51,
|
| 1218 |
+
"learning_rate": 1.7461340206185568e-05,
|
| 1219 |
+
"loss": 1.0853,
|
| 1220 |
+
"step": 101000
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 6.54,
|
| 1224 |
+
"learning_rate": 1.7300257731958764e-05,
|
| 1225 |
+
"loss": 1.0632,
|
| 1226 |
+
"step": 101500
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 6.57,
|
| 1230 |
+
"learning_rate": 1.7139175257731958e-05,
|
| 1231 |
+
"loss": 1.0379,
|
| 1232 |
+
"step": 102000
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 6.6,
|
| 1236 |
+
"learning_rate": 1.6978092783505155e-05,
|
| 1237 |
+
"loss": 1.052,
|
| 1238 |
+
"step": 102500
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 6.64,
|
| 1242 |
+
"learning_rate": 1.6817010309278352e-05,
|
| 1243 |
+
"loss": 1.0583,
|
| 1244 |
+
"step": 103000
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 6.67,
|
| 1248 |
+
"learning_rate": 1.6655927835051546e-05,
|
| 1249 |
+
"loss": 1.1123,
|
| 1250 |
+
"step": 103500
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 6.7,
|
| 1254 |
+
"learning_rate": 1.6494845360824743e-05,
|
| 1255 |
+
"loss": 1.0807,
|
| 1256 |
+
"step": 104000
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 6.73,
|
| 1260 |
+
"learning_rate": 1.633376288659794e-05,
|
| 1261 |
+
"loss": 1.0447,
|
| 1262 |
+
"step": 104500
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 6.77,
|
| 1266 |
+
"learning_rate": 1.6172680412371137e-05,
|
| 1267 |
+
"loss": 1.0392,
|
| 1268 |
+
"step": 105000
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 6.8,
|
| 1272 |
+
"learning_rate": 1.601159793814433e-05,
|
| 1273 |
+
"loss": 1.0944,
|
| 1274 |
+
"step": 105500
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 6.83,
|
| 1278 |
+
"learning_rate": 1.5850515463917527e-05,
|
| 1279 |
+
"loss": 1.0617,
|
| 1280 |
+
"step": 106000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 6.86,
|
| 1284 |
+
"learning_rate": 1.568943298969072e-05,
|
| 1285 |
+
"loss": 1.0552,
|
| 1286 |
+
"step": 106500
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 6.89,
|
| 1290 |
+
"learning_rate": 1.5528350515463918e-05,
|
| 1291 |
+
"loss": 1.1154,
|
| 1292 |
+
"step": 107000
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 6.93,
|
| 1296 |
+
"learning_rate": 1.5367268041237115e-05,
|
| 1297 |
+
"loss": 1.0879,
|
| 1298 |
+
"step": 107500
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 6.96,
|
| 1302 |
+
"learning_rate": 1.5206185567010308e-05,
|
| 1303 |
+
"loss": 1.0709,
|
| 1304 |
+
"step": 108000
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 6.99,
|
| 1308 |
+
"learning_rate": 1.5045103092783505e-05,
|
| 1309 |
+
"loss": 1.0835,
|
| 1310 |
+
"step": 108500
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 7.02,
|
| 1314 |
+
"learning_rate": 1.4884020618556702e-05,
|
| 1315 |
+
"loss": 0.974,
|
| 1316 |
+
"step": 109000
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 7.06,
|
| 1320 |
+
"learning_rate": 1.4722938144329898e-05,
|
| 1321 |
+
"loss": 0.9417,
|
| 1322 |
+
"step": 109500
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 7.09,
|
| 1326 |
+
"learning_rate": 1.4561855670103095e-05,
|
| 1327 |
+
"loss": 0.9393,
|
| 1328 |
+
"step": 110000
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 7.12,
|
| 1332 |
+
"learning_rate": 1.4400773195876288e-05,
|
| 1333 |
+
"loss": 0.9574,
|
| 1334 |
+
"step": 110500
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 7.15,
|
| 1338 |
+
"learning_rate": 1.4239690721649485e-05,
|
| 1339 |
+
"loss": 0.9629,
|
| 1340 |
+
"step": 111000
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 7.18,
|
| 1344 |
+
"learning_rate": 1.407860824742268e-05,
|
| 1345 |
+
"loss": 0.97,
|
| 1346 |
+
"step": 111500
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 7.22,
|
| 1350 |
+
"learning_rate": 1.3917525773195878e-05,
