End of training
Browse files- all_results.json +8 -0
- train_results.json +8 -0
- train_results.txt +6 -0
- trainer_state.json +3359 -0
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"train_loss": 1.7010083939733822,
|
| 4 |
+
"train_runtime": 6451.2169,
|
| 5 |
+
"train_samples": 42367,
|
| 6 |
+
"train_samples_per_second": 32.836,
|
| 7 |
+
"train_steps_per_second": 8.209
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"train_loss": 1.7010083939733822,
|
| 4 |
+
"train_runtime": 6451.2169,
|
| 5 |
+
"train_samples": 42367,
|
| 6 |
+
"train_samples_per_second": 32.836,
|
| 7 |
+
"train_steps_per_second": 8.209
|
| 8 |
+
}
|
train_results.txt
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
epoch = 5.0
|
| 2 |
+
train_loss = 1.7010083939733822
|
| 3 |
+
train_runtime = 6451.2169
|
| 4 |
+
train_samples = 42367
|
| 5 |
+
train_samples_per_second = 32.836
|
| 6 |
+
train_steps_per_second = 8.209
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3359 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 2.720076322555542,
|
| 3 |
+
"best_model_checkpoint": "./models/kobart_4_5.6e-5_datav2_min30_lp5.0_temperature1.0/checkpoint-20000",
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"global_step": 52960,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.01,
|
| 12 |
+
"learning_rate": 1.0574018126888216e-06,
|
| 13 |
+
"loss": 9.5934,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.02,
|
| 18 |
+
"learning_rate": 2.1148036253776433e-06,
|
| 19 |
+
"loss": 4.9023,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.03,
|
| 24 |
+
"learning_rate": 3.1722054380664653e-06,
|
| 25 |
+
"loss": 3.5742,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.04,
|
| 30 |
+
"learning_rate": 4.2296072507552865e-06,
|
| 31 |
+
"loss": 3.2585,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.05,
|
| 36 |
+
"learning_rate": 5.287009063444109e-06,
|
| 37 |
+
"loss": 3.044,
|
| 38 |
+
"step": 500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.06,
|
| 42 |
+
"learning_rate": 6.344410876132931e-06,
|
| 43 |
+
"loss": 2.9663,
|
| 44 |
+
"step": 600
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.07,
|
| 48 |
+
"learning_rate": 7.401812688821753e-06,
|
| 49 |
+
"loss": 2.7921,
|
| 50 |
+
"step": 700
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.08,
|
| 54 |
+
"learning_rate": 8.459214501510573e-06,
|
| 55 |
+
"loss": 2.794,
|
| 56 |
+
"step": 800
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.08,
|
| 60 |
+
"learning_rate": 9.516616314199396e-06,
|
| 61 |
+
"loss": 2.6951,
|
| 62 |
+
"step": 900
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.09,
|
| 66 |
+
"learning_rate": 1.0574018126888217e-05,
|
| 67 |
+
"loss": 2.7445,
|
| 68 |
+
"step": 1000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.1,
|
| 72 |
+
"learning_rate": 1.1631419939577038e-05,
|
| 73 |
+
"loss": 2.6744,
|
| 74 |
+
"step": 1100
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.11,
|
| 78 |
+
"learning_rate": 1.2688821752265861e-05,
|
| 79 |
+
"loss": 2.644,
|
| 80 |
+
"step": 1200
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.12,
|
| 84 |
+
"learning_rate": 1.3746223564954682e-05,
|
| 85 |
+
"loss": 2.5615,
|
| 86 |
+
"step": 1300
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.13,
|
| 90 |
+
"learning_rate": 1.4803625377643505e-05,
|
| 91 |
+
"loss": 2.6068,
|
| 92 |
+
"step": 1400
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.14,
|
| 96 |
+
"learning_rate": 1.5861027190332325e-05,
|
| 97 |
+
"loss": 2.5422,
|
| 98 |
+
"step": 1500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.15,
|
| 102 |
+
"learning_rate": 1.6918429003021146e-05,
|
| 103 |
+
"loss": 2.5506,
|
| 104 |
+
"step": 1600
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.16,
|
| 108 |
+
"learning_rate": 1.797583081570997e-05,
|
| 109 |
+
"loss": 2.5316,
|
| 110 |
+
"step": 1700
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.17,
|
| 114 |
+
"learning_rate": 1.9033232628398792e-05,
|
| 115 |
+
"loss": 2.4983,
|
| 116 |
+
"step": 1800
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.18,
|
| 120 |
+
"learning_rate": 2.0090634441087613e-05,
|
| 121 |
+
"loss": 2.4222,
|
| 122 |
+
"step": 1900
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.19,
|
| 126 |
+
"learning_rate": 2.1148036253776434e-05,
|
| 127 |
+
"loss": 2.5043,
|
| 128 |
+
"step": 2000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.2,
|
| 132 |
+
"learning_rate": 2.2205438066465256e-05,
|
| 133 |
+
"loss": 2.4956,
|
| 134 |
+
"step": 2100
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.21,
|
| 138 |
+
"learning_rate": 2.3262839879154077e-05,
|
| 139 |
+
"loss": 2.503,
|
| 140 |
+
"step": 2200
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.22,
|
| 144 |
+
"learning_rate": 2.43202416918429e-05,
|
| 145 |
+
"loss": 2.4681,
|
| 146 |
+
"step": 2300
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.23,
|
| 150 |
+
"learning_rate": 2.5377643504531723e-05,
|
| 151 |
+
"loss": 2.4417,
|
| 152 |
+
"step": 2400
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.24,
|
| 156 |
+
"learning_rate": 2.6435045317220544e-05,
|
| 157 |
+
"loss": 2.3993,
|
| 158 |
+
"step": 2500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.25,
|
| 162 |
+
"learning_rate": 2.7492447129909365e-05,
|
| 163 |
+
"loss": 2.424,
|
| 164 |
+
"step": 2600
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.25,
|
| 168 |
+
"learning_rate": 2.8549848942598183e-05,
|
| 169 |
+
"loss": 2.4301,
|
| 170 |
+
"step": 2700
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.26,
|
| 174 |
+
"learning_rate": 2.960725075528701e-05,
|
| 175 |
+
"loss": 2.5465,
|
| 176 |
+
"step": 2800
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.27,
|
| 180 |
+
"learning_rate": 3.066465256797583e-05,
|
| 181 |
+
"loss": 2.398,
|
| 182 |
+
"step": 2900
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.28,
|
| 186 |
+
"learning_rate": 3.172205438066465e-05,
|
| 187 |
+
"loss": 2.4061,
|
| 188 |
+
"step": 3000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.29,
|
| 192 |
+
"learning_rate": 3.277945619335347e-05,
|
| 193 |
+
"loss": 2.4014,
|
| 194 |
+
"step": 3100
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.3,
|
| 198 |
+
"learning_rate": 3.383685800604229e-05,
|
| 199 |
+
"loss": 2.4134,
|
| 200 |
+
"step": 3200
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.31,
|
| 204 |
+
"learning_rate": 3.4894259818731113e-05,
|
| 205 |
+
"loss": 2.4089,
|
| 206 |
+
"step": 3300
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.32,
|
| 210 |
+
"learning_rate": 3.595166163141994e-05,
|
| 211 |
+
"loss": 2.4397,
|
| 212 |
+
"step": 3400
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.33,
|
| 216 |
+
"learning_rate": 3.700906344410876e-05,
|
| 217 |
+
"loss": 2.3923,
|
| 218 |
+
"step": 3500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.34,
|
| 222 |
+
"learning_rate": 3.8066465256797584e-05,
|
| 223 |
+
"loss": 2.3635,
|
| 224 |
+
"step": 3600
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.35,
|
| 228 |
+
"learning_rate": 3.9123867069486405e-05,
|
| 229 |
+
"loss": 2.3589,
|
| 230 |
+
"step": 3700
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.36,
|
| 234 |
+
"learning_rate": 4.0181268882175226e-05,
|
| 235 |
+
"loss": 2.3856,
|
| 236 |
+
"step": 3800
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.37,
|
| 240 |
+
"learning_rate": 4.123867069486405e-05,
|
| 241 |
+
"loss": 2.3656,
|
| 242 |
+
"step": 3900
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.38,
|
| 246 |
+
"learning_rate": 4.229607250755287e-05,
|
| 247 |
+
"loss": 2.402,
|
| 248 |
+
"step": 4000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.39,
|
| 252 |
+
"learning_rate": 4.335347432024169e-05,
|
| 253 |
+
"loss": 2.4819,
|
| 254 |
+
"step": 4100
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.4,
|
| 258 |
+
"learning_rate": 4.441087613293051e-05,
|
| 259 |
+
"loss": 2.4714,
|
| 260 |
+
"step": 4200
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.41,
|
| 264 |
+
"learning_rate": 4.546827794561933e-05,
|
| 265 |
+
"loss": 2.3861,
|
| 266 |
+
"step": 4300
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.42,
|
| 270 |
+
"learning_rate": 4.6525679758308154e-05,
|
| 271 |
+
"loss": 2.4166,
|
| 272 |
+
"step": 4400
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.42,
|
| 276 |
+
"learning_rate": 4.7583081570996975e-05,
|
| 277 |
+
"loss": 2.3908,
|
| 278 |
+
"step": 4500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.43,
|
| 282 |
+
"learning_rate": 4.86404833836858e-05,
|
| 283 |
+
"loss": 2.4096,
|
| 284 |
+
"step": 4600
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.44,
|
| 288 |
+
"learning_rate": 4.9697885196374624e-05,
|
| 289 |
+
"loss": 2.3846,
|
| 290 |
+
"step": 4700
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.45,
|
| 294 |
+
"learning_rate": 5.0755287009063445e-05,
|
| 295 |
+
"loss": 2.3477,
|
| 296 |
+
"step": 4800
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.46,
|
| 300 |
+
"learning_rate": 5.1812688821752266e-05,
|
| 301 |
+
"loss": 2.4371,
|
| 302 |
+
"step": 4900
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.47,
|
| 306 |
+
"learning_rate": 5.287009063444109e-05,
|
| 307 |
+
"loss": 2.3968,
|
| 308 |
+
"step": 5000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.47,
|
| 312 |
+
"eval_bleu1": 27.0594,
|
| 313 |
+
"eval_bleu2": 15.1133,
|
| 314 |
+
"eval_bleu3": 8.4503,
|
| 315 |
+
"eval_bleu4": 4.564,
|
| 316 |
+
"eval_gen_len": 48.5501,
|
| 317 |
+
"eval_loss": 2.9096498489379883,
|
| 318 |
+
"eval_rouge1": 32.7469,
|
| 319 |
+
"eval_rouge2": 10.9679,
|
| 320 |
+
"eval_rougeL": 21.4954,
|
| 321 |
+
"eval_runtime": 125.8926,
|
| 322 |
+
"eval_samples_per_second": 3.408,
|
| 323 |
+
"eval_steps_per_second": 0.032,
|
| 324 |
+
"step": 5000
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.48,
|
| 328 |
+
"learning_rate": 5.392749244712991e-05,
|
| 329 |
+
"loss": 2.3874,
|
| 330 |
+
"step": 5100
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.49,
|
| 334 |
+
"learning_rate": 5.498489425981873e-05,
|
| 335 |
+
"loss": 2.3827,
|
| 336 |
+
"step": 5200
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 0.5,
|
| 340 |
+
"learning_rate": 5.599530043638805e-05,
|
| 341 |
+
"loss": 2.3515,
|
| 342 |
+
"step": 5300
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.51,
|
| 346 |
+
"learning_rate": 5.587781134608929e-05,
|
| 347 |
+
"loss": 2.4104,
|
| 348 |
+
"step": 5400
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.52,
|
| 352 |
+
"learning_rate": 5.5760322255790535e-05,
|
| 353 |
+
"loss": 2.423,
|
| 354 |
+
"step": 5500
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.53,
|
| 358 |
+
"learning_rate": 5.564283316549178e-05,
|
| 359 |
+
"loss": 2.4084,
|
| 360 |
+
"step": 5600
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.54,
|
| 364 |
+
"learning_rate": 5.5525344075193014e-05,
|
| 365 |
+
"loss": 2.3811,
|
| 366 |
+
"step": 5700
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.55,
|
| 370 |
+
"learning_rate": 5.5407854984894264e-05,
|
| 371 |
+
"loss": 2.4017,
|
| 372 |
+
"step": 5800
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 0.56,
|
| 376 |
+
"learning_rate": 5.52903658945955e-05,
|
| 377 |
+
"loss": 2.3276,
|
| 378 |
+
"step": 5900
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.57,
|
| 382 |
+
"learning_rate": 5.517287680429674e-05,
|
| 383 |
+
"loss": 2.4217,
|
| 384 |
+
"step": 6000
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.58,
|
| 388 |
+
"learning_rate": 5.5055387713997986e-05,
|
| 389 |
+
"loss": 2.3477,
|
| 390 |
+
"step": 6100
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 0.59,
|
| 394 |
+
"learning_rate": 5.493789862369923e-05,
|
| 395 |
+
"loss": 2.3782,
|
| 396 |
+
"step": 6200
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.59,
|
| 400 |
+
"learning_rate": 5.4820409533400465e-05,
|
| 401 |
+
"loss": 2.3795,
|
| 402 |
+
"step": 6300
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.6,
|
| 406 |
+
