End of training
Browse files- all_results.json +18 -0
- eval_results.json +13 -0
- train_results.json +8 -0
- trainer_state.json +1497 -0
all_results.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"eval_gen_len": 13.1493370551291,
|
| 4 |
+
"eval_loss": 1.0642070770263672,
|
| 5 |
+
"eval_rouge1": 58.9516,
|
| 6 |
+
"eval_rouge2": 41.8006,
|
| 7 |
+
"eval_rougeL": 56.8249,
|
| 8 |
+
"eval_rougeLsum": 56.9171,
|
| 9 |
+
"eval_runtime": 235.2294,
|
| 10 |
+
"eval_samples": 2866,
|
| 11 |
+
"eval_samples_per_second": 12.184,
|
| 12 |
+
"eval_steps_per_second": 3.048,
|
| 13 |
+
"train_loss": 1.2681677364512702,
|
| 14 |
+
"train_runtime": 3900.1166,
|
| 15 |
+
"train_samples": 23167,
|
| 16 |
+
"train_samples_per_second": 11.88,
|
| 17 |
+
"train_steps_per_second": 0.186
|
| 18 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"eval_gen_len": 13.1493370551291,
|
| 4 |
+
"eval_loss": 1.0642070770263672,
|
| 5 |
+
"eval_rouge1": 58.9516,
|
| 6 |
+
"eval_rouge2": 41.8006,
|
| 7 |
+
"eval_rougeL": 56.8249,
|
| 8 |
+
"eval_rougeLsum": 56.9171,
|
| 9 |
+
"eval_runtime": 235.2294,
|
| 10 |
+
"eval_samples": 2866,
|
| 11 |
+
"eval_samples_per_second": 12.184,
|
| 12 |
+
"eval_steps_per_second": 3.048
|
| 13 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"train_loss": 1.2681677364512702,
|
| 4 |
+
"train_runtime": 3900.1166,
|
| 5 |
+
"train_samples": 23167,
|
| 6 |
+
"train_samples_per_second": 11.88,
|
| 7 |
+
"train_steps_per_second": 0.186
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1497 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"global_step": 724,
|
| 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": 1.6000000000000003e-05,
|
| 13 |
+
"loss": 2.5925,
|
| 14 |
+
"step": 3
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.02,
|
| 18 |
+
"learning_rate": 3.2000000000000005e-05,
|
| 19 |
+
"loss": 2.521,
|
| 20 |
+
"step": 6
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.02,
|
| 24 |
+
"learning_rate": 4.8e-05,
|
| 25 |
+
"loss": 2.3332,
|
| 26 |
+
"step": 9
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.03,
|
| 30 |
+
"learning_rate": 6.400000000000001e-05,
|
| 31 |
+
"loss": 2.3091,
|
| 32 |
+
"step": 12
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.04,
|
| 36 |
+
"learning_rate": 8e-05,
|
| 37 |
+
"loss": 2.0732,
|
| 38 |
+
"step": 15
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.05,
|
| 42 |
+
"learning_rate": 7.999646594434211e-05,
|
| 43 |
+
"loss": 2.0075,
|
| 44 |
+
"step": 18
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.06,
|
| 48 |
+
"learning_rate": 7.998586440184589e-05,
|
| 49 |
+
"loss": 1.821,
|
| 50 |
+
"step": 21
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.07,
|
| 54 |
+
"learning_rate": 7.996819724583341e-05,
|
| 55 |
+
"loss": 1.7247,
|
| 56 |
+
"step": 24
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.07,
|
| 60 |
+
"learning_rate": 7.99434675981403e-05,
|
| 61 |
+
"loss": 1.6434,
|
| 62 |
+
"step": 27
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.08,
|
| 66 |
+
"learning_rate": 7.991167982856416e-05,
|
| 67 |
+
"loss": 1.6819,
|
| 68 |
+
"step": 30
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.09,
|
| 72 |
+
"learning_rate": 7.987283955409229e-05,
|
| 73 |
+
"loss": 1.5917,
|
| 74 |
+
"step": 33
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.1,
|
| 78 |
+
"learning_rate": 7.982695363790929e-05,
|
| 79 |
+
"loss": 1.6361,
|
| 80 |
+
"step": 36
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.11,
|
| 84 |
+
"learning_rate": 7.977403018818425e-05,
|
| 85 |
+
"loss": 1.6557,
|
| 86 |
+
"step": 39
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.12,
|
| 90 |
+
"learning_rate": 7.971407855663803e-05,
|
| 91 |
+
"loss": 1.5521,
|
| 92 |
+
"step": 42
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.12,
|
| 96 |
+
"learning_rate": 7.964710933689073e-05,
|
| 97 |
+
"loss": 1.5572,
|
| 98 |
+
"step": 45
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.13,