|
| 1351 |
+
"loss": 0.9266,
|
| 1352 |
+
"step": 112000
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 7.25,
|
| 1356 |
+
"learning_rate": 1.3756443298969075e-05,
|
| 1357 |
+
"loss": 0.9694,
|
| 1358 |
+
"step": 112500
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 7.28,
|
| 1362 |
+
"learning_rate": 1.3595360824742268e-05,
|
| 1363 |
+
"loss": 0.9429,
|
| 1364 |
+
"step": 113000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 7.31,
|
| 1368 |
+
"learning_rate": 1.3434278350515463e-05,
|
| 1369 |
+
"loss": 0.9373,
|
| 1370 |
+
"step": 113500
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 7.35,
|
| 1374 |
+
"learning_rate": 1.327319587628866e-05,
|
| 1375 |
+
"loss": 0.9613,
|
| 1376 |
+
"step": 114000
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 7.38,
|
| 1380 |
+
"learning_rate": 1.3112113402061857e-05,
|
| 1381 |
+
"loss": 0.9507,
|
| 1382 |
+
"step": 114500
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 7.41,
|
| 1386 |
+
"learning_rate": 1.2951030927835053e-05,
|
| 1387 |
+
"loss": 0.9546,
|
| 1388 |
+
"step": 115000
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 7.44,
|
| 1392 |
+
"learning_rate": 1.2789948453608246e-05,
|
| 1393 |
+
"loss": 0.9737,
|
| 1394 |
+
"step": 115500
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 7.47,
|
| 1398 |
+
"learning_rate": 1.2628865979381443e-05,
|
| 1399 |
+
"loss": 0.9765,
|
| 1400 |
+
"step": 116000
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 7.51,
|
| 1404 |
+
"learning_rate": 1.246778350515464e-05,
|
| 1405 |
+
"loss": 0.9772,
|
| 1406 |
+
"step": 116500
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 7.54,
|
| 1410 |
+
"learning_rate": 1.2306701030927836e-05,
|
| 1411 |
+
"loss": 0.9745,
|
| 1412 |
+
"step": 117000
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 7.57,
|
| 1416 |
+
"learning_rate": 1.2145618556701031e-05,
|
| 1417 |
+
"loss": 0.9693,
|
| 1418 |
+
"step": 117500
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 7.6,
|
| 1422 |
+
"learning_rate": 1.1984536082474228e-05,
|
| 1423 |
+
"loss": 0.9641,
|
| 1424 |
+
"step": 118000
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 7.64,
|
| 1428 |
+
"learning_rate": 1.1823453608247423e-05,
|
| 1429 |
+
"loss": 0.9696,
|
| 1430 |
+
"step": 118500
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 7.67,
|
| 1434 |
+
"learning_rate": 1.1662371134020619e-05,
|
| 1435 |
+
"loss": 0.9849,
|
| 1436 |
+
"step": 119000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 7.7,
|
| 1440 |
+
"learning_rate": 1.1501288659793814e-05,
|
| 1441 |
+
"loss": 1.0011,
|
| 1442 |
+
"step": 119500
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 7.73,
|
| 1446 |
+
"learning_rate": 1.134020618556701e-05,
|
| 1447 |
+
"loss": 0.9824,
|
| 1448 |
+
"step": 120000
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 7.76,
|
| 1452 |
+
"learning_rate": 1.1179123711340208e-05,
|
| 1453 |
+
"loss": 0.9685,
|
| 1454 |
+
"step": 120500
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 7.8,
|
| 1458 |
+
"learning_rate": 1.1018041237113403e-05,
|
| 1459 |
+
"loss": 0.9807,
|
| 1460 |
+
"step": 121000
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 7.83,
|
| 1464 |
+
"learning_rate": 1.0856958762886598e-05,
|
| 1465 |
+
"loss": 0.9776,
|
| 1466 |
+
"step": 121500
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 7.86,
|
| 1470 |
+
"learning_rate": 1.0695876288659794e-05,
|
| 1471 |
+
"loss": 0.9665,
|
| 1472 |
+
"step": 122000
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 7.89,
|
| 1476 |
+
"learning_rate": 1.053479381443299e-05,
|
| 1477 |
+
"loss": 0.9841,
|
| 1478 |
+
"step": 122500
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 7.93,
|
| 1482 |
+
"learning_rate": 1.0373711340206186e-05,
|
| 1483 |
+
"loss": 0.978,
|
| 1484 |
+
"step": 123000