"learning_rate": 5.4702920443101714e-05,
|
| 407 |
+
"loss": 2.3955,
|
| 408 |
+
"step": 6400
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.61,
|
| 412 |
+
"learning_rate": 5.458543135280295e-05,
|
| 413 |
+
"loss": 2.3809,
|
| 414 |
+
"step": 6500
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.62,
|
| 418 |
+
"learning_rate": 5.446794226250419e-05,
|
| 419 |
+
"loss": 2.3147,
|
| 420 |
+
"step": 6600
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.63,
|
| 424 |
+
"learning_rate": 5.435045317220544e-05,
|
| 425 |
+
"loss": 2.4039,
|
| 426 |
+
"step": 6700
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.64,
|
| 430 |
+
"learning_rate": 5.423296408190668e-05,
|
| 431 |
+
"loss": 2.3404,
|
| 432 |
+
"step": 6800
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 0.65,
|
| 436 |
+
"learning_rate": 5.411547499160792e-05,
|
| 437 |
+
"loss": 2.2851,
|
| 438 |
+
"step": 6900
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.66,
|
| 442 |
+
"learning_rate": 5.3997985901309164e-05,
|
| 443 |
+
"loss": 2.3282,
|
| 444 |
+
"step": 7000
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.67,
|
| 448 |
+
"learning_rate": 5.388049681101041e-05,
|
| 449 |
+
"loss": 2.3535,
|
| 450 |
+
"step": 7100
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.68,
|
| 454 |
+
"learning_rate": 5.376300772071164e-05,
|
| 455 |
+
"loss": 2.359,
|
| 456 |
+
"step": 7200
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 0.69,
|
| 460 |
+
"learning_rate": 5.364551863041289e-05,
|
| 461 |
+
"loss": 2.3116,
|
| 462 |
+
"step": 7300
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 0.7,
|
| 466 |
+
"learning_rate": 5.352802954011413e-05,
|
| 467 |
+
"loss": 2.3784,
|
| 468 |
+
"step": 7400
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 0.71,
|
| 472 |
+
"learning_rate": 5.341054044981537e-05,
|
| 473 |
+
"loss": 2.3451,
|
| 474 |
+
"step": 7500
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.72,
|
| 478 |
+
"learning_rate": 5.3293051359516615e-05,
|
| 479 |
+
"loss": 2.3174,
|
| 480 |
+
"step": 7600
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 0.73,
|
| 484 |
+
"learning_rate": 5.317556226921786e-05,
|
| 485 |
+
"loss": 2.3198,
|
| 486 |
+
"step": 7700
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.74,
|
| 490 |
+
"learning_rate": 5.3058073178919094e-05,
|
| 491 |
+
"loss": 2.3014,
|
| 492 |
+
"step": 7800
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.75,
|
| 496 |
+
"learning_rate": 5.294058408862034e-05,
|
| 497 |
+
"loss": 2.3443,
|
| 498 |
+
"step": 7900
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 0.76,
|
| 502 |
+
"learning_rate": 5.2823094998321586e-05,
|
| 503 |
+
"loss": 2.3317,
|
| 504 |
+
"step": 8000
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.76,
|
| 508 |
+
"learning_rate": 5.270560590802282e-05,
|
| 509 |
+
"loss": 2.2974,
|
| 510 |
+
"step": 8100
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 0.77,
|
| 514 |
+
"learning_rate": 5.258811681772407e-05,
|
| 515 |
+
"loss": 2.3403,
|
| 516 |
+
"step": 8200
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 0.78,
|
| 520 |
+
"learning_rate": 5.247062772742531e-05,
|
| 521 |
+
"loss": 2.3128,
|
| 522 |
+
"step": 8300
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.79,
|
| 526 |
+
"learning_rate": 5.235313863712655e-05,
|
| 527 |
+
"loss": 2.3456,
|
| 528 |
+
"step": 8400
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.8,
|
| 532 |
+
"learning_rate": 5.2235649546827793e-05,
|
| 533 |
+
"loss": 2.3053,
|
| 534 |
+
"step": 8500
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.81,
|
| 538 |
+
"learning_rate": 5.2118160456529036e-05,
|
| 539 |
+
"loss": 2.3246,
|
| 540 |
+
"step": 8600
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 0.82,
|
| 544 |
+
"learning_rate": 5.200067136623028e-05,
|
| 545 |
+
"loss": 2.3298,
|
| 546 |
+
"step": 8700
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 0.83,
|
| 550 |
+
"learning_rate": 5.188318227593152e-05,
|
| 551 |
+
"loss": 2.3137,
|
| 552 |
+
"step": 8800
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.84,
|
| 556 |
+
"learning_rate": 5.176569318563276e-05,
|
| 557 |
+
"loss": 2.2685,
|
| 558 |
+
"step": 8900
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 0.85,
|
| 562 |
+
"learning_rate": 5.164820409533401e-05,
|
| 563 |
+
"loss": 2.3339,
|
| 564 |
+
"step": 9000
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.86,
|
| 568 |
+
"learning_rate": 5.1530715005035244e-05,
|
| 569 |
+
"loss": 2.2938,
|
| 570 |
+
"step": 9100
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.87,
|
| 574 |
+
"learning_rate": 5.1413225914736487e-05,
|
| 575 |
+
"loss": 2.2663,
|
| 576 |
+
"step": 9200
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.88,
|
| 580 |
+
"learning_rate": 5.1295736824437736e-05,
|
| 581 |
+
"loss": 2.275,
|
| 582 |
+
"step": 9300
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.89,
|
| 586 |
+
"learning_rate": 5.117824773413897e-05,
|
| 587 |
+
"loss": 2.286,
|
| 588 |
+
"step": 9400
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 0.9,
|
| 592 |
+
"learning_rate": 5.1060758643840215e-05,
|
| 593 |
+
"loss": 2.3234,
|
| 594 |
+
"step": 9500
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.91,
|
| 598 |
+
"learning_rate": 5.094326955354146e-05,
|
| 599 |
+
"loss": 2.3009,
|
| 600 |
+
"step": 9600
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 0.92,
|
| 604 |
+
"learning_rate": 5.08257804632427e-05,
|
| 605 |
+
"loss": 2.1689,
|
| 606 |
+
"step": 9700
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 0.93,
|
| 610 |
+
"learning_rate": 5.070829137294394e-05,
|
| 611 |
+
"loss": 2.2665,
|
| 612 |
+
"step": 9800
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.93,
|
| 616 |
+
"learning_rate": 5.0590802282645186e-05,
|
| 617 |
+
"loss": 2.2566,
|
| 618 |
+
"step": 9900
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.94,
|
| 622 |
+
"learning_rate": 5.047331319234642e-05,
|
| 623 |
+
"loss": 2.2338,
|
| 624 |
+
"step": 10000
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.94,
|
| 628 |
+
"eval_bleu1": 26.4886,
|
| 629 |
+
"eval_bleu2": 15.0125,
|
| 630 |
+
"eval_bleu3": 8.5792,
|
| 631 |
+
"eval_bleu4": 4.8523,
|
| 632 |
+
"eval_gen_len": 41.1049,
|
| 633 |
+
"eval_loss": 2.8001697063446045,
|
| 634 |
+
"eval_rouge1": 33.2148,
|
| 635 |
+
"eval_rouge2": 11.5121,
|
| 636 |
+
"eval_rougeL": 22.7066,
|
| 637 |
+
"eval_runtime": 101.3032,
|
| 638 |
+
"eval_samples_per_second": 4.235,
|
| 639 |
+
"eval_steps_per_second": 0.039,
|
| 640 |
+
"step": 10000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.95,
|
| 644 |
+
"learning_rate": 5.0355824102047665e-05,
|
| 645 |
+
"loss": 2.2267,
|
| 646 |
+
"step": 10100
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.96,
|
| 650 |
+
"learning_rate": 5.023833501174891e-05,
|
| 651 |
+
"loss": 2.198,
|
| 652 |
+
"step": 10200
|
| 653 |
+
},
|
| 654 |
+
{
|
| 655 |
+
"epoch": 0.97,
|
| 656 |
+
"learning_rate": 5.012084592145015e-05,
|
| 657 |
+
"loss": 2.2451,
|
| 658 |
+
"step": 10300
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 0.98,
|
| 662 |
+
"learning_rate": 5.000335683115139e-05,
|
| 663 |
+
"loss": 2.2533,
|
| 664 |
+
"step": 10400
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 0.99,
|
| 668 |
+
"learning_rate": 4.988586774085264e-05,
|
| 669 |
+
"loss": 2.3167,
|
| 670 |
+
"step": 10500
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 1.0,
|
| 674 |
+
"learning_rate": 4.976837865055387e-05,
|
| 675 |
+
"loss": 2.2111,
|
| 676 |
+
"step": 10600
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 1.01,
|
| 680 |
+
"learning_rate": 4.9650889560255116e-05,
|
| 681 |
+
"loss": 2.0181,
|
| 682 |
+
"step": 10700
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 1.02,
|
| 686 |
+
"learning_rate": 4.9533400469956365e-05,
|
| 687 |
+
"loss": 1.9661,
|
| 688 |
+
"step": 10800
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 1.03,
|
| 692 |
+
"learning_rate": 4.94159113796576e-05,
|
| 693 |
+
"loss": 1.9449,
|
| 694 |
+
"step": 10900
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 1.04,
|
| 698 |
+
"learning_rate": 4.9298422289358844e-05,
|
| 699 |
+
"loss": 2.0047,
|
| 700 |
+
"step": 11000
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 1.05,
|
| 704 |
+
"learning_rate": 4.918093319906009e-05,
|
| 705 |
+
"loss": 1.9197,
|
| 706 |
+
"step": 11100
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.06,
|
| 710 |
+
"learning_rate": 4.906344410876133e-05,
|
| 711 |
+
"loss": 1.9867,
|
| 712 |
+
"step": 11200
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 1.07,
|
| 716 |
+
"learning_rate": 4.8945955018462566e-05,
|
| 717 |
+
"loss": 2.0066,
|
| 718 |
+
"step": 11300
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 1.08,
|
| 722 |
+
"learning_rate": 4.8828465928163816e-05,
|
| 723 |
+
"loss": 2.0198,
|
| 724 |
+
"step": 11400
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 1.09,
|
| 728 |
+
"learning_rate": 4.871097683786505e-05,
|
| 729 |
+
"loss": 1.9711,
|
| 730 |
+
"step": 11500
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.1,
|
| 734 |
+
"learning_rate": 4.8593487747566294e-05,
|
| 735 |
+
"loss": 1.9725,
|
| 736 |
+
"step": 11600
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 1.1,
|
| 740 |
+
"learning_rate": 4.847599865726754e-05,
|
| 741 |
+
"loss": 2.032,
|
| 742 |
+
"step": 11700
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 1.11,
|
| 746 |
+
"learning_rate": 4.835850956696878e-05,
|
| 747 |
+
"loss": 1.9697,
|
| 748 |
+
"step": 11800
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 1.12,
|
| 752 |
+
"learning_rate": 4.824102047667002e-05,
|
| 753 |
+
"loss": 1.9865,
|
| 754 |
+
"step": 11900
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 1.13,
|
| 758 |
+
"learning_rate": 4.8123531386371266e-05,
|
| 759 |
+
"loss": 2.0073,
|
| 760 |
+
"step": 12000
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 1.14,
|
| 764 |
+
"learning_rate": 4.800604229607251e-05,
|
| 765 |
+
"loss": 1.9924,
|
| 766 |
+
"step": 12100
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.15,
|
| 770 |
+
"learning_rate": 4.788855320577375e-05,
|
| 771 |
+
"loss": 1.9583,
|
| 772 |
+
"step": 12200
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 1.16,
|
| 776 |
+
"learning_rate": 4.7771064115474994e-05,
|
| 777 |
+
"loss": 1.9727,
|
| 778 |
+
"step": 12300
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 1.17,
|
| 782 |
+
"learning_rate": 4.765357502517623e-05,
|
| 783 |
+
"loss": 1.9346,
|
| 784 |
+
"step": 12400
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 1.18,
|
| 788 |
+
"learning_rate": 4.753608593487748e-05,
|
| 789 |
+
"loss": 1.9879,
|
| 790 |
+
"step": 12500
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 1.19,
|
| 794 |
+
"learning_rate": 4.7418596844578716e-05,
|
| 795 |
+
"loss": 2.0059,
|
| 796 |
+
"step": 12600
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 1.2,
|
| 800 |
+
"learning_rate": 4.730110775427996e-05,
|
| 801 |
+
"loss": 1.9787,
|
| 802 |
+
"step": 12700
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.21,
|
| 806 |
+
"learning_rate": 4.71836186639812e-05,
|
| 807 |
+
"loss": 2.0469,
|
| 808 |
+
"step": 12800
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 1.22,
|
| 812 |
+
"learning_rate": 4.7066129573682445e-05,
|
| 813 |
+
"loss": 1.9941,
|
| 814 |
+
"step": 12900
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 1.23,
|
| 818 |
+
"learning_rate": 4.694864048338368e-05,
|
| 819 |
+
"loss": 1.9872,
|
| 820 |
+
"step": 13000
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 1.24,
|
| 824 |
+
"learning_rate": 4.683115139308493e-05,
|
| 825 |
+
"loss": 1.9744,
|
| 826 |
+
"step": 13100
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 1.25,
|
| 830 |
+
"learning_rate": 4.6713662302786166e-05,
|
| 831 |
+
"loss": 1.924,
|
| 832 |
+
"step": 13200
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 1.26,
|
| 836 |
+
"learning_rate": 4.659617321248741e-05,
|
| 837 |
+
"loss": 1.9875,
|
| 838 |
+