|
| 102 |
+
"learning_rate": 7.95731343625899e-05,
|
| 103 |
+
"loss": 1.4677,
|
| 104 |
+
"step": 48
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.14,
|
| 108 |
+
"learning_rate": 7.94921667053193e-05,
|
| 109 |
+
"loss": 1.4902,
|
| 110 |
+
"step": 51
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.15,
|
| 114 |
+
"learning_rate": 7.940422067228933e-05,
|
| 115 |
+
"loss": 1.3551,
|
| 116 |
+
"step": 54
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.16,
|
| 120 |
+
"learning_rate": 7.930931180380879e-05,
|
| 121 |
+
"loss": 1.5743,
|
| 122 |
+
"step": 57
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.17,
|
| 126 |
+
"learning_rate": 7.920745687053881e-05,
|
| 127 |
+
"loss": 1.5111,
|
| 128 |
+
"step": 60
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.17,
|
| 132 |
+
"learning_rate": 7.909867387052959e-05,
|
| 133 |
+
"loss": 1.3469,
|
| 134 |
+
"step": 63
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.18,
|
| 138 |
+
"learning_rate": 7.898298202603996e-05,
|
| 139 |
+
"loss": 1.4434,
|
| 140 |
+
"step": 66
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.19,
|
| 144 |
+
"learning_rate": 7.886040178014079e-05,
|
| 145 |
+
"loss": 1.359,
|
| 146 |
+
"step": 69
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.2,
|
| 150 |
+
"learning_rate": 7.873095479310265e-05,
|
| 151 |
+
"loss": 1.4179,
|
| 152 |
+
"step": 72
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.21,
|
| 156 |
+
"learning_rate": 7.859466393856842e-05,
|
| 157 |
+
"loss": 1.4197,
|
| 158 |
+
"step": 75
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.22,
|
| 162 |
+
"learning_rate": 7.845155329951134e-05,
|
| 163 |
+
"loss": 1.4091,
|
| 164 |
+
"step": 78
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.22,
|
| 168 |
+
"learning_rate": 7.830164816397961e-05,
|
| 169 |
+
"loss": 1.4557,
|
| 170 |
+
"step": 81
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.23,
|
| 174 |
+
"learning_rate": 7.814497502062784e-05,
|
| 175 |
+
"loss": 1.3837,
|
| 176 |
+
"step": 84
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.24,
|
| 180 |
+
"learning_rate": 7.798156155403649e-05,
|
| 181 |
+
"loss": 1.3812,
|
| 182 |
+
"step": 87
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.25,
|
| 186 |
+
"learning_rate": 7.781143663981985e-05,
|
| 187 |
+
"loss": 1.4313,
|
| 188 |
+
"step": 90
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.26,
|
| 192 |
+
"learning_rate": 7.76346303395237e-05,
|
| 193 |
+
"loss": 1.2811,
|
| 194 |
+
"step": 93
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.27,
|
| 198 |
+
"learning_rate": 7.745117389531335e-05,
|
| 199 |
+
"loss": 1.2625,
|
| 200 |
+
"step": 96
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.27,
|
| 204 |
+
"learning_rate": 7.726109972445301e-05,
|
| 205 |
+
"loss": 1.3855,
|
| 206 |
+
"step": 99
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.28,
|
| 210 |
+
"learning_rate": 7.706444141357764e-05,
|
| 211 |
+
"loss": 1.234,
|
| 212 |
+
"step": 102
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.29,
|
| 216 |
+
"learning_rate": 7.686123371275806e-05,
|
| 217 |
+
"loss": 1.3975,
|
| 218 |
+
"step": 105
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.3,
|
| 222 |
+
"learning_rate": 7.665151252936049e-05,
|
| 223 |
+
"loss": 1.2287,
|
| 224 |
+
"step": 108
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.31,
|
| 228 |
+
"learning_rate": 7.643531492170168e-05,
|
| 229 |
+
"loss": 1.3446,
|
| 230 |
+
"step": 111
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.31,
|
| 234 |
+
"learning_rate": 7.621267909250057e-05,
|
| 235 |
+
"loss": 1.2702,
|
| 236 |
+
"step": 114
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.32,
|
| 240 |
+
"learning_rate": 7.598364438212773e-05,
|
| 241 |
+
"loss": 1.2986,
|
| 242 |
+
"step": 117
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.33,
|
| 246 |
+
"learning_rate": 7.574825126165386e-05,
|
| 247 |
+
"loss": 1.3812,
|
| 248 |