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 7.96,
|
| 1488 |
+
"learning_rate": 1.0212628865979381e-05,
|
| 1489 |
+
"loss": 0.9936,
|
| 1490 |
+
"step": 123500
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 7.99,
|
| 1494 |
+
"learning_rate": 1.0051546391752578e-05,
|
| 1495 |
+
"loss": 0.979,
|
| 1496 |
+
"step": 124000
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 8.02,
|
| 1500 |
+
"learning_rate": 9.890463917525774e-06,
|
| 1501 |
+
"loss": 0.9301,
|
| 1502 |
+
"step": 124500
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 8.05,
|
| 1506 |
+
"learning_rate": 9.729381443298969e-06,
|
| 1507 |
+
"loss": 0.8839,
|
| 1508 |
+
"step": 125000
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 8.09,
|
| 1512 |
+
"learning_rate": 9.568298969072164e-06,
|
| 1513 |
+
"loss": 0.8927,
|
| 1514 |
+
"step": 125500
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 8.12,
|
| 1518 |
+
"learning_rate": 9.407216494845361e-06,
|
| 1519 |
+
"loss": 0.8764,
|
| 1520 |
+
"step": 126000
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 8.15,
|
| 1524 |
+
"learning_rate": 9.246134020618558e-06,
|
| 1525 |
+
"loss": 0.8763,
|
| 1526 |
+
"step": 126500
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 8.18,
|
| 1530 |
+
"learning_rate": 9.085051546391753e-06,
|
| 1531 |
+
"loss": 0.8848,
|
| 1532 |
+
"step": 127000
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 8.22,
|
| 1536 |
+
"learning_rate": 8.923969072164949e-06,
|
| 1537 |
+
"loss": 0.8908,
|
| 1538 |
+
"step": 127500
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 8.25,
|
| 1542 |
+
"learning_rate": 8.762886597938144e-06,
|
| 1543 |
+
"loss": 0.9014,
|
| 1544 |
+
"step": 128000
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 8.28,
|
| 1548 |
+
"learning_rate": 8.601804123711341e-06,
|
| 1549 |
+
"loss": 0.882,
|
| 1550 |
+
"step": 128500
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 8.31,
|
| 1554 |
+
"learning_rate": 8.440721649484536e-06,
|
| 1555 |
+
"loss": 0.9012,
|
| 1556 |
+
"step": 129000
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 8.34,
|
| 1560 |
+
"learning_rate": 8.279639175257732e-06,
|
| 1561 |
+
"loss": 0.859,
|
| 1562 |
+
"step": 129500
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 8.38,
|
| 1566 |
+
"learning_rate": 8.118556701030929e-06,
|
| 1567 |
+
"loss": 0.8917,
|
| 1568 |
+
"step": 130000
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 8.41,
|
| 1572 |
+
"learning_rate": 7.957474226804124e-06,
|
| 1573 |
+
"loss": 0.9153,
|
| 1574 |
+
"step": 130500
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 8.44,
|
| 1578 |
+
"learning_rate": 7.79639175257732e-06,
|
| 1579 |
+
"loss": 0.8929,
|
| 1580 |
+
"step": 131000
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 8.47,
|
| 1584 |
+
"learning_rate": 7.635309278350516e-06,
|
| 1585 |
+
"loss": 0.8982,
|
| 1586 |
+
"step": 131500
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 8.51,
|
| 1590 |
+
"learning_rate": 7.4742268041237115e-06,
|
| 1591 |
+
"loss": 0.8929,
|
| 1592 |
+
"step": 132000
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 8.54,
|
| 1596 |
+
"learning_rate": 7.313144329896908e-06,
|
| 1597 |
+
"loss": 0.9249,
|
| 1598 |
+
"step": 132500
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 8.57,
|
| 1602 |
+
"learning_rate": 7.152061855670103e-06,
|
| 1603 |
+
"loss": 0.8743,
|
| 1604 |
+
"step": 133000
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 8.6,
|
| 1608 |
+
"learning_rate": 6.990979381443299e-06,
|
| 1609 |
+
"loss": 0.8956,
|
| 1610 |
+
"step": 133500
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 8.63,
|
| 1614 |
+
"learning_rate": 6.829896907216494e-06,
|
| 1615 |
+
"loss": 0.8994,
|
| 1616 |
+
"step": 134000
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 8.67,
|
| 1620 |