"step": 13300
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.27,
|
| 842 |
+
"learning_rate": 4.647868412218866e-05,
|
| 843 |
+
"loss": 1.9952,
|
| 844 |
+
"step": 13400
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"epoch": 1.27,
|
| 848 |
+
"learning_rate": 4.6361195031889895e-05,
|
| 849 |
+
"loss": 2.0532,
|
| 850 |
+
"step": 13500
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 1.28,
|
| 854 |
+
"learning_rate": 4.624370594159114e-05,
|
| 855 |
+
"loss": 2.0257,
|
| 856 |
+
"step": 13600
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 1.29,
|
| 860 |
+
"learning_rate": 4.612621685129238e-05,
|
| 861 |
+
"loss": 1.9944,
|
| 862 |
+
"step": 13700
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 1.3,
|
| 866 |
+
"learning_rate": 4.6008727760993623e-05,
|
| 867 |
+
"loss": 1.9851,
|
| 868 |
+
"step": 13800
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 1.31,
|
| 872 |
+
"learning_rate": 4.589123867069486e-05,
|
| 873 |
+
"loss": 1.9465,
|
| 874 |
+
"step": 13900
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.32,
|
| 878 |
+
"learning_rate": 4.577374958039611e-05,
|
| 879 |
+
"loss": 1.9918,
|
| 880 |
+
"step": 14000
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"epoch": 1.33,
|
| 884 |
+
"learning_rate": 4.5656260490097345e-05,
|
| 885 |
+
"loss": 1.9753,
|
| 886 |
+
"step": 14100
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"epoch": 1.34,
|
| 890 |
+
"learning_rate": 4.553877139979859e-05,
|
| 891 |
+
"loss": 2.0087,
|
| 892 |
+
"step": 14200
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 1.35,
|
| 896 |
+
"learning_rate": 4.542128230949983e-05,
|
| 897 |
+
"loss": 2.0146,
|
| 898 |
+
"step": 14300
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 1.36,
|
| 902 |
+
"learning_rate": 4.5303793219201074e-05,
|
| 903 |
+
"loss": 1.9441,
|
| 904 |
+
"step": 14400
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.37,
|
| 908 |
+
"learning_rate": 4.518630412890231e-05,
|
| 909 |
+
"loss": 1.9914,
|
| 910 |
+
"step": 14500
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 1.38,
|
| 914 |
+
"learning_rate": 4.506881503860356e-05,
|
| 915 |
+
"loss": 2.0245,
|
| 916 |
+
"step": 14600
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 1.39,
|
| 920 |
+
"learning_rate": 4.49513259483048e-05,
|
| 921 |
+
"loss": 1.9698,
|
| 922 |
+
"step": 14700
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 1.4,
|
| 926 |
+
"learning_rate": 4.483383685800604e-05,
|
| 927 |
+
"loss": 1.9747,
|
| 928 |
+
"step": 14800
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 1.41,
|
| 932 |
+
"learning_rate": 4.471634776770729e-05,
|
| 933 |
+
"loss": 2.0027,
|
| 934 |
+
"step": 14900
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 1.42,
|
| 938 |
+
"learning_rate": 4.4598858677408524e-05,
|
| 939 |
+
"loss": 1.9652,
|
| 940 |
+
"step": 15000
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 1.42,
|
| 944 |
+
"eval_bleu1": 28.2628,
|
| 945 |
+
"eval_bleu2": 16.0909,
|
| 946 |
+
"eval_bleu3": 9.0427,
|
| 947 |
+
"eval_bleu4": 4.9254,
|
| 948 |
+
"eval_gen_len": 46.9744,
|
| 949 |
+
"eval_loss": 2.7699384689331055,
|
| 950 |
+
"eval_rouge1": 34.4269,
|
| 951 |
+
"eval_rouge2": 11.8551,
|
| 952 |
+
"eval_rougeL": 22.8478,
|
| 953 |
+
"eval_runtime": 110.644,
|
| 954 |
+
"eval_samples_per_second": 3.877,
|
| 955 |
+
"eval_steps_per_second": 0.036,
|
| 956 |
+
"step": 15000
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 1.43,
|
| 960 |
+
"learning_rate": 4.448136958710977e-05,
|
| 961 |
+
"loss": 2.015,
|
| 962 |
+
"step": 15100
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 1.44,
|
| 966 |
+
"learning_rate": 4.436388049681101e-05,
|
| 967 |
+
"loss": 1.9913,
|
| 968 |
+
"step": 15200
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 1.44,
|
| 972 |
+
"learning_rate": 4.424639140651225e-05,
|
| 973 |
+
"loss": 1.9891,
|
| 974 |
+
"step": 15300
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 1.45,
|
| 978 |
+
"learning_rate": 4.4128902316213495e-05,
|
| 979 |
+
"loss": 2.0327,
|
| 980 |
+
"step": 15400
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 1.46,
|
| 984 |
+
"learning_rate": 4.401141322591474e-05,
|
| 985 |
+
"loss": 1.9824,
|
| 986 |
+
"step": 15500
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 1.47,
|
| 990 |
+
"learning_rate": 4.3893924135615974e-05,
|
| 991 |
+
"loss": 2.0338,
|
| 992 |
+
"step": 15600
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 1.48,
|
| 996 |
+
"learning_rate": 4.3776435045317224e-05,
|
| 997 |
+
"loss": 2.015,
|
| 998 |
+
"step": 15700
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 1.49,
|
| 1002 |
+
"learning_rate": 4.365894595501846e-05,
|
| 1003 |
+
"loss": 2.0183,
|
| 1004 |
+
"step": 15800
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 1.5,
|
| 1008 |
+
"learning_rate": 4.35414568647197e-05,
|
| 1009 |
+
"loss": 1.9903,
|
| 1010 |
+
"step": 15900
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 1.51,
|
| 1014 |
+
"learning_rate": 4.3423967774420946e-05,
|
| 1015 |
+
"loss": 2.0171,
|
| 1016 |
+
"step": 16000
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 1.52,
|
| 1020 |
+
"learning_rate": 4.330647868412219e-05,
|
| 1021 |
+
"loss": 2.0266,
|
| 1022 |
+
"step": 16100
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 1.53,
|
| 1026 |
+
"learning_rate": 4.318898959382343e-05,
|
| 1027 |
+
"loss": 1.9222,
|
| 1028 |
+
"step": 16200
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 1.54,
|
| 1032 |
+
"learning_rate": 4.3071500503524674e-05,
|
| 1033 |
+
"loss": 1.9535,
|
| 1034 |
+
"step": 16300
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 1.55,
|
| 1038 |
+
"learning_rate": 4.295401141322592e-05,
|
| 1039 |
+
"loss": 2.009,
|
| 1040 |
+
"step": 16400
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 1.56,
|
| 1044 |
+
"learning_rate": 4.283652232292715e-05,
|
| 1045 |
+
"loss": 1.9915,
|
| 1046 |
+
"step": 16500
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 1.57,
|
| 1050 |
+
"learning_rate": 4.27190332326284e-05,
|
| 1051 |
+
"loss": 1.9631,
|
| 1052 |
+
"step": 16600
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 1.58,
|
| 1056 |
+
"learning_rate": 4.260154414232964e-05,
|
| 1057 |
+
"loss": 2.0534,
|
| 1058 |
+
"step": 16700
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.59,
|
| 1062 |
+
"learning_rate": 4.248405505203088e-05,
|
| 1063 |
+
"loss": 1.9991,
|
| 1064 |
+
"step": 16800
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 1.6,
|
| 1068 |
+
"learning_rate": 4.2366565961732124e-05,
|
| 1069 |
+
"loss": 1.9622,
|
| 1070 |
+
"step": 16900
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 1.6,
|
| 1074 |
+
"learning_rate": 4.224907687143337e-05,
|
| 1075 |
+
"loss": 2.0058,
|
| 1076 |
+
"step": 17000
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 1.61,
|
| 1080 |
+
"learning_rate": 4.21315877811346e-05,
|
| 1081 |
+
"loss": 1.9901,
|
| 1082 |
+
"step": 17100
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 1.62,
|
| 1086 |
+
"learning_rate": 4.201409869083585e-05,
|
| 1087 |
+
"loss": 1.9951,
|
| 1088 |
+
"step": 17200
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 1.63,
|
| 1092 |
+
"learning_rate": 4.189660960053709e-05,
|
| 1093 |
+
"loss": 1.9505,
|
| 1094 |
+
"step": 17300
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 1.64,
|
| 1098 |
+
"learning_rate": 4.177912051023833e-05,
|
| 1099 |
+
"loss": 1.9171,
|
| 1100 |
+
"step": 17400
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 1.65,
|
| 1104 |
+
"learning_rate": 4.166163141993958e-05,
|
| 1105 |
+
"loss": 2.0443,
|
| 1106 |
+
"step": 17500
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 1.66,
|
| 1110 |
+
"learning_rate": 4.154414232964082e-05,
|
| 1111 |
+
"loss": 2.0238,
|
| 1112 |
+
"step": 17600
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 1.67,
|
| 1116 |
+
"learning_rate": 4.142665323934206e-05,
|
| 1117 |
+
"loss": 1.9011,
|
| 1118 |
+
"step": 17700
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 1.68,
|
| 1122 |
+
"learning_rate": 4.13091641490433e-05,
|
| 1123 |
+
"loss": 2.0077,
|
| 1124 |
+
"step": 17800
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 1.69,
|
| 1128 |
+
"learning_rate": 4.1191675058744546e-05,
|
| 1129 |
+
"loss": 1.9958,
|
| 1130 |
+
"step": 17900
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 1.7,
|
| 1134 |
+
"learning_rate": 4.107418596844578e-05,
|
| 1135 |
+
"loss": 1.9903,
|
| 1136 |
+
"step": 18000
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 1.71,
|
| 1140 |
+
"learning_rate": 4.095669687814703e-05,
|
| 1141 |
+
"loss": 1.9198,
|
| 1142 |
+
"step": 18100
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 1.72,
|
| 1146 |
+
"learning_rate": 4.083920778784827e-05,
|
| 1147 |
+
"loss": 1.9204,
|
| 1148 |
+
"step": 18200
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 1.73,
|
| 1152 |
+
"learning_rate": 4.072171869754951e-05,
|
| 1153 |
+
"loss": 1.955,
|
| 1154 |
+
"step": 18300
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 1.74,
|
| 1158 |
+
"learning_rate": 4.0604229607250753e-05,
|
| 1159 |
+
"loss": 2.0038,
|
| 1160 |
+
"step": 18400
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 1.75,
|
| 1164 |
+
"learning_rate": 4.0486740516951996e-05,
|
| 1165 |
+
"loss": 2.0018,
|
| 1166 |
+
"step": 18500
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 1.76,
|
| 1170 |
+
"learning_rate": 4.036925142665324e-05,
|
| 1171 |
+
"loss": 2.0406,
|
| 1172 |
+
"step": 18600
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 1.77,
|
| 1176 |
+
"learning_rate": 4.025176233635448e-05,
|
| 1177 |
+
"loss": 1.9958,
|
| 1178 |
+
"step": 18700
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 1.77,
|
| 1182 |
+
"learning_rate": 4.0134273246055725e-05,
|
| 1183 |
+
"loss": 1.951,
|
| 1184 |
+
"step": 18800
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 1.78,
|
| 1188 |
+
"learning_rate": 4.001678415575697e-05,
|
| 1189 |
+
"loss": 1.9721,
|
| 1190 |
+
"step": 18900
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 1.79,
|
| 1194 |
+
"learning_rate": 3.989929506545821e-05,
|
| 1195 |
+
"loss": 1.9323,
|
| 1196 |
+
"step": 19000
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 1.8,
|
| 1200 |
+
"learning_rate": 3.9781805975159447e-05,
|
| 1201 |
+
"loss": 1.9773,
|
| 1202 |
+
"step": 19100
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 1.81,
|
| 1206 |
+
"learning_rate": 3.9664316884860696e-05,
|
| 1207 |
+
"loss": 2.0322,
|
| 1208 |
+
"step": 19200
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 1.82,
|
| 1212 |
+
"learning_rate": 3.954682779456193e-05,
|
| 1213 |
+
"loss": 1.9171,
|
| 1214 |
+
"step": 19300
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 1.83,
|
| 1218 |
+
"learning_rate": 3.9429338704263175e-05,
|
| 1219 |
+
"loss": 1.9805,
|
| 1220 |
+
"step": 19400
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 1.84,
|
| 1224 |
+
"learning_rate": 3.931184961396442e-05,
|
| 1225 |
+
"loss": 1.9344,
|
| 1226 |
+
"step": 19500
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 1.85,
|
| 1230 |
+
"learning_rate": 3.919436052366566e-05,
|
| 1231 |
+
"loss": 1.9163,
|
| 1232 |
+
"step": 19600
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 1.86,
|
| 1236 |
+
"learning_rate": 3.90768714333669e-05,
|
| 1237 |
+
"loss": 1.9857,
|
| 1238 |
+
"step": 19700
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 1.87,
|
| 1242 |
+
"learning_rate": 3.8959382343068146e-05,
|
| 1243 |
+
"loss": 1.9088,
|
| 1244 |
+
"step": 19800
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 1.88,
|
| 1248 |
+
"learning_rate": 3.884189325276938e-05,
|
| 1249 |
+
"loss": 1.9217,
|
| 1250 |
+
"step": 19900
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 1.89,
|
| 1254 |
+
"learning_rate": 3.8724404162470625e-05,
|
| 1255 |
+
"loss": 2.001,
|
| 1256 |
+
"step": 20000
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 1.89,
|
| 1260 |
+
"eval_bleu1": 28.3593,
|
| 1261 |
+
"eval_bleu2": 16.1361,
|
| 1262 |
+
"eval_bleu3": 9.221,