+
"step": 120
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.34,
|
| 252 |
+
"learning_rate": 7.550654132569846e-05,
|
| 253 |
+
"loss": 1.354,
|
| 254 |
+
"step": 123
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.35,
|
| 258 |
+
"learning_rate": 7.525855728507984e-05,
|
| 259 |
+
"loss": 1.4568,
|
| 260 |
+
"step": 126
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.36,
|
| 264 |
+
"learning_rate": 7.500434295926807e-05,
|
| 265 |
+
"loss": 1.4748,
|
| 266 |
+
"step": 129
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.36,
|
| 270 |
+
"learning_rate": 7.474394326864201e-05,
|
| 271 |
+
"loss": 1.4357,
|
| 272 |
+
"step": 132
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.37,
|
| 276 |
+
"learning_rate": 7.447740422655164e-05,
|
| 277 |
+
"loss": 1.3519,
|
| 278 |
+
"step": 135
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.38,
|
| 282 |
+
"learning_rate": 7.420477293118745e-05,
|
| 283 |
+
"loss": 1.2764,
|
| 284 |
+
"step": 138
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.39,
|
| 288 |
+
"learning_rate": 7.392609755725803e-05,
|
| 289 |
+
"loss": 1.2988,
|
| 290 |
+
"step": 141
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.4,
|
| 294 |
+
"learning_rate": 7.36414273474775e-05,
|
| 295 |
+
"loss": 1.4085,
|
| 296 |
+
"step": 144
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.41,
|
| 300 |
+
"learning_rate": 7.33508126038641e-05,
|
| 301 |
+
"loss": 1.3397,
|
| 302 |
+
"step": 147
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.41,
|
| 306 |
+
"learning_rate": 7.305430467885182e-05,
|
| 307 |
+
"loss": 1.2507,
|
| 308 |
+
"step": 150
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.42,
|
| 312 |
+
"learning_rate": 7.275195596621611e-05,
|
| 313 |
+
"loss": 1.3038,
|
| 314 |
+
"step": 153
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.43,
|
| 318 |
+
"learning_rate": 7.244381989181594e-05,
|
| 319 |
+
"loss": 1.2589,
|
| 320 |
+
"step": 156
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.44,
|
| 324 |
+
"learning_rate": 7.212995090415312e-05,
|
| 325 |
+
"loss": 1.2187,
|
| 326 |
+
"step": 159
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.45,
|
| 330 |
+
"learning_rate": 7.181040446475129e-05,
|
| 331 |
+
"loss": 1.2102,
|
| 332 |
+
"step": 162
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.46,
|
| 336 |
+
"learning_rate": 7.148523703835553e-05,
|
| 337 |
+
"loss": 1.3638,
|
| 338 |
+
"step": 165
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.46,
|
| 342 |
+
"learning_rate": 7.115450608295498e-05,
|
| 343 |
+
"loss": 1.32,
|
| 344 |
+
"step": 168
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.47,
|
| 348 |
+
"learning_rate": 7.081827003962987e-05,
|
| 349 |
+
"loss": 1.3807,
|
| 350 |
+
"step": 171
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.48,
|
| 354 |
+
"learning_rate": 7.047658832222475e-05,
|
| 355 |
+
"loss": 1.3778,
|
| 356 |
+
"step": 174
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.49,
|
| 360 |
+
"learning_rate": 7.012952130684995e-05,
|
| 361 |
+
"loss": 1.2008,
|
| 362 |
+
"step": 177
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.5,
|
| 366 |
+
"learning_rate": 6.977713032121295e-05,
|
| 367 |
+
"loss": 1.3183,
|
| 368 |
+
"step": 180
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.51,
|
| 372 |
+
"learning_rate": 6.941947763378157e-05,
|
| 373 |
+
"loss": 1.2617,
|
| 374 |
+
"step": 183
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.51,
|
| 378 |
+
"learning_rate": 6.905662644278099e-05,
|
| 379 |
+
"loss": 1.355,
|
| 380 |
+
"step": 186
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.52,
|
| 384 |
+
"learning_rate": 6.868864086502643e-05,
|
| 385 |
+
"loss": 1.2293,
|
| 386 |
+
"step": 189
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.53,
|
| 390 |
+
"learning_rate": 6.831558592459356e-05,
|
| 391 |
+
"loss": 1.4429,
|
| 392 |
+
"step": 192
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.54,
|
| 396 |