+
"learning_rate": 6.668814432989691e-06,
|
| 1621 |
+
"loss": 0.8952,
|
| 1622 |
+
"step": 134500
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 8.7,
|
| 1626 |
+
"learning_rate": 6.5077319587628875e-06,
|
| 1627 |
+
"loss": 0.9033,
|
| 1628 |
+
"step": 135000
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 8.73,
|
| 1632 |
+
"learning_rate": 6.346649484536083e-06,
|
| 1633 |
+
"loss": 0.9032,
|
| 1634 |
+
"step": 135500
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 8.76,
|
| 1638 |
+
"learning_rate": 6.185567010309279e-06,
|
| 1639 |
+
"loss": 0.8964,
|
| 1640 |
+
"step": 136000
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 8.8,
|
| 1644 |
+
"learning_rate": 6.024484536082474e-06,
|
| 1645 |
+
"loss": 0.8843,
|
| 1646 |
+
"step": 136500
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 8.83,
|
| 1650 |
+
"learning_rate": 5.86340206185567e-06,
|
| 1651 |
+
"loss": 0.9027,
|
| 1652 |
+
"step": 137000
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 8.86,
|
| 1656 |
+
"learning_rate": 5.7023195876288665e-06,
|
| 1657 |
+
"loss": 0.9043,
|
| 1658 |
+
"step": 137500
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 8.89,
|
| 1662 |
+
"learning_rate": 5.541237113402062e-06,
|
| 1663 |
+
"loss": 0.8974,
|
| 1664 |
+
"step": 138000
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 8.92,
|
| 1668 |
+
"learning_rate": 5.380154639175258e-06,
|
| 1669 |
+
"loss": 0.8937,
|
| 1670 |
+
"step": 138500
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 8.96,
|
| 1674 |
+
"learning_rate": 5.219072164948454e-06,
|
| 1675 |
+
"loss": 0.8971,
|
| 1676 |
+
"step": 139000
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 8.99,
|
| 1680 |
+
"learning_rate": 5.057989690721649e-06,
|
| 1681 |
+
"loss": 0.9233,
|
| 1682 |
+
"step": 139500
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 9.02,
|
| 1686 |
+
"learning_rate": 4.8969072164948455e-06,
|
| 1687 |
+
"loss": 0.852,
|
| 1688 |
+
"step": 140000
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 9.05,
|
| 1692 |
+
"learning_rate": 4.735824742268042e-06,
|
| 1693 |
+
"loss": 0.8142,
|
| 1694 |
+
"step": 140500
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 9.09,
|
| 1698 |
+
"learning_rate": 4.574742268041237e-06,
|
| 1699 |
+
"loss": 0.8369,
|
| 1700 |
+
"step": 141000
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 9.12,
|
| 1704 |
+
"learning_rate": 4.413659793814433e-06,
|
| 1705 |
+
"loss": 0.8492,
|
| 1706 |
+
"step": 141500
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 9.15,
|
| 1710 |
+
"learning_rate": 4.252577319587629e-06,
|
| 1711 |
+
"loss": 0.8308,
|
| 1712 |
+
"step": 142000
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 9.18,
|
| 1716 |
+
"learning_rate": 4.0914948453608246e-06,
|
| 1717 |
+
"loss": 0.8706,
|
| 1718 |
+
"step": 142500
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 9.21,
|
| 1722 |
+
"learning_rate": 3.930412371134021e-06,
|
| 1723 |
+
"loss": 0.8536,
|
| 1724 |
+
"step": 143000
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 9.25,
|
| 1728 |
+
"learning_rate": 3.7693298969072164e-06,
|
| 1729 |
+
"loss": 0.8231,
|
| 1730 |
+
"step": 143500
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 9.28,
|
| 1734 |
+
"learning_rate": 3.608247422680412e-06,
|
| 1735 |
+
"loss": 0.8492,
|
| 1736 |
+
"step": 144000
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 9.31,
|
| 1740 |
+
"learning_rate": 3.4471649484536087e-06,
|
| 1741 |
+
"loss": 0.8233,
|
| 1742 |
+
"step": 144500
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 9.34,
|
| 1746 |
+
"learning_rate": 3.2860824742268044e-06,
|
| 1747 |
+
"loss": 0.8217,
|
| 1748 |
+
"step": 145000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 9.38,
|
| 1752 |
+
"learning_rate": 3.125e-06,
|
| 1753 |
+
"loss": 0.8694,
|
| 1754 |