|
| 1263 |
+
"eval_bleu4": 4.8616,
|
| 1264 |
+
"eval_gen_len": 46.979,
|
| 1265 |
+
"eval_loss": 2.720076322555542,
|
| 1266 |
+
"eval_rouge1": 34.157,
|
| 1267 |
+
"eval_rouge2": 11.8683,
|
| 1268 |
+
"eval_rougeL": 22.6775,
|
| 1269 |
+
"eval_runtime": 113.1379,
|
| 1270 |
+
"eval_samples_per_second": 3.792,
|
| 1271 |
+
"eval_steps_per_second": 0.035,
|
| 1272 |
+
"step": 20000
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 1.9,
|
| 1276 |
+
"learning_rate": 3.8606915072171875e-05,
|
| 1277 |
+
"loss": 1.9878,
|
| 1278 |
+
"step": 20100
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 1.91,
|
| 1282 |
+
"learning_rate": 3.848942598187311e-05,
|
| 1283 |
+
"loss": 1.9526,
|
| 1284 |
+
"step": 20200
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 1.92,
|
| 1288 |
+
"learning_rate": 3.8371936891574354e-05,
|
| 1289 |
+
"loss": 1.9721,
|
| 1290 |
+
"step": 20300
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 1.93,
|
| 1294 |
+
"learning_rate": 3.82544478012756e-05,
|
| 1295 |
+
"loss": 2.0391,
|
| 1296 |
+
"step": 20400
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 1.94,
|
| 1300 |
+
"learning_rate": 3.813695871097684e-05,
|
| 1301 |
+
"loss": 1.9484,
|
| 1302 |
+
"step": 20500
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 1.94,
|
| 1306 |
+
"learning_rate": 3.8019469620678076e-05,
|
| 1307 |
+
"loss": 1.9701,
|
| 1308 |
+
"step": 20600
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 1.95,
|
| 1312 |
+
"learning_rate": 3.7901980530379325e-05,
|
| 1313 |
+
"loss": 2.015,
|
| 1314 |
+
"step": 20700
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 1.96,
|
| 1318 |
+
"learning_rate": 3.778449144008056e-05,
|
| 1319 |
+
"loss": 1.9754,
|
| 1320 |
+
"step": 20800
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 1.97,
|
| 1324 |
+
"learning_rate": 3.7667002349781804e-05,
|
| 1325 |
+
"loss": 1.9644,
|
| 1326 |
+
"step": 20900
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 1.98,
|
| 1330 |
+
"learning_rate": 3.754951325948305e-05,
|
| 1331 |
+
"loss": 1.9766,
|
| 1332 |
+
"step": 21000
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 1.99,
|
| 1336 |
+
"learning_rate": 3.743202416918429e-05,
|
| 1337 |
+
"loss": 1.9929,
|
| 1338 |
+
"step": 21100
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 2.0,
|
| 1342 |
+
"learning_rate": 3.7314535078885526e-05,
|
| 1343 |
+
"loss": 1.937,
|
| 1344 |
+
"step": 21200
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 2.01,
|
| 1348 |
+
"learning_rate": 3.7197045988586775e-05,
|
| 1349 |
+
"loss": 1.5585,
|
| 1350 |
+
"step": 21300
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 2.02,
|
| 1354 |
+
"learning_rate": 3.707955689828801e-05,
|
| 1355 |
+
"loss": 1.6113,
|
| 1356 |
+
"step": 21400
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 2.03,
|
| 1360 |
+
"learning_rate": 3.6962067807989254e-05,
|
| 1361 |
+
"loss": 1.5737,
|
| 1362 |
+
"step": 21500
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 2.04,
|
| 1366 |
+
"learning_rate": 3.6844578717690504e-05,
|
| 1367 |
+
"loss": 1.6545,
|
| 1368 |
+
"step": 21600
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 2.05,
|
| 1372 |
+
"learning_rate": 3.672708962739174e-05,
|
| 1373 |
+
"loss": 1.6024,
|
| 1374 |
+
"step": 21700
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 2.06,
|
| 1378 |
+
"learning_rate": 3.660960053709298e-05,
|
| 1379 |
+
"loss": 1.6431,
|
| 1380 |
+
"step": 21800
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 2.07,
|
| 1384 |
+
"learning_rate": 3.6492111446794226e-05,
|
| 1385 |
+
"loss": 1.5974,
|
| 1386 |
+
"step": 21900
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 2.08,
|
| 1390 |
+
"learning_rate": 3.637462235649547e-05,
|
| 1391 |
+
"loss": 1.6111,
|
| 1392 |
+
"step": 22000
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 2.09,
|
| 1396 |
+
"learning_rate": 3.625713326619671e-05,
|
| 1397 |
+
"loss": 1.6215,
|
| 1398 |
+
"step": 22100
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 2.1,
|
| 1402 |
+
"learning_rate": 3.6139644175897954e-05,
|
| 1403 |
+
"loss": 1.5874,
|
| 1404 |
+
"step": 22200
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 2.11,
|
| 1408 |
+
"learning_rate": 3.602215508559919e-05,
|
| 1409 |
+
"loss": 1.5834,
|
| 1410 |
+
"step": 22300
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 2.11,
|
| 1414 |
+
"learning_rate": 3.590466599530044e-05,
|
| 1415 |
+
"loss": 1.6405,
|
| 1416 |
+
"step": 22400
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 2.12,
|
| 1420 |
+
"learning_rate": 3.5787176905001676e-05,
|
| 1421 |
+
"loss": 1.5939,
|
| 1422 |
+
"step": 22500
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 2.13,
|
| 1426 |
+
"learning_rate": 3.566968781470292e-05,
|
| 1427 |
+
"loss": 1.6135,
|
| 1428 |
+
"step": 22600
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 2.14,
|
| 1432 |
+
"learning_rate": 3.555219872440416e-05,
|
| 1433 |
+
"loss": 1.6091,
|
| 1434 |
+
"step": 22700
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 2.15,
|
| 1438 |
+
"learning_rate": 3.5434709634105405e-05,
|
| 1439 |
+
"loss": 1.6102,
|
| 1440 |
+
"step": 22800
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 2.16,
|
| 1444 |
+
"learning_rate": 3.531722054380665e-05,
|
| 1445 |
+
"loss": 1.6148,
|
| 1446 |
+
"step": 22900
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 2.17,
|
| 1450 |
+
"learning_rate": 3.519973145350789e-05,
|
| 1451 |
+
"loss": 1.6351,
|
| 1452 |
+
"step": 23000
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 2.18,
|
| 1456 |
+
"learning_rate": 3.508224236320913e-05,
|
| 1457 |
+
"loss": 1.5941,
|
| 1458 |
+
"step": 23100
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 2.19,
|
| 1462 |
+
"learning_rate": 3.496475327291037e-05,
|
| 1463 |
+
"loss": 1.6088,
|
| 1464 |
+
"step": 23200
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 2.2,
|
| 1468 |
+
"learning_rate": 3.484726418261162e-05,
|
| 1469 |
+
"loss": 1.6255,
|
| 1470 |
+
"step": 23300
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 2.21,
|
| 1474 |
+
"learning_rate": 3.4729775092312855e-05,
|
| 1475 |
+
"loss": 1.671,
|
| 1476 |
+
"step": 23400
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 2.22,
|
| 1480 |
+
"learning_rate": 3.46122860020141e-05,
|
| 1481 |
+
"loss": 1.6096,
|
| 1482 |
+
"step": 23500
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 2.23,
|
| 1486 |
+
"learning_rate": 3.449479691171534e-05,
|
| 1487 |
+
"loss": 1.6399,
|
| 1488 |
+
"step": 23600
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 2.24,
|
| 1492 |
+
"learning_rate": 3.437730782141658e-05,
|
| 1493 |
+
"loss": 1.5641,
|
| 1494 |
+
"step": 23700
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 2.25,
|
| 1498 |
+
"learning_rate": 3.425981873111782e-05,
|
| 1499 |
+
"loss": 1.6423,
|
| 1500 |
+
"step": 23800
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 2.26,
|
| 1504 |
+
"learning_rate": 3.414232964081907e-05,
|
| 1505 |
+
"loss": 1.5897,
|
| 1506 |
+
"step": 23900
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 2.27,
|
| 1510 |
+
"learning_rate": 3.4024840550520305e-05,
|
| 1511 |
+
"loss": 1.6378,
|
| 1512 |
+
"step": 24000
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"epoch": 2.28,
|
| 1516 |
+
"learning_rate": 3.390735146022155e-05,
|
| 1517 |
+
"loss": 1.6512,
|
| 1518 |
+
"step": 24100
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 2.28,
|
| 1522 |
+
"learning_rate": 3.37898623699228e-05,
|
| 1523 |
+
"loss": 1.5813,
|
| 1524 |
+
"step": 24200
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 2.29,
|
| 1528 |
+
"learning_rate": 3.3672373279624034e-05,
|
| 1529 |
+
"loss": 1.6065,
|
| 1530 |
+
"step": 24300
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 2.3,
|
| 1534 |
+
"learning_rate": 3.3554884189325276e-05,
|
| 1535 |
+
"loss": 1.6269,
|
| 1536 |
+
"step": 24400
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 2.31,
|
| 1540 |
+
"learning_rate": 3.343739509902652e-05,
|
| 1541 |
+
"loss": 1.6715,
|
| 1542 |
+
"step": 24500
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 2.32,
|
| 1546 |
+
"learning_rate": 3.331990600872776e-05,
|
| 1547 |
+
"loss": 1.645,
|
| 1548 |
+
"step": 24600
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 2.33,
|
| 1552 |
+
"learning_rate": 3.3202416918429e-05,
|
| 1553 |
+
"loss": 1.592,
|
| 1554 |
+
"step": 24700
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 2.34,
|
| 1558 |
+
"learning_rate": 3.308492782813025e-05,
|
| 1559 |
+
"loss": 1.6165,
|
| 1560 |
+
"step": 24800
|
| 1561 |
+
},
|
| 1562 |
+
{
|
| 1563 |
+
"epoch": 2.35,
|
| 1564 |
+
"learning_rate": 3.2967438737831484e-05,
|
| 1565 |
+
"loss": 1.6097,
|
| 1566 |
+
"step": 24900
|
| 1567 |
+
},
|
| 1568 |
+
{
|
| 1569 |
+
"epoch": 2.36,
|
| 1570 |
+
"learning_rate": 3.284994964753273e-05,
|
| 1571 |
+
"loss": 1.6433,
|
| 1572 |
+
"step": 25000
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 2.36,
|
| 1576 |
+
"eval_bleu1": 27.6475,
|
| 1577 |
+
"eval_bleu2": 15.6571,
|
| 1578 |
+
"eval_bleu3": 8.8372,
|
| 1579 |
+
"eval_bleu4": 4.8672,
|
| 1580 |
+
"eval_gen_len": 43.9953,
|
| 1581 |
+
"eval_loss": 2.7900760173797607,
|
| 1582 |
+
"eval_rouge1": 33.6354,
|
| 1583 |
+
"eval_rouge2": 11.5761,
|
| 1584 |
+
"eval_rougeL": 22.6878,
|
| 1585 |
+
"eval_runtime": 88.3325,
|
| 1586 |
+
"eval_samples_per_second": 4.857,
|
| 1587 |
+
"eval_steps_per_second": 0.045,
|
| 1588 |
+
"step": 25000
|
| 1589 |
+
},
|
| 1590 |
+
{
|
| 1591 |
+
"epoch": 2.37,
|
| 1592 |
+
"learning_rate": 3.273246055723397e-05,
|
| 1593 |
+
"loss": 1.616,
|
| 1594 |
+
"step": 25100
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 2.38,
|
| 1598 |
+
"learning_rate": 3.261497146693521e-05,
|
| 1599 |
+
"loss": 1.5917,
|
| 1600 |
+
"step": 25200
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"epoch": 2.39,
|
| 1604 |
+
"learning_rate": 3.2497482376636455e-05,
|
| 1605 |
+
"loss": 1.5815,
|
| 1606 |
+
"step": 25300
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 2.4,
|
| 1610 |
+
"learning_rate": 3.23799932863377e-05,
|
| 1611 |
+
"loss": 1.5837,
|
| 1612 |
+
"step": 25400
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 2.41,
|
| 1616 |
+
"learning_rate": 3.226250419603894e-05,
|
| 1617 |
+
"loss": 1.6637,
|
| 1618 |
+
"step": 25500
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 2.42,
|
| 1622 |
+
"learning_rate": 3.2145015105740184e-05,
|
| 1623 |
+
"loss": 1.6237,
|
| 1624 |
+
"step": 25600
|
| 1625 |
+
},
|
| 1626 |
+
{
|
| 1627 |
+
"epoch": 2.43,
|
| 1628 |
+
"learning_rate": 3.2027526015441427e-05,
|
| 1629 |
+
"loss": 1.629,
|
| 1630 |
+
"step": 25700
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 2.44,
|
| 1634 |
+
"learning_rate": 3.191003692514266e-05,
|
| 1635 |
+
"loss": 1.6329,
|
| 1636 |
+
"step": 25800
|
| 1637 |
+
},
|
| 1638 |
+
{
|
| 1639 |
+
"epoch": 2.45,
|
| 1640 |
+
"learning_rate": 3.179254783484391e-05,
|
| 1641 |
+
"loss": 1.6019,
|
| 1642 |
+
"step": 25900
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 2.45,
|
| 1646 |
+
"learning_rate": 3.167505874454515e-05,
|
| 1647 |
+
"loss": 1.6123,
|
| 1648 |
+
"step": 26000
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 2.46,
|
| 1652 |
+
"learning_rate": 3.155756965424639e-05,
|
| 1653 |
+
"loss": 1.7134,
|
| 1654 |
+
"step": 26100
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 2.47,
|
| 1658 |
+
"learning_rate": 3.1440080563947634e-05,
|
| 1659 |
+
"loss": 1.6092,
|
| 1660 |
+
"step": 26200
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 2.48,
|
| 1664 |
+
"learning_rate": 3.132259147364888e-05,
|
| 1665 |
+
"loss": 1.6641,
|
| 1666 |
+
"step": 26300
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 2.49,
|
| 1670 |
+
"learning_rate": 3.120510238335011e-05,
|
| 1671 |
+
"loss": 1.6283,
|
| 1672 |
+
"step": 26400
|
| 1673 |
+
},
|
| 1674 |
+
{
|
| 1675 |
+
"epoch": 2.5,
|
| 1676 |
+
"learning_rate": 3.108761329305136e-05,
|
| 1677 |
+
"loss": 1.5878,
|
| 1678 |
+
"step": 26500
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 2.51,
|
| 1682 |
+
"learning_rate": 3.09701242027526e-05,
|
| 1683 |
+
"loss": 1.602,
|
| 1684 |
+
"step": 26600
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 2.52,
|
| 1688 |
+
"learning_rate": 3.085263511245384e-05,
|
| 1689 |
+
"loss": 1.6025,
|
| 1690 |
+
"step": 26700
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 2.53,
|
| 1694 |
+
"learning_rate": 3.0735146022155084e-05,
|
| 1695 |
+
"loss": 1.6628,
|
| 1696 |
+
"step": 26800
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 2.54,
|
| 1700 |
+
"learning_rate": 3.061765693185633e-05,
|
| 1701 |
+
"loss": 1.6723,
|
| 1702 |
+
"step": 26900
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 2.55,
|
| 1706 |
+
"learning_rate": 3.0500167841557573e-05,
|
| 1707 |
+
"loss": 1.6593,
|
| 1708 |
+
"step": 27000
|
| 1709 |
+
},
|
| 1710 |
+
{
|
| 1711 |
+
"epoch": 2.56,
|
| 1712 |
+
"learning_rate": 3.038267875125881e-05,
|
| 1713 |
+
"loss": 1.6143,
|
| 1714 |
+
"step": 27100
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"epoch": 2.57,
|
| 1718 |
+
"learning_rate": 3.0265189660960056e-05,
|
| 1719 |
+
"loss": 1.6176,
|
| 1720 |
+
"step": 27200
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 2.58,
|
| 1724 |
+
"learning_rate": 3.0147700570661295e-05,
|
| 1725 |
+
"loss": 1.6353,
|
| 1726 |
+
"step": 27300
|
| 1727 |
+
},
|
| 1728 |
+
{
|
| 1729 |
+
"epoch": 2.59,
|
| 1730 |
+
"learning_rate": 3.0030211480362538e-05,
|
| 1731 |
+
"loss": 1.6185,
|
| 1732 |
+
"step": 27400
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 2.6,
|
| 1736 |
+
"learning_rate": 2.9912722390063777e-05,
|
| 1737 |
+
"loss": 1.6467,
|
| 1738 |
+
"step": 27500
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 2.61,
|
| 1742 |
+
"learning_rate": 2.9795233299765024e-05,
|
| 1743 |
+
"loss": 1.6083,
|
| 1744 |
+
"step": 27600
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"epoch": 2.62,
|
| 1748 |
+
"learning_rate": 2.967774420946626e-05,
|
| 1749 |
+
"loss": 1.701,
|
| 1750 |
+
"step": 27700
|
| 1751 |
+
},
|
| 1752 |
+
{
|
| 1753 |
+
"epoch": 2.62,
|
| 1754 |
+
"learning_rate": 2.9560255119167506e-05,
|
| 1755 |
+
"loss": 1.6665,
|
| 1756 |
+
"step": 27800
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 2.63,
|
| 1760 |
+
"learning_rate": 2.9442766028868745e-05,
|
| 1761 |
+
"loss": 1.5699,
|
| 1762 |
+
"step": 27900
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 2.64,
|
| 1766 |
+
"learning_rate": 2.9325276938569988e-05,
|
| 1767 |
+
"loss": 1.7012,
|
| 1768 |
+
"step": 28000
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"epoch": 2.65,
|
| 1772 |
+
"learning_rate": 2.9207787848271228e-05,
|
| 1773 |
+
"loss": 1.6608,
|
| 1774 |
+
"step": 28100
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 2.66,
|
| 1778 |
+
"learning_rate": 2.9090298757972474e-05,
|
| 1779 |
+
"loss": 1.6536,
|
| 1780 |
+
"step": 28200
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 2.67,
|
| 1784 |
+
"learning_rate": 2.8972809667673717e-05,
|
| 1785 |
+
"loss": 1.6572,
|
| 1786 |
+
"step": 28300
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 2.68,
|
| 1790 |
+
"learning_rate": 2.8855320577374956e-05,
|
| 1791 |
+
"loss": 1.6291,
|
| 1792 |
+
"step": 28400
|
| 1793 |
+
},
|
| 1794 |
+
{
|
| 1795 |
+
"epoch": 2.69,
|
| 1796 |
+
"learning_rate": 2.8737831487076202e-05,
|
| 1797 |
+
"loss": 1.6501,
|
| 1798 |
+
"step": 28500
|
| 1799 |
+
},
|
| 1800 |
+
{
|
| 1801 |
+
"epoch": 2.7,
|
| 1802 |
+
"learning_rate": 2.8620342396777442e-05,
|
| 1803 |
+
"loss": 1.6456,
|
| 1804 |
+
"step": 28600
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 2.71,
|
| 1808 |
+
"learning_rate": 2.8502853306478685e-05,
|
| 1809 |
+
"loss": 1.6701,
|
| 1810 |
+
"step": 28700
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 2.72,
|
| 1814 |
+
"learning_rate": 2.8385364216179924e-05,
|
| 1815 |
+
"loss": 1.6873,
|
| 1816 |
+
"step": 28800
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 2.73,
|
| 1820 |
+
"learning_rate": 2.826787512588117e-05,
|
| 1821 |
+
"loss": 1.5811,
|
| 1822 |
+
"step": 28900
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 2.74,
|
| 1826 |
+
"learning_rate": 2.8150386035582406e-05,
|
| 1827 |
+
"loss": 1.6128,
|
| 1828 |
+
"step": 29000
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 2.75,
|
| 1832 |
+
"learning_rate": 2.8032896945283653e-05,
|
| 1833 |
+
"loss": 1.653,
|
| 1834 |
+
"step": 29100
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 2.76,
|
| 1838 |
+
"learning_rate": 2.7915407854984896e-05,
|
| 1839 |
+
"loss": 1.6867,
|
| 1840 |
+
"step": 29200
|
| 1841 |
+
},
|
| 1842 |
+
{
|
| 1843 |
+
"epoch": 2.77,
|
| 1844 |
+
"learning_rate": 2.7797918764686135e-05,
|
| 1845 |
+
"loss": 1.6462,
|
| 1846 |
+
"step": 29300
|
| 1847 |
+
},
|
| 1848 |
+
{
|
| 1849 |
+
"epoch": 2.78,
|
| 1850 |
+
"learning_rate": 2.7680429674387378e-05,
|
| 1851 |
+
"loss": 1.5985,
|
| 1852 |
+
"step": 29400
|
| 1853 |
+
},
|
| 1854 |
+
{
|
| 1855 |
+
"epoch": 2.79,
|
| 1856 |
+
"learning_rate": 2.756294058408862e-05,
|
| 1857 |
+
"loss": 1.6274,
|
| 1858 |
+
"step": 29500
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 2.79,
|
| 1862 |
+
"learning_rate": 2.744545149378986e-05,
|
| 1863 |
+
"loss": 1.6512,
|
| 1864 |
+
"step": 29600
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 2.8,
|
| 1868 |
+
"learning_rate": 2.7327962403491103e-05,
|
| 1869 |
+
"loss": 1.6,
|
| 1870 |
+
"step": 29700
|
| 1871 |
+
},
|
| 1872 |
+
{
|
| 1873 |
+
"epoch": 2.81,
|
| 1874 |
+
"learning_rate": 2.7210473313192346e-05,
|
| 1875 |
+
"loss": 1.7005,
|
| 1876 |
+
"step": 29800
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 2.82,
|
| 1880 |
+
"learning_rate": 2.709298422289359e-05,
|
| 1881 |
+
"loss": 1.6796,
|
| 1882 |
+
"step": 29900
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 2.83,
|
| 1886 |
+
"learning_rate": 2.6975495132594828e-05,
|
| 1887 |
+
"loss": 1.6204,
|
| 1888 |
+
"step": 30000
|
| 1889 |
+
},
|
| 1890 |
+
{
|
| 1891 |
+
"epoch": 2.83,
|
| 1892 |
+
"eval_bleu1": 29.1014,
|
| 1893 |
+
"eval_bleu2": 16.6689,
|
| 1894 |
+
"eval_bleu3": 9.3661,
|
| 1895 |
+
"eval_bleu4": 5.1916,
|
| 1896 |
+
"eval_gen_len": 48.8811,
|
| 1897 |
+
"eval_loss": 2.7724153995513916,
|
| 1898 |
+
"eval_rouge1": 34.9611,
|
| 1899 |
+
"eval_rouge2": 12.1606,
|
| 1900 |
+
"eval_rougeL": 23.0246,
|
| 1901 |
+
"eval_runtime": 137.6409,
|
| 1902 |
+
"eval_samples_per_second": 3.117,
|
| 1903 |
+
"eval_steps_per_second": 0.029,
|
| 1904 |
+
"step": 30000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 2.84,
|
| 1908 |
+
"learning_rate": 2.685800604229607e-05,
|
| 1909 |
+
"loss": 1.6336,
|
| 1910 |
+
"step": 30100
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 2.85,
|
| 1914 |
+
"learning_rate": 2.6740516951997317e-05,
|
| 1915 |
+
"loss": 1.6349,
|
| 1916 |
+
"step": 30200
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 2.86,
|
| 1920 |
+
"learning_rate": 2.6623027861698557e-05,
|
| 1921 |
+
"loss": 1.5879,
|
| 1922 |
+
"step": 30300
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 2.87,
|
| 1926 |
+
"learning_rate": 2.65055387713998e-05,
|
| 1927 |
+
"loss": 1.6111,
|
| 1928 |
+
"step": 30400
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 2.88,
|
| 1932 |
+
"learning_rate": 2.6388049681101042e-05,
|
| 1933 |
+
"loss": 1.6168,
|
| 1934 |
+
"step": 30500
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 2.89,
|
| 1938 |
+
"learning_rate": 2.6270560590802282e-05,
|
| 1939 |
+
"loss": 1.6425,
|
| 1940 |
+
"step": 30600
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 2.9,
|
| 1944 |
+
"learning_rate": 2.6153071500503525e-05,
|
| 1945 |
+
"loss": 1.6359,
|
| 1946 |
+
"step": 30700
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 2.91,
|
| 1950 |
+
"learning_rate": 2.6035582410204767e-05,
|
| 1951 |
+
"loss": 1.6269,
|
| 1952 |
+
"step": 30800
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 2.92,
|
| 1956 |
+
"learning_rate": 2.5918093319906007e-05,
|
| 1957 |
+
"loss": 1.6668,
|
| 1958 |
+
"step": 30900
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 2.93,
|
| 1962 |
+
"learning_rate": 2.580060422960725e-05,
|
| 1963 |
+
"loss": 1.6364,
|
| 1964 |
+
"step": 31000
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 2.94,
|
| 1968 |
+
"learning_rate": 2.5683115139308493e-05,
|
| 1969 |
+
"loss": 1.5957,
|
| 1970 |
+
"step": 31100
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 2.95,
|
| 1974 |
+
"learning_rate": 2.5565626049009732e-05,
|
| 1975 |
+
"loss": 1.5837,
|
| 1976 |
+
"step": 31200
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 2.96,
|
| 1980 |
+
"learning_rate": 2.5448136958710975e-05,
|
| 1981 |
+
"loss": 1.6382,
|
| 1982 |
+
"step": 31300
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 2.96,
|
| 1986 |
+
"learning_rate": 2.5330647868412218e-05,
|
| 1987 |
+
"loss": 1.6077,
|
| 1988 |
+
"step": 31400
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 2.97,
|
| 1992 |
+
"learning_rate": 2.521315877811346e-05,
|
| 1993 |
+
"loss": 1.6355,
|
| 1994 |
+
"step": 31500
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 2.98,
|
| 1998 |
+
"learning_rate": 2.5095669687814703e-05,
|
| 1999 |
+
"loss": 1.6633,
|
| 2000 |
+
"step": 31600
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 2.99,
|
| 2004 |
+
"learning_rate": 2.4978180597515946e-05,
|
| 2005 |
+
"loss": 1.5906,
|
| 2006 |
+
"step": 31700
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 3.0,
|
| 2010 |
+
"learning_rate": 2.486069150721719e-05,
|
| 2011 |
+
"loss": 1.5347,
|
| 2012 |
+
"step": 31800
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 3.01,
|
| 2016 |
+
"learning_rate": 2.474320241691843e-05,
|
| 2017 |
+
"loss": 1.2498,
|
| 2018 |
+
"step": 31900
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 3.02,
|
| 2022 |
+
"learning_rate": 2.462571332661967e-05,
|
| 2023 |
+
"loss": 1.3091,
|
| 2024 |
+
"step": 32000
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 3.03,
|
| 2028 |
+
"learning_rate": 2.4508224236320914e-05,
|
| 2029 |
+
"loss": 1.3252,
|
| 2030 |
+
"step": 32100
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 3.04,
|
| 2034 |
+
"learning_rate": 2.4390735146022154e-05,
|
| 2035 |
+
"loss": 1.2999,
|
| 2036 |
+
"step": 32200
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 3.05,
|
| 2040 |
+
"learning_rate": 2.4273246055723396e-05,
|
| 2041 |
+
"loss": 1.288,
|
| 2042 |
+
"step": 32300
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 3.06,
|
| 2046 |
+
"learning_rate": 2.415575696542464e-05,
|
| 2047 |
+
"loss": 1.3072,
|
| 2048 |
+
"step": 32400
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 3.07,
|
| 2052 |
+
"learning_rate": 2.403826787512588e-05,
|
| 2053 |
+
"loss": 1.2572,
|
| 2054 |
+
"step": 32500
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 3.08,
|
| 2058 |
+
"learning_rate": 2.392077878482712e-05,
|
| 2059 |
+
"loss": 1.3139,
|
| 2060 |
+
"step": 32600
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 3.09,
|
| 2064 |
+
"learning_rate": 2.3803289694528364e-05,
|
| 2065 |
+
"loss": 1.344,
|
| 2066 |
+
"step": 32700
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 3.1,
|
| 2070 |
+
"learning_rate": 2.3685800604229604e-05,
|
| 2071 |
+
"loss": 1.3242,
|
| 2072 |
+
"step": 32800
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 3.11,
|
| 2076 |
+
"learning_rate": 2.356831151393085e-05,
|
| 2077 |
+
"loss": 1.2924,
|
| 2078 |
+
"step": 32900
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 3.12,
|
| 2082 |
+
"learning_rate": 2.3450822423632093e-05,
|
| 2083 |
+
"loss": 1.2989,
|
| 2084 |
+
"step": 33000
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 3.12,
|
| 2088 |
+
"learning_rate": 2.3333333333333332e-05,
|
| 2089 |
+
"loss": 1.2983,
|
| 2090 |
+
"step": 33100
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 3.13,
|
| 2094 |
+
"learning_rate": 2.3215844243034575e-05,
|
| 2095 |
+
"loss": 1.2902,
|
| 2096 |
+
"step": 33200
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 3.14,
|
| 2100 |
+
"learning_rate": 2.3098355152735818e-05,