+
"step": 145500
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 9.41,
|
| 1758 |
+
"learning_rate": 2.963917525773196e-06,
|
| 1759 |
+
"loss": 0.866,
|
| 1760 |
+
"step": 146000
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 9.44,
|
| 1764 |
+
"learning_rate": 2.802835051546392e-06,
|
| 1765 |
+
"loss": 0.8402,
|
| 1766 |
+
"step": 146500
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 9.47,
|
| 1770 |
+
"learning_rate": 2.6417525773195877e-06,
|
| 1771 |
+
"loss": 0.84,
|
| 1772 |
+
"step": 147000
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 9.5,
|
| 1776 |
+
"learning_rate": 2.4806701030927835e-06,
|
| 1777 |
+
"loss": 0.8501,
|
| 1778 |
+
"step": 147500
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 9.54,
|
| 1782 |
+
"learning_rate": 2.3195876288659796e-06,
|
| 1783 |
+
"loss": 0.8374,
|
| 1784 |
+
"step": 148000
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 9.57,
|
| 1788 |
+
"learning_rate": 2.1585051546391753e-06,
|
| 1789 |
+
"loss": 0.8645,
|
| 1790 |
+
"step": 148500
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 9.6,
|
| 1794 |
+
"learning_rate": 1.997422680412371e-06,
|
| 1795 |
+
"loss": 0.8513,
|
| 1796 |
+
"step": 149000
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 9.63,
|
| 1800 |
+
"learning_rate": 1.8363402061855672e-06,
|
| 1801 |
+
"loss": 0.8475,
|
| 1802 |
+
"step": 149500
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 9.66,
|
| 1806 |
+
"learning_rate": 1.675257731958763e-06,
|
| 1807 |
+
"loss": 0.8474,
|
| 1808 |
+
"step": 150000
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 9.7,
|
| 1812 |
+
"learning_rate": 1.5141752577319588e-06,
|
| 1813 |
+
"loss": 0.8414,
|
| 1814 |
+
"step": 150500
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 9.73,
|
| 1818 |
+
"learning_rate": 1.3530927835051548e-06,
|
| 1819 |
+
"loss": 0.8441,
|
| 1820 |
+
"step": 151000
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 9.76,
|
| 1824 |
+
"learning_rate": 1.1920103092783505e-06,
|
| 1825 |
+
"loss": 0.8762,
|
| 1826 |
+
"step": 151500
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 9.79,
|
| 1830 |
+
"learning_rate": 1.0309278350515464e-06,
|
| 1831 |
+
"loss": 0.842,
|
| 1832 |
+
"step": 152000
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 9.83,
|
| 1836 |
+
"learning_rate": 8.698453608247423e-07,
|
| 1837 |
+
"loss": 0.8571,
|
| 1838 |
+
"step": 152500
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 9.86,
|
| 1842 |
+
"learning_rate": 7.087628865979382e-07,
|
| 1843 |
+
"loss": 0.8309,
|
| 1844 |
+
"step": 153000
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 9.89,
|
| 1848 |
+
"learning_rate": 5.47680412371134e-07,
|
| 1849 |
+
"loss": 0.8334,
|
| 1850 |
+
"step": 153500
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 9.92,
|
| 1854 |
+
"learning_rate": 3.8659793814432993e-07,
|
| 1855 |
+
"loss": 0.8501,
|
| 1856 |
+
"step": 154000
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 9.95,
|
| 1860 |
+
"learning_rate": 2.2551546391752576e-07,
|
| 1861 |
+
"loss": 0.8423,
|
| 1862 |
+
"step": 154500
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 9.99,
|
| 1866 |
+
"learning_rate": 6.443298969072165e-08,
|
| 1867 |
+
"loss": 0.8466,
|
| 1868 |
+
"step": 155000
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 10.0,
|
| 1872 |
+
"step": 155200,
|
| 1873 |
+
"total_flos": 1.441343477907456e+17,
|
| 1874 |
+
"train_loss": 1.3494602605485424,
|
| 1875 |
+
"train_runtime": 31681.5151,
|
| 1876 |
+
"train_samples_per_second": 4.899,
|
| 1877 |
+
"train_steps_per_second": 4.899
|
| 1878 |
+
}
|
| 1879 |
+
],
|
| 1880 |
+
"max_steps": 155200,
|
| 1881 |
+
"num_train_epochs": 10,
|
| 1882 |
+
"total_flos": 1.441343477907456e+17,
|
| 1883 |
+
"trial_name": null,
|
| 1884 |
+
"trial_params": null
|
| 1885 |
+
}
|