|
| 2101 |
+
"loss": 1.2479,
|
| 2102 |
+
"step": 33300
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 3.15,
|
| 2106 |
+
"learning_rate": 2.298086606243706e-05,
|
| 2107 |
+
"loss": 1.2555,
|
| 2108 |
+
"step": 33400
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 3.16,
|
| 2112 |
+
"learning_rate": 2.28633769721383e-05,
|
| 2113 |
+
"loss": 1.2613,
|
| 2114 |
+
"step": 33500
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 3.17,
|
| 2118 |
+
"learning_rate": 2.2745887881839543e-05,
|
| 2119 |
+
"loss": 1.2782,
|
| 2120 |
+
"step": 33600
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 3.18,
|
| 2124 |
+
"learning_rate": 2.2628398791540786e-05,
|
| 2125 |
+
"loss": 1.2686,
|
| 2126 |
+
"step": 33700
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 3.19,
|
| 2130 |
+
"learning_rate": 2.2510909701242026e-05,
|
| 2131 |
+
"loss": 1.3088,
|
| 2132 |
+
"step": 33800
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 3.2,
|
| 2136 |
+
"learning_rate": 2.239342061094327e-05,
|
| 2137 |
+
"loss": 1.3223,
|
| 2138 |
+
"step": 33900
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 3.21,
|
| 2142 |
+
"learning_rate": 2.227593152064451e-05,
|
| 2143 |
+
"loss": 1.3449,
|
| 2144 |
+
"step": 34000
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 3.22,
|
| 2148 |
+
"learning_rate": 2.215844243034575e-05,
|
| 2149 |
+
"loss": 1.3136,
|
| 2150 |
+
"step": 34100
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 3.23,
|
| 2154 |
+
"learning_rate": 2.2040953340046994e-05,
|
| 2155 |
+
"loss": 1.2793,
|
| 2156 |
+
"step": 34200
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 3.24,
|
| 2160 |
+
"learning_rate": 2.192346424974824e-05,
|
| 2161 |
+
"loss": 1.3015,
|
| 2162 |
+
"step": 34300
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 3.25,
|
| 2166 |
+
"learning_rate": 2.180597515944948e-05,
|
| 2167 |
+
"loss": 1.2883,
|
| 2168 |
+
"step": 34400
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 3.26,
|
| 2172 |
+
"learning_rate": 2.1688486069150722e-05,
|
| 2173 |
+
"loss": 1.3133,
|
| 2174 |
+
"step": 34500
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 3.27,
|
| 2178 |
+
"learning_rate": 2.1570996978851965e-05,
|
| 2179 |
+
"loss": 1.3283,
|
| 2180 |
+
"step": 34600
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 3.28,
|
| 2184 |
+
"learning_rate": 2.1453507888553204e-05,
|
| 2185 |
+
"loss": 1.267,
|
| 2186 |
+
"step": 34700
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 3.29,
|
| 2190 |
+
"learning_rate": 2.1336018798254447e-05,
|
| 2191 |
+
"loss": 1.2919,
|
| 2192 |
+
"step": 34800
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 3.29,
|
| 2196 |
+
"learning_rate": 2.121852970795569e-05,
|
| 2197 |
+
"loss": 1.3182,
|
| 2198 |
+
"step": 34900
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 3.3,
|
| 2202 |
+
"learning_rate": 2.1101040617656933e-05,
|
| 2203 |
+
"loss": 1.2955,
|
| 2204 |
+
"step": 35000
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 3.3,
|
| 2208 |
+
"eval_bleu1": 29.9701,
|
| 2209 |
+
"eval_bleu2": 17.3963,
|
| 2210 |
+
"eval_bleu3": 10.2978,
|
| 2211 |
+
"eval_bleu4": 5.9339,
|
| 2212 |
+
"eval_gen_len": 49.5921,
|
| 2213 |
+
"eval_loss": 2.8970282077789307,
|
| 2214 |
+
"eval_rouge1": 35.896,
|
| 2215 |
+
"eval_rouge2": 12.7037,
|
| 2216 |
+
"eval_rougeL": 23.3781,
|
| 2217 |
+
"eval_runtime": 133.8073,
|
| 2218 |
+
"eval_samples_per_second": 3.206,
|
| 2219 |
+
"eval_steps_per_second": 0.03,
|
| 2220 |
+
"step": 35000
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 3.31,
|
| 2224 |
+
"learning_rate": 2.0983551527358172e-05,
|
| 2225 |
+
"loss": 1.3059,
|
| 2226 |
+
"step": 35100
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 3.32,
|
| 2230 |
+
"learning_rate": 2.0866062437059415e-05,
|
| 2231 |
+
"loss": 1.2751,
|
| 2232 |
+
"step": 35200
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 3.33,
|
| 2236 |
+
"learning_rate": 2.0748573346760658e-05,
|
| 2237 |
+
"loss": 1.3223,
|
| 2238 |
+
"step": 35300
|
| 2239 |
+
},
|
| 2240 |
+
{
|
| 2241 |
+
"epoch": 3.34,
|
| 2242 |
+
"learning_rate": 2.0631084256461897e-05,
|
| 2243 |
+
"loss": 1.2946,
|
| 2244 |
+
"step": 35400
|
| 2245 |
+
},
|
| 2246 |
+
{
|
| 2247 |
+
"epoch": 3.35,
|
| 2248 |
+
"learning_rate": 2.051359516616314e-05,
|
| 2249 |
+
"loss": 1.3418,
|
| 2250 |
+
"step": 35500
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 3.36,
|
| 2254 |
+
"learning_rate": 2.0396106075864386e-05,
|
| 2255 |
+
"loss": 1.3068,
|
| 2256 |
+
"step": 35600
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 3.37,
|
| 2260 |
+
"learning_rate": 2.0278616985565626e-05,
|
| 2261 |
+
"loss": 1.3108,
|
| 2262 |
+
"step": 35700
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 3.38,
|
| 2266 |
+
"learning_rate": 2.016112789526687e-05,
|
| 2267 |
+
"loss": 1.3569,
|
| 2268 |
+
"step": 35800
|
| 2269 |
+
},
|
| 2270 |
+
{
|
| 2271 |
+
"epoch": 3.39,
|
| 2272 |
+
"learning_rate": 2.004363880496811e-05,
|
| 2273 |
+
"loss": 1.338,
|
| 2274 |
+
"step": 35900
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 3.4,
|
| 2278 |
+
"learning_rate": 1.992614971466935e-05,
|
| 2279 |
+
"loss": 1.2918,
|
| 2280 |
+
"step": 36000
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"epoch": 3.41,
|
| 2284 |
+
"learning_rate": 1.9808660624370594e-05,
|
| 2285 |
+
"loss": 1.3495,
|
| 2286 |
+
"step": 36100
|
| 2287 |
+
},
|
| 2288 |
+
{
|
| 2289 |
+
"epoch": 3.42,
|
| 2290 |
+
"learning_rate": 1.9691171534071837e-05,
|
| 2291 |
+
"loss": 1.2948,
|
| 2292 |
+
"step": 36200
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 3.43,
|
| 2296 |
+
"learning_rate": 1.9573682443773076e-05,
|
| 2297 |
+
"loss": 1.3311,
|
| 2298 |
+
"step": 36300
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 3.44,
|
| 2302 |
+
"learning_rate": 1.945619335347432e-05,
|
| 2303 |
+
"loss": 1.3683,
|
| 2304 |
+
"step": 36400
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 3.45,
|
| 2308 |
+
"learning_rate": 1.9338704263175562e-05,
|
| 2309 |
+
"loss": 1.2935,
|
| 2310 |
+
"step": 36500
|
| 2311 |
+
},
|
| 2312 |
+
{
|
| 2313 |
+
"epoch": 3.46,
|
| 2314 |
+
"learning_rate": 1.9221215172876805e-05,
|
| 2315 |
+
"loss": 1.3249,
|
| 2316 |
+
"step": 36600
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 3.46,
|
| 2320 |
+
"learning_rate": 1.9103726082578044e-05,
|
| 2321 |
+
"loss": 1.3301,
|
| 2322 |
+
"step": 36700
|
| 2323 |
+
},
|
| 2324 |
+
{
|
| 2325 |
+
"epoch": 3.47,
|
| 2326 |
+
"learning_rate": 1.8986236992279287e-05,
|
| 2327 |
+
"loss": 1.2982,
|
| 2328 |
+
"step": 36800
|
| 2329 |
+
},
|
| 2330 |
+
{
|
| 2331 |
+
"epoch": 3.48,
|
| 2332 |
+
"learning_rate": 1.886874790198053e-05,
|
| 2333 |
+
"loss": 1.3244,
|
| 2334 |
+
"step": 36900
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 3.49,
|
| 2338 |
+
"learning_rate": 1.8751258811681773e-05,
|
| 2339 |
+
"loss": 1.286,
|
| 2340 |
+
"step": 37000
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 3.5,
|
| 2344 |
+
"learning_rate": 1.8633769721383016e-05,
|
| 2345 |
+
"loss": 1.343,
|
| 2346 |
+
"step": 37100
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 3.51,
|
| 2350 |
+
"learning_rate": 1.851628063108426e-05,
|
| 2351 |
+
"loss": 1.3298,
|
| 2352 |
+
"step": 37200
|
| 2353 |
+
},
|
| 2354 |
+
{
|
| 2355 |
+
"epoch": 3.52,
|
| 2356 |
+
"learning_rate": 1.8398791540785498e-05,
|
| 2357 |
+
"loss": 1.3453,
|
| 2358 |
+
"step": 37300
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 3.53,
|
| 2362 |
+
"learning_rate": 1.828130245048674e-05,
|
| 2363 |
+
"loss": 1.344,
|
| 2364 |
+
"step": 37400
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 3.54,
|
| 2368 |
+
"learning_rate": 1.8163813360187984e-05,
|
| 2369 |
+
"loss": 1.3198,
|
| 2370 |
+
"step": 37500
|
| 2371 |
+
},
|
| 2372 |
+
{
|
| 2373 |
+
"epoch": 3.55,
|
| 2374 |
+
"learning_rate": 1.8046324269889223e-05,
|
| 2375 |
+
"loss": 1.3162,
|
| 2376 |
+
"step": 37600
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 3.56,
|
| 2380 |
+
"learning_rate": 1.7928835179590466e-05,
|
| 2381 |
+
"loss": 1.3382,
|
| 2382 |
+
"step": 37700
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 3.57,
|
| 2386 |
+
"learning_rate": 1.781134608929171e-05,
|
| 2387 |
+
"loss": 1.3602,
|
| 2388 |
+
"step": 37800
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 3.58,
|
| 2392 |
+
"learning_rate": 1.7693856998992948e-05,
|
| 2393 |
+
"loss": 1.2834,
|
| 2394 |
+
"step": 37900
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 3.59,
|
| 2398 |
+
"learning_rate": 1.757636790869419e-05,
|
| 2399 |
+
"loss": 1.3467,
|
| 2400 |
+
"step": 38000
|
| 2401 |
+
},
|
| 2402 |
+
{
|
| 2403 |
+
"epoch": 3.6,
|
| 2404 |
+
"learning_rate": 1.7458878818395434e-05,
|
| 2405 |
+
"loss": 1.3505,
|
| 2406 |
+
"step": 38100
|
| 2407 |
+
},
|
| 2408 |
+
{
|
| 2409 |
+
"epoch": 3.61,
|
| 2410 |
+
"learning_rate": 1.7341389728096677e-05,
|
| 2411 |
+
"loss": 1.3005,
|
| 2412 |
+
"step": 38200
|
| 2413 |
+
},
|
| 2414 |
+
{
|
| 2415 |
+
"epoch": 3.62,
|
| 2416 |
+
"learning_rate": 1.722390063779792e-05,
|
| 2417 |
+
"loss": 1.3155,
|
| 2418 |
+
"step": 38300
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 3.63,
|
| 2422 |
+
"learning_rate": 1.7106411547499162e-05,
|
| 2423 |
+
"loss": 1.3164,
|
| 2424 |
+
"step": 38400
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 3.63,
|
| 2428 |
+
"learning_rate": 1.6988922457200405e-05,
|
| 2429 |
+
"loss": 1.2915,
|
| 2430 |
+
"step": 38500
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 3.64,
|
| 2434 |
+
"learning_rate": 1.6871433366901645e-05,
|
| 2435 |
+
"loss": 1.3501,
|
| 2436 |
+
"step": 38600
|
| 2437 |
+
},
|
| 2438 |
+
{
|
| 2439 |
+
"epoch": 3.65,
|
| 2440 |
+
"learning_rate": 1.6753944276602887e-05,
|
| 2441 |
+
"loss": 1.345,
|
| 2442 |
+
"step": 38700
|
| 2443 |
+
},
|
| 2444 |
+
{
|
| 2445 |
+
"epoch": 3.66,
|
| 2446 |
+
"learning_rate": 1.663645518630413e-05,
|
| 2447 |
+
"loss": 1.3215,
|
| 2448 |
+
"step": 38800
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 3.67,
|
| 2452 |
+
"learning_rate": 1.651896609600537e-05,
|
| 2453 |
+
"loss": 1.3088,
|
| 2454 |
+
"step": 38900
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 3.68,
|
| 2458 |
+
"learning_rate": 1.6401477005706613e-05,
|
| 2459 |
+
"loss": 1.266,
|
| 2460 |
+
"step": 39000
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 3.69,
|
| 2464 |
+
"learning_rate": 1.6283987915407855e-05,
|
| 2465 |
+
"loss": 1.3522,
|
| 2466 |
+
"step": 39100
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 3.7,
|
| 2470 |
+
"learning_rate": 1.6166498825109095e-05,
|
| 2471 |
+
"loss": 1.298,
|
| 2472 |
+
"step": 39200
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 3.71,
|
| 2476 |
+
"learning_rate": 1.6049009734810338e-05,
|
| 2477 |
+
"loss": 1.255,
|
| 2478 |
+
"step": 39300
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 3.72,
|
| 2482 |
+
"learning_rate": 1.593152064451158e-05,
|
| 2483 |
+
"loss": 1.2987,
|
| 2484 |
+
"step": 39400
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 3.73,
|
| 2488 |
+
"learning_rate": 1.581403155421282e-05,
|
| 2489 |
+
"loss": 1.3309,
|
| 2490 |
+
"step": 39500
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 3.74,
|
| 2494 |
+
"learning_rate": 1.5696542463914063e-05,
|
| 2495 |
+
"loss": 1.2839,
|
| 2496 |
+
"step": 39600
|
| 2497 |
+
},
|
| 2498 |
+
{
|
| 2499 |
+
"epoch": 3.75,
|
| 2500 |
+
"learning_rate": 1.557905337361531e-05,
|
| 2501 |
+
"loss": 1.3109,
|
| 2502 |
+
"step": 39700
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 3.76,
|
| 2506 |
+
"learning_rate": 1.546156428331655e-05,
|
| 2507 |
+
"loss": 1.3073,
|
| 2508 |
+
"step": 39800
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 3.77,
|
| 2512 |
+
"learning_rate": 1.534407519301779e-05,
|
| 2513 |
+
"loss": 1.339,
|
| 2514 |
+
"step": 39900
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 3.78,
|
| 2518 |
+
"learning_rate": 1.5226586102719034e-05,
|
| 2519 |
+
"loss": 1.3501,
|
| 2520 |
+
"step": 40000
|
| 2521 |
+
},
|
| 2522 |
+
{
|
| 2523 |
+
"epoch": 3.78,
|
| 2524 |
+
"eval_bleu1": 29.483,
|
| 2525 |
+
"eval_bleu2": 16.7795,
|
| 2526 |
+
"eval_bleu3": 9.4124,
|
| 2527 |
+
"eval_bleu4": 5.2042,
|
| 2528 |
+
"eval_gen_len": 48.5897,
|
| 2529 |
+
"eval_loss": 2.8853633403778076,
|
| 2530 |
+
"eval_rouge1": 35.2981,
|
| 2531 |
+
"eval_rouge2": 12.1133,
|
| 2532 |
+
"eval_rougeL": 23.1845,
|
| 2533 |
+
"eval_runtime": 112.9023,
|
| 2534 |
+
"eval_samples_per_second": 3.8,
|
| 2535 |
+
"eval_steps_per_second": 0.035,
|
| 2536 |
+
"step": 40000
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 3.79,
|
| 2540 |
+
"learning_rate": 1.5109097012420275e-05,
|
| 2541 |
+
"loss": 1.2452,
|
| 2542 |
+
"step": 40100
|
| 2543 |
+
},
|
| 2544 |
+
{
|
| 2545 |
+
"epoch": 3.8,
|
| 2546 |
+
"learning_rate": 1.4991607922121517e-05,
|
| 2547 |
+
"loss": 1.3348,
|
| 2548 |
+
"step": 40200
|
| 2549 |
+
},
|
| 2550 |
+
{
|
| 2551 |
+
"epoch": 3.8,
|
| 2552 |
+
"learning_rate": 1.487411883182276e-05,
|
| 2553 |
+
"loss": 1.3156,
|
| 2554 |
+
"step": 40300
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"epoch": 3.81,
|
| 2558 |
+
"learning_rate": 1.4756629741524e-05,
|
| 2559 |
+
"loss": 1.3597,
|
| 2560 |
+
"step": 40400
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"epoch": 3.82,
|
| 2564 |
+
"learning_rate": 1.4639140651225243e-05,
|
| 2565 |
+
"loss": 1.362,
|
| 2566 |
+
"step": 40500
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 3.83,
|
| 2570 |
+
"learning_rate": 1.4521651560926484e-05,
|
| 2571 |
+
"loss": 1.2591,
|
| 2572 |
+
"step": 40600
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 3.84,
|
| 2576 |
+
"learning_rate": 1.4404162470627726e-05,
|
| 2577 |
+
"loss": 1.3309,
|
| 2578 |
+
"step": 40700
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 3.85,
|
| 2582 |
+
"learning_rate": 1.4286673380328968e-05,
|
| 2583 |
+
"loss": 1.3623,
|
| 2584 |
+
"step": 40800
|
| 2585 |
+
},
|
| 2586 |
+
{
|
| 2587 |
+
"epoch": 3.86,
|
| 2588 |
+
"learning_rate": 1.416918429003021e-05,
|
| 2589 |
+
"loss": 1.3032,
|
| 2590 |
+
"step": 40900
|
| 2591 |
+
},
|
| 2592 |
+
{
|
| 2593 |
+
"epoch": 3.87,
|
| 2594 |
+
"learning_rate": 1.4051695199731454e-05,
|
| 2595 |
+
"loss": 1.3075,
|
| 2596 |
+
"step": 41000
|
| 2597 |
+
},
|
| 2598 |
+
{
|
| 2599 |
+
"epoch": 3.88,
|
| 2600 |
+
"learning_rate": 1.3934206109432695e-05,
|
| 2601 |
+
"loss": 1.3549,
|
| 2602 |
+
"step": 41100
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 3.89,
|
| 2606 |
+
"learning_rate": 1.3816717019133936e-05,
|
| 2607 |
+
"loss": 1.3257,
|
| 2608 |
+
"step": 41200
|
| 2609 |
+
},
|
| 2610 |
+
{
|
| 2611 |
+
"epoch": 3.9,
|
| 2612 |
+
"learning_rate": 1.369922792883518e-05,
|
| 2613 |
+
"loss": 1.3262,
|
| 2614 |
+
"step": 41300
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 3.91,
|
| 2618 |
+
"learning_rate": 1.3581738838536422e-05,
|
| 2619 |
+
"loss": 1.3143,
|
| 2620 |
+
"step": 41400
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 3.92,
|
| 2624 |
+
"learning_rate": 1.3464249748237663e-05,
|
| 2625 |
+
"loss": 1.341,
|
| 2626 |
+
"step": 41500
|
| 2627 |
+
},
|
| 2628 |
+
{
|
| 2629 |
+
"epoch": 3.93,
|
| 2630 |
+
"learning_rate": 1.3346760657938906e-05,
|
| 2631 |
+
"loss": 1.3282,
|
| 2632 |
+
"step": 41600
|
| 2633 |
+
},
|
| 2634 |
+
{
|
| 2635 |
+
"epoch": 3.94,
|
| 2636 |
+
"learning_rate": 1.3229271567640147e-05,
|
| 2637 |
+
"loss": 1.3146,
|
| 2638 |
+
"step": 41700
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 3.95,
|
| 2642 |
+
"learning_rate": 1.3111782477341388e-05,
|
| 2643 |
+
"loss": 1.3032,
|
| 2644 |
+
"step": 41800
|
| 2645 |
+
},
|
| 2646 |
+
{
|
| 2647 |
+
"epoch": 3.96,
|
| 2648 |
+
"learning_rate": 1.2994293387042631e-05,
|
| 2649 |
+
"loss": 1.3441,
|
| 2650 |
+
"step": 41900
|
| 2651 |
+
},
|
| 2652 |
+
{
|
| 2653 |
+
"epoch": 3.97,
|
| 2654 |
+
"learning_rate": 1.2876804296743874e-05,
|
| 2655 |
+
"loss": 1.3165,
|
| 2656 |
+
"step": 42000
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 3.97,
|
| 2660 |
+
"learning_rate": 1.2759315206445117e-05,
|
| 2661 |
+
"loss": 1.3309,
|
| 2662 |
+
"step": 42100
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 3.98,
|
| 2666 |
+
"learning_rate": 1.2641826116146358e-05,
|
| 2667 |
+
"loss": 1.3448,
|
| 2668 |
+
"step": 42200
|
| 2669 |
+
},
|
| 2670 |
+
{
|
| 2671 |
+
"epoch": 3.99,
|
| 2672 |
+
"learning_rate": 1.25243370258476e-05,
|
| 2673 |
+
"loss": 1.3181,
|
| 2674 |
+
"step": 42300
|
| 2675 |
+
},
|
| 2676 |
+
{
|
| 2677 |
+
"epoch": 4.0,
|
| 2678 |
+
"learning_rate": 1.2406847935548842e-05,
|
| 2679 |
+
"loss": 1.2496,
|
| 2680 |
+
"step": 42400
|
| 2681 |
+
},
|
| 2682 |
+
{
|
| 2683 |
+
"epoch": 4.01,
|
| 2684 |
+
"learning_rate": 1.2289358845250083e-05,
|
| 2685 |
+
"loss": 1.0748,
|
| 2686 |
+
"step": 42500
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 4.02,
|
| 2690 |
+
"learning_rate": 1.2171869754951324e-05,
|
| 2691 |
+
"loss": 1.0485,
|
| 2692 |
+
"step": 42600
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"epoch": 4.03,
|
| 2696 |
+
"learning_rate": 1.2054380664652569e-05,
|
| 2697 |
+
"loss": 1.0712,
|
| 2698 |
+
"step": 42700
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 4.04,
|
| 2702 |
+
"learning_rate": 1.193689157435381e-05,
|
| 2703 |
+
"loss": 1.0456,
|
| 2704 |
+
"step": 42800
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 4.05,
|
| 2708 |
+
"learning_rate": 1.1819402484055053e-05,
|
| 2709 |
+
"loss": 1.0662,
|
| 2710 |
+
"step": 42900
|
| 2711 |
+
},
|
| 2712 |
+
{
|
| 2713 |
+
"epoch": 4.06,
|
| 2714 |
+
"learning_rate": 1.1701913393756294e-05,
|
| 2715 |
+
"loss": 1.0658,
|
| 2716 |
+
"step": 43000
|
| 2717 |
+
},
|
| 2718 |
+
{
|
| 2719 |
+
"epoch": 4.07,
|
| 2720 |
+
"learning_rate": 1.1584424303457535e-05,
|
| 2721 |
+
"loss": 1.0439,
|
| 2722 |
+
"step": 43100
|
| 2723 |
+
},
|
| 2724 |
+
{
|
| 2725 |
+
"epoch": 4.08,
|
| 2726 |
+
"learning_rate": 1.1466935213158778e-05,
|
| 2727 |
+
"loss": 1.0558,
|
| 2728 |
+
"step": 43200
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 4.09,
|
| 2732 |
+
"learning_rate": 1.134944612286002e-05,
|
| 2733 |
+
"loss": 1.0442,
|
| 2734 |
+
"step": 43300
|
| 2735 |
+
},
|
| 2736 |
+
{
|
| 2737 |
+
"epoch": 4.1,
|
| 2738 |
+
"learning_rate": 1.1231957032561262e-05,
|
| 2739 |
+
"loss": 1.1046,
|
| 2740 |
+
"step": 43400
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 4.11,
|
| 2744 |
+
"learning_rate": 1.1114467942262505e-05,
|
| 2745 |
+
"loss": 1.0772,
|
| 2746 |
+
"step": 43500
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 4.12,
|
| 2750 |
+
"learning_rate": 1.0996978851963746e-05,
|
| 2751 |
+
"loss": 1.0464,
|
| 2752 |
+
"step": 43600
|
| 2753 |
+
},
|
| 2754 |
+
{
|
| 2755 |
+
"epoch": 4.13,
|
| 2756 |
+
"learning_rate": 1.0879489761664989e-05,
|
| 2757 |
+
"loss": 1.049,
|
| 2758 |
+
"step": 43700
|
| 2759 |
+
},
|
| 2760 |
+
{
|
| 2761 |
+
"epoch": 4.14,
|
| 2762 |
+
"learning_rate": 1.076200067136623e-05,
|
| 2763 |
+
"loss": 1.0382,
|
| 2764 |
+
"step": 43800
|
| 2765 |
+
},
|
| 2766 |
+
{
|
| 2767 |
+
"epoch": 4.14,
|
| 2768 |
+
"learning_rate": 1.0644511581067471e-05,
|
| 2769 |
+
"loss": 1.0537,
|
| 2770 |
+
"step": 43900
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 4.15,
|
| 2774 |
+
"learning_rate": 1.0527022490768716e-05,
|
| 2775 |
+
"loss": 1.0735,
|
| 2776 |
+
"step": 44000
|
| 2777 |
+
},
|
| 2778 |
+
{
|
| 2779 |
+
"epoch": 4.16,
|
| 2780 |
+
"learning_rate": 1.0409533400469957e-05,
|
| 2781 |
+
"loss": 1.0729,
|
| 2782 |
+
"step": 44100
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 4.17,
|
| 2786 |
+
"learning_rate": 1.0292044310171198e-05,
|
| 2787 |
+
"loss": 1.0703,
|
| 2788 |
+
"step": 44200
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 4.18,
|
| 2792 |
+
"learning_rate": 1.017455521987244e-05,
|
| 2793 |
+
"loss": 1.0869,
|
| 2794 |
+
"step": 44300
|
| 2795 |
+
},
|
| 2796 |
+
{
|
| 2797 |
+
"epoch": 4.19,
|
| 2798 |
+
"learning_rate": 1.0057066129573682e-05,
|
| 2799 |
+
"loss": 1.0986,
|
| 2800 |
+
"step": 44400
|
| 2801 |
+
},
|
| 2802 |
+
{
|
| 2803 |
+
"epoch": 4.2,
|
| 2804 |
+
"learning_rate": 9.939577039274923e-06,
|
| 2805 |
+
"loss": 1.1168,
|
| 2806 |
+
"step": 44500
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"epoch": 4.21,
|
| 2810 |
+
"learning_rate": 9.822087948976166e-06,
|
| 2811 |
+
"loss": 1.0136,
|
| 2812 |
+
"step": 44600
|
| 2813 |
+
},
|
| 2814 |
+
{
|
| 2815 |
+
"epoch": 4.22,
|
| 2816 |
+
"learning_rate": 9.704598858677409e-06,
|
| 2817 |
+
"loss": 1.0433,
|
| 2818 |
+
"step": 44700
|
| 2819 |
+
},
|
| 2820 |
+
{
|
| 2821 |
+
"epoch": 4.23,
|
| 2822 |
+
"learning_rate": 9.587109768378652e-06,
|
| 2823 |
+
"loss": 1.0302,
|
| 2824 |
+
"step": 44800
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 4.24,
|
| 2828 |
+
"learning_rate": 9.469620678079893e-06,
|
| 2829 |
+
"loss": 1.0834,
|
| 2830 |
+
"step": 44900
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 4.25,
|
| 2834 |
+
"learning_rate": 9.352131587781134e-06,
|
| 2835 |
+
"loss": 1.0865,
|
| 2836 |
+
"step": 45000
|
| 2837 |
+
},
|
| 2838 |
+
{
|
| 2839 |
+
"epoch": 4.25,
|
| 2840 |
+
"eval_bleu1": 29.9364,
|
| 2841 |
+
"eval_bleu2": 17.2064,
|
| 2842 |
+
"eval_bleu3": 10.0427,
|
| 2843 |
+
"eval_bleu4": 5.62,
|
| 2844 |
+
"eval_gen_len": 48.31,
|
| 2845 |
+
"eval_loss": 2.9911699295043945,
|
| 2846 |
+
"eval_rouge1": 35.581,
|
| 2847 |
+
"eval_rouge2": 12.5145,
|
| 2848 |
+
"eval_rougeL": 23.2262,
|
| 2849 |
+
"eval_runtime": 132.7683,
|
| 2850 |
+
"eval_samples_per_second": 3.231,
|
| 2851 |
+
"eval_steps_per_second": 0.03,
|
| 2852 |
+
"step": 45000
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 4.26,
|
| 2856 |
+
"learning_rate": 9.234642497482377e-06,
|
| 2857 |
+
"loss": 1.0786,
|
| 2858 |
+
"step": 45100
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 4.27,
|
| 2862 |
+
"learning_rate": 9.117153407183618e-06,
|
| 2863 |
+
"loss": 1.0742,
|
| 2864 |
+
"step": 45200
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 4.28,
|
| 2868 |
+
"learning_rate": 8.999664316884859e-06,
|
| 2869 |
+
"loss": 1.0674,
|
| 2870 |
+
"step": 45300
|
| 2871 |
+
},
|
| 2872 |
+
{
|
| 2873 |
+
"epoch": 4.29,
|
| 2874 |
+
"learning_rate": 8.882175226586104e-06,
|
| 2875 |
+
"loss": 1.0379,
|
| 2876 |
+
"step": 45400
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 4.3,
|
| 2880 |
+
"learning_rate": 8.764686136287345e-06,
|
| 2881 |
+
"loss": 1.0856,
|
| 2882 |
+
"step": 45500
|
| 2883 |
+
},
|
| 2884 |
+
{
|
| 2885 |
+
"epoch": 4.31,
|
| 2886 |
+
"learning_rate": 8.647197045988588e-06,
|
| 2887 |
+
"loss": 1.0653,
|
| 2888 |
+
"step": 45600
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 4.31,
|
| 2892 |
+
"learning_rate": 8.529707955689829e-06,
|
| 2893 |
+
"loss": 1.0765,
|
| 2894 |
+
"step": 45700
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 4.32,
|
| 2898 |
+
"learning_rate": 8.41221886539107e-06,
|
| 2899 |
+
"loss": 1.0777,
|
| 2900 |
+
"step": 45800
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 4.33,
|
| 2904 |
+
"learning_rate": 8.294729775092313e-06,
|
| 2905 |
+
"loss": 1.0399,
|
| 2906 |
+
"step": 45900
|
| 2907 |
+
},
|
| 2908 |
+
{
|
| 2909 |
+
"epoch": 4.34,
|
| 2910 |
+
"learning_rate": 8.177240684793554e-06,
|
| 2911 |
+
"loss": 1.0091,
|
| 2912 |
+
"step": 46000
|
| 2913 |
+
},
|
| 2914 |
+
{
|
| 2915 |
+
"epoch": 4.35,
|
| 2916 |
+
"learning_rate": 8.059751594494797e-06,
|
| 2917 |
+
"loss": 1.0784,
|
| 2918 |
+
"step": 46100
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 4.36,
|
| 2922 |
+
"learning_rate": 7.94226250419604e-06,
|
| 2923 |
+
"loss": 1.0458,
|
| 2924 |
+
"step": 46200
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 4.37,
|
| 2928 |
+
"learning_rate": 7.82477341389728e-06,
|
| 2929 |
+
"loss": 1.0492,
|
| 2930 |
+
"step": 46300
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 4.38,
|
| 2934 |
+
"learning_rate": 7.707284323598524e-06,
|
| 2935 |
+
"loss": 1.1092,
|
| 2936 |
+
"step": 46400
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 4.39,
|
| 2940 |
+
"learning_rate": 7.589795233299765e-06,
|
| 2941 |
+
"loss": 1.0529,
|
| 2942 |
+
"step": 46500
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 4.4,
|
| 2946 |
+
"learning_rate": 7.472306143001007e-06,
|
| 2947 |
+
"loss": 1.0548,
|
| 2948 |
+
"step": 46600
|
| 2949 |
+
},
|
| 2950 |
+
{
|
| 2951 |
+
"epoch": 4.41,
|
| 2952 |
+
"learning_rate": 7.3548170527022495e-06,
|
| 2953 |
+
"loss": 1.036,
|
| 2954 |
+
"step": 46700
|
| 2955 |
+
},
|
| 2956 |
+
{
|
| 2957 |
+
"epoch": 4.42,
|
| 2958 |
+
"learning_rate": 7.2373279624034915e-06,
|
| 2959 |
+
"loss": 1.0559,
|
| 2960 |
+
"step": 46800
|
| 2961 |
+
},
|
| 2962 |
+
{
|
| 2963 |
+
"epoch": 4.43,
|
| 2964 |
+
"learning_rate": 7.1198388721047335e-06,
|
| 2965 |
+
"loss": 1.0745,
|
| 2966 |
+
"step": 46900
|
| 2967 |
+
},
|
| 2968 |
+
{
|
| 2969 |
+
"epoch": 4.44,
|
| 2970 |
+
"learning_rate": 7.0023497818059755e-06,
|
| 2971 |
+
"loss": 1.0568,
|
| 2972 |
+
"step": 47000
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 4.45,
|
| 2976 |
+
"learning_rate": 6.884860691507217e-06,
|
| 2977 |
+
"loss": 1.092,
|
| 2978 |
+
"step": 47100
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 4.46,
|
| 2982 |
+
"learning_rate": 6.7673716012084595e-06,
|
| 2983 |
+
"loss": 1.0348,
|
| 2984 |
+
"step": 47200
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 4.47,
|
| 2988 |
+
"learning_rate": 6.6498825109097015e-06,
|
| 2989 |
+
"loss": 1.0851,
|
| 2990 |
+
"step": 47300
|
| 2991 |
+
},
|
| 2992 |
+
{
|
| 2993 |
+
"epoch": 4.48,
|
| 2994 |
+
"learning_rate": 6.5323934206109435e-06,
|
| 2995 |
+
"loss": 1.0879,
|
| 2996 |
+
"step": 47400
|
| 2997 |
+
},
|
| 2998 |
+
{
|
| 2999 |
+
"epoch": 4.48,
|
| 3000 |
+
"learning_rate": 6.414904330312185e-06,
|
| 3001 |
+
"loss": 1.0162,
|
| 3002 |
+
"step": 47500
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 4.49,
|
| 3006 |
+
"learning_rate": 6.2974152400134274e-06,
|
| 3007 |
+
"loss": 1.0917,
|
| 3008 |
+
"step": 47600
|
| 3009 |
+
},
|
| 3010 |
+
{
|
| 3011 |
+
"epoch": 4.5,
|
| 3012 |
+
"learning_rate": 6.1799261497146694e-06,
|
| 3013 |
+
"loss": 1.0483,
|
| 3014 |
+
"step": 47700
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 4.51,
|
| 3018 |
+
"learning_rate": 6.0624370594159114e-06,
|
| 3019 |
+
"loss": 1.0914,
|
| 3020 |
+
"step": 47800
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 4.52,
|
| 3024 |
+
"learning_rate": 5.944947969117153e-06,
|
| 3025 |
+
"loss": 1.0819,
|
| 3026 |
+
"step": 47900
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 4.53,
|
| 3030 |
+
"learning_rate": 5.827458878818395e-06,
|
| 3031 |
+
"loss": 1.0699,
|
| 3032 |
+
"step": 48000
|
| 3033 |
+
},
|
| 3034 |
+
{
|
| 3035 |
+
"epoch": 4.54,
|
| 3036 |
+
"learning_rate": 5.709969788519637e-06,
|
| 3037 |
+
"loss": 1.0507,
|
| 3038 |
+
"step": 48100
|
| 3039 |
+
},
|
| 3040 |
+
{
|
| 3041 |
+
"epoch": 4.55,
|
| 3042 |
+
"learning_rate": 5.592480698220879e-06,
|
| 3043 |
+
"loss": 1.1011,
|
| 3044 |
+
"step": 48200
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 4.56,
|
| 3048 |
+
"learning_rate": 5.474991607922121e-06,
|
| 3049 |
+
"loss": 1.0611,
|
| 3050 |
+
"step": 48300
|
| 3051 |
+
},
|
| 3052 |
+
{
|
| 3053 |
+
"epoch": 4.57,
|
| 3054 |
+
"learning_rate": 5.357502517623363e-06,
|
| 3055 |
+
"loss": 1.0708,
|
| 3056 |
+
"step": 48400
|
| 3057 |
+
},
|
| 3058 |
+
{
|
| 3059 |
+
"epoch": 4.58,
|
| 3060 |
+
"learning_rate": 5.240013427324605e-06,
|
| 3061 |
+
"loss": 1.0579,
|
| 3062 |
+
"step": 48500
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 4.59,
|
| 3066 |
+
"learning_rate": 5.122524337025847e-06,
|
| 3067 |
+
"loss": 1.0582,
|
| 3068 |
+
"step": 48600
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 4.6,
|
| 3072 |
+
"learning_rate": 5.005035246727089e-06,
|
| 3073 |
+
"loss": 1.0748,
|
| 3074 |
+
"step": 48700
|
| 3075 |
+
},
|
| 3076 |
+
{
|
| 3077 |
+
"epoch": 4.61,
|
| 3078 |
+
"learning_rate": 4.887546156428331e-06,
|
| 3079 |
+
"loss": 1.0763,
|
| 3080 |
+
"step": 48800
|
| 3081 |
+
},
|
| 3082 |
+
{
|
| 3083 |
+
"epoch": 4.62,
|
| 3084 |
+
"learning_rate": 4.770057066129574e-06,
|
| 3085 |
+
"loss": 1.085,
|
| 3086 |
+
"step": 48900
|
| 3087 |
+
},
|
| 3088 |
+
{
|
| 3089 |
+
"epoch": 4.63,
|
| 3090 |
+
"learning_rate": 4.652567975830815e-06,
|
| 3091 |
+
"loss": 1.059,
|
| 3092 |
+
"step": 49000
|
| 3093 |
+
},
|
| 3094 |
+
{
|
| 3095 |
+
"epoch": 4.64,
|
| 3096 |
+
"learning_rate": 4.535078885532057e-06,
|
| 3097 |
+
"loss": 1.119,
|
| 3098 |
+
"step": 49100
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 4.65,
|
| 3102 |
+
"learning_rate": 4.417589795233299e-06,
|
| 3103 |
+
"loss": 1.0784,
|
| 3104 |
+
"step": 49200
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 4.65,
|
| 3108 |
+
"learning_rate": 4.300100704934542e-06,
|
| 3109 |
+
"loss": 1.0702,
|
| 3110 |
+
"step": 49300
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 4.66,
|
| 3114 |
+
"learning_rate": 4.182611614635783e-06,
|
| 3115 |
+
"loss": 1.0722,
|
| 3116 |
+
"step": 49400
|
| 3117 |
+
},
|
| 3118 |
+
{
|
| 3119 |
+
"epoch": 4.67,
|
| 3120 |
+
"learning_rate": 4.065122524337025e-06,
|
| 3121 |
+
"loss": 1.0578,
|
| 3122 |
+
"step": 49500
|
| 3123 |
+
},
|
| 3124 |
+
{
|
| 3125 |
+
"epoch": 4.68,
|
| 3126 |
+
"learning_rate": 3.947633434038268e-06,
|
| 3127 |
+
"loss": 1.0838,
|
| 3128 |
+
"step": 49600
|
| 3129 |
+
},
|
| 3130 |
+
{
|
| 3131 |
+
"epoch": 4.69,
|
| 3132 |
+
"learning_rate": 3.83014434373951e-06,
|
| 3133 |
+
"loss": 1.0704,
|
| 3134 |
+
"step": 49700
|
| 3135 |
+
},
|
| 3136 |
+
{
|
| 3137 |
+
"epoch": 4.7,
|
| 3138 |
+
"learning_rate": 3.7126552534407517e-06,
|
| 3139 |
+
"loss": 1.0318,
|
| 3140 |
+
"step": 49800
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 4.71,
|
| 3144 |
+
"learning_rate": 3.595166163141994e-06,
|
| 3145 |
+
"loss": 1.0619,
|
| 3146 |
+
"step": 49900
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 4.72,
|
| 3150 |
+
"learning_rate": 3.477677072843236e-06,
|
| 3151 |
+
"loss": 1.052,
|
| 3152 |
+
"step": 50000
|
| 3153 |
+
},
|
| 3154 |
+
{
|
| 3155 |
+
"epoch": 4.72,
|
| 3156 |
+
"eval_bleu1": 29.793,
|
| 3157 |
+
"eval_bleu2": 16.882,
|
| 3158 |
+
"eval_bleu3": 9.6468,
|
| 3159 |
+
"eval_bleu4": 5.3654,
|
| 3160 |
+
"eval_gen_len": 50.6014,
|
| 3161 |
+
"eval_loss": 2.989078998565674,
|
| 3162 |
+
"eval_rouge1": 35.4597,
|
| 3163 |
+
"eval_rouge2": 12.0824,
|
| 3164 |
+
"eval_rougeL": 23.0161,
|
| 3165 |
+
"eval_runtime": 113.3288,
|
| 3166 |
+
"eval_samples_per_second": 3.785,
|
| 3167 |
+
"eval_steps_per_second": 0.035,
|
| 3168 |
+
"step": 50000
|
| 3169 |
+
},
|
| 3170 |
+
{
|
| 3171 |
+
"epoch": 4.73,
|
| 3172 |
+
"learning_rate": 3.3601879825444777e-06,
|
| 3173 |
+
"loss": 1.0508,
|
| 3174 |
+
"step": 50100
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 4.74,
|
| 3178 |
+
"learning_rate": 3.24269889224572e-06,
|
| 3179 |
+
"loss": 1.052,
|
| 3180 |
+
"step": 50200
|
| 3181 |
+
},
|
| 3182 |
+
{
|
| 3183 |
+
"epoch": 4.75,
|
| 3184 |
+
"learning_rate": 3.1252098019469617e-06,
|
| 3185 |
+
"loss": 1.0357,
|
| 3186 |
+
"step": 50300
|
| 3187 |
+
},
|
| 3188 |
+
{
|
| 3189 |
+
"epoch": 4.76,
|
| 3190 |
+
"learning_rate": 3.007720711648204e-06,
|
| 3191 |
+
"loss": 1.0977,
|
| 3192 |
+
"step": 50400
|
| 3193 |
+
},
|
| 3194 |
+
{
|
| 3195 |
+
"epoch": 4.77,
|
| 3196 |
+
"learning_rate": 2.890231621349446e-06,
|
| 3197 |
+
"loss": 1.0845,
|
| 3198 |
+
"step": 50500
|
| 3199 |
+
},
|
| 3200 |
+
{
|
| 3201 |
+
"epoch": 4.78,
|
| 3202 |
+
"learning_rate": 2.772742531050688e-06,
|
| 3203 |
+
"loss": 1.0882,
|
| 3204 |
+
"step": 50600
|
| 3205 |
+
},
|
| 3206 |
+
{
|
| 3207 |
+
"epoch": 4.79,
|
| 3208 |
+
"learning_rate": 2.65525344075193e-06,
|
| 3209 |
+
"loss": 1.0633,
|
| 3210 |
+
"step": 50700
|
| 3211 |
+
},
|
| 3212 |
+
{
|
| 3213 |
+
"epoch": 4.8,
|
| 3214 |
+
"learning_rate": 2.537764350453172e-06,
|
| 3215 |
+
"loss": 1.0854,
|
| 3216 |
+
"step": 50800
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 4.81,
|
| 3220 |
+
"learning_rate": 2.420275260154414e-06,
|
| 3221 |
+
"loss": 1.0948,
|
| 3222 |
+
"step": 50900
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 4.81,
|
| 3226 |
+
"learning_rate": 2.3027861698556565e-06,
|
| 3227 |
+
"loss": 1.0605,
|
| 3228 |
+
"step": 51000
|
| 3229 |
+
},
|
| 3230 |
+
{
|
| 3231 |
+
"epoch": 4.82,
|
| 3232 |
+
"learning_rate": 2.185297079556898e-06,
|
| 3233 |
+
"loss": 1.0726,
|
| 3234 |
+
"step": 51100
|
| 3235 |
+
},
|
| 3236 |
+
{
|
| 3237 |
+
"epoch": 4.83,
|
| 3238 |
+
"learning_rate": 2.0678079892581405e-06,
|
| 3239 |
+
"loss": 1.0495,
|
| 3240 |
+
"step": 51200
|
| 3241 |
+
},
|
| 3242 |
+
{
|
| 3243 |
+
"epoch": 4.84,
|
| 3244 |
+
"learning_rate": 1.950318898959382e-06,
|
| 3245 |
+
"loss": 1.0376,
|
| 3246 |
+
"step": 51300
|
| 3247 |
+
},
|
| 3248 |
+
{
|
| 3249 |
+
"epoch": 4.85,
|
| 3250 |
+
"learning_rate": 1.8328298086606243e-06,
|
| 3251 |
+
"loss": 1.0678,
|
| 3252 |
+
"step": 51400
|
| 3253 |
+
},
|
| 3254 |
+
{
|
| 3255 |
+
"epoch": 4.86,
|
| 3256 |
+
"learning_rate": 1.7153407183618663e-06,
|
| 3257 |
+
"loss": 1.0557,
|
| 3258 |
+
"step": 51500
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 4.87,
|
| 3262 |
+
"learning_rate": 1.5978516280631082e-06,
|
| 3263 |
+
"loss": 1.0985,
|
| 3264 |
+
"step": 51600
|
| 3265 |
+
},
|
| 3266 |
+
{
|
| 3267 |
+
"epoch": 4.88,
|
| 3268 |
+
"learning_rate": 1.4803625377643502e-06,
|
| 3269 |
+
"loss": 1.0851,
|
| 3270 |
+
"step": 51700
|
| 3271 |
+
},
|
| 3272 |
+
{
|
| 3273 |
+
"epoch": 4.89,
|
| 3274 |
+
"learning_rate": 1.3628734474655924e-06,
|
| 3275 |
+
"loss": 1.0664,
|
| 3276 |
+
"step": 51800
|
| 3277 |
+
},
|
| 3278 |
+
{
|
| 3279 |
+
"epoch": 4.9,
|
| 3280 |
+
"learning_rate": 1.2453843571668344e-06,
|
| 3281 |
+
"loss": 1.0403,
|
| 3282 |
+
"step": 51900
|
| 3283 |
+
},
|
| 3284 |
+
{
|
| 3285 |
+
"epoch": 4.91,
|
| 3286 |
+
"learning_rate": 1.1278952668680764e-06,
|
| 3287 |
+
"loss": 1.0449,
|
| 3288 |
+
"step": 52000
|
| 3289 |
+
},
|
| 3290 |
+
{
|
| 3291 |
+
"epoch": 4.92,
|
| 3292 |
+
"learning_rate": 1.0104061765693184e-06,
|
| 3293 |
+
"loss": 1.0334,
|
| 3294 |
+
"step": 52100
|
| 3295 |
+
},
|
| 3296 |
+
{
|
| 3297 |
+
"epoch": 4.93,
|
| 3298 |
+
"learning_rate": 8.929170862705605e-07,
|
| 3299 |
+
"loss": 1.0725,
|
| 3300 |
+
"step": 52200
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 4.94,
|
| 3304 |
+
"learning_rate": 7.754279959718026e-07,
|
| 3305 |
+
"loss": 1.0981,
|
| 3306 |
+
"step": 52300
|
| 3307 |
+
},
|
| 3308 |
+
{
|
| 3309 |
+
"epoch": 4.95,
|
| 3310 |
+
"learning_rate": 6.579389056730447e-07,
|
| 3311 |
+
"loss": 1.0442,
|
| 3312 |
+
"step": 52400
|
| 3313 |
+
},
|
| 3314 |
+
{
|
| 3315 |
+
"epoch": 4.96,
|
| 3316 |
+
"learning_rate": 5.404498153742867e-07,
|
| 3317 |
+
"loss": 1.0475,
|
| 3318 |
+
"step": 52500
|
| 3319 |
+
},
|
| 3320 |
+
{
|
| 3321 |
+
"epoch": 4.97,
|
| 3322 |
+
"learning_rate": 4.229607250755287e-07,
|
| 3323 |
+
"loss": 1.0857,
|
| 3324 |
+
"step": 52600
|
| 3325 |
+
},
|
| 3326 |
+
{
|
| 3327 |
+
"epoch": 4.98,
|
| 3328 |
+
"learning_rate": 3.054716347767707e-07,
|
| 3329 |
+
"loss": 1.1035,
|
| 3330 |
+
"step": 52700
|
| 3331 |
+
},
|
| 3332 |
+
{
|
| 3333 |
+
"epoch": 4.98,
|
| 3334 |
+
"learning_rate": 1.8798254447801276e-07,
|
| 3335 |
+
"loss": 1.0935,
|
| 3336 |
+
"step": 52800
|
| 3337 |
+
},
|
| 3338 |
+
{
|
| 3339 |
+
"epoch": 4.99,
|
| 3340 |
+
"learning_rate": 7.049345417925478e-08,
|
| 3341 |
+
"loss": 1.0707,
|
| 3342 |
+
"step": 52900
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 5.0,
|
| 3346 |
+
"step": 52960,
|
| 3347 |
+
"total_flos": 1.489824167150592e+16,
|
| 3348 |
+
"train_loss": 1.7010083939733822,
|
| 3349 |
+
"train_runtime": 6451.2169,
|
| 3350 |
+
"train_samples_per_second": 32.836,
|
| 3351 |
+
"train_steps_per_second": 8.209
|
| 3352 |
+
}
|
| 3353 |
+
],
|
| 3354 |
+
"max_steps": 52960,
|
| 3355 |
+
"num_train_epochs": 5,
|
| 3356 |
+
"total_flos": 1.489824167150592e+16,
|
| 3357 |
+
"trial_name": null,
|
| 3358 |
+
"trial_params": null
|
| 3359 |
+
}
|