+
"learning_rate": 6.793752754132852e-05,
|
| 397 |
+
"loss": 1.2577,
|
| 398 |
+
"step": 195
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.55,
|
| 402 |
+
"learning_rate": 6.755453251919973e-05,
|
| 403 |
+
"loss": 1.3414,
|
| 404 |
+
"step": 198
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 0.56,
|
| 408 |
+
"learning_rate": 6.716666853449342e-05,
|
| 409 |
+
"loss": 1.2173,
|
| 410 |
+
"step": 201
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.56,
|
| 414 |
+
"learning_rate": 6.67740041238551e-05,
|
| 415 |
+
"loss": 1.2659,
|
| 416 |
+
"step": 204
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.57,
|
| 420 |
+
"learning_rate": 6.637660867217884e-05,
|
| 421 |
+
"loss": 1.3387,
|
| 422 |
+
"step": 207
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.58,
|
| 426 |
+
"learning_rate": 6.59745524003469e-05,
|
| 427 |
+
"loss": 1.213,
|
| 428 |
+
"step": 210
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.59,
|
| 432 |
+
"learning_rate": 6.556790635282136e-05,
|
| 433 |
+
"loss": 1.303,
|
| 434 |
+
"step": 213
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.6,
|
| 438 |
+
"learning_rate": 6.515674238509048e-05,
|
| 439 |
+
"loss": 1.3497,
|
| 440 |
+
"step": 216
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 0.6,
|
| 444 |
+
"learning_rate": 6.474113315097161e-05,
|
| 445 |
+
"loss": 1.3096,
|
| 446 |
+
"step": 219
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.61,
|
| 450 |
+
"learning_rate": 6.432115208977297e-05,
|
| 451 |
+
"loss": 1.2827,
|
| 452 |
+
"step": 222
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.62,
|
| 456 |
+
"learning_rate": 6.389687341331688e-05,
|
| 457 |
+
"loss": 1.2684,
|
| 458 |
+
"step": 225
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.63,
|
| 462 |
+
"learning_rate": 6.346837209282615e-05,
|
| 463 |
+
"loss": 1.2377,
|
| 464 |
+
"step": 228
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.64,
|
| 468 |
+
"learning_rate": 6.303572384567662e-05,
|
| 469 |
+
"loss": 1.1533,
|
| 470 |
+
"step": 231
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.65,
|
| 474 |
+
"learning_rate": 6.259900512201756e-05,
|
| 475 |
+
"loss": 1.259,
|
| 476 |
+
"step": 234
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.65,
|
| 480 |
+
"learning_rate": 6.215829309126279e-05,
|
| 481 |
+
"loss": 1.2757,
|
| 482 |
+
"step": 237
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.66,
|
| 486 |
+
"learning_rate": 6.17136656284546e-05,
|
| 487 |
+
"loss": 1.1996,
|
| 488 |
+
"step": 240
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 0.67,
|
| 492 |
+
"learning_rate": 6.1265201300503e-05,
|
| 493 |
+
"loss": 1.185,
|
| 494 |
+
"step": 243
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.68,
|
| 498 |
+
"learning_rate": 6.081297935230281e-05,
|
| 499 |
+
"loss": 1.2503,
|
| 500 |
+
"step": 246
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.69,
|
| 504 |
+
"learning_rate": 6.035707969273072e-05,
|
| 505 |
+
"loss": 1.3338,
|
| 506 |
+
"step": 249
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.7,
|
| 510 |
+
"learning_rate": 5.989758288052531e-05,
|
| 511 |
+
"loss": 1.2651,
|
| 512 |
+
"step": 252
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 0.7,
|
| 516 |
+
"learning_rate": 5.9434570110052036e-05,
|
| 517 |
+
"loss": 1.2211,
|
| 518 |
+
"step": 255
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 0.71,
|
| 522 |
+
"learning_rate": 5.8968123196955955e-05,
|
| 523 |
+
"loss": 1.2736,
|
| 524 |
+
"step": 258
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.72,
|
| 528 |
+
"learning_rate": 5.8498324563704676e-05,
|
| 529 |
+
"loss": 1.2947,
|
| 530 |
+
"step": 261
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.73,
|
| 534 |
+
"learning_rate": 5.80252572250241e-05,
|
| 535 |
+
"loss": 1.2006,
|
| 536 |
+
"step": 264
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.74,
|
| 540 |
+
"learning_rate": 5.7549004773229474e-05,
|
| 541 |
+
"loss": 1.2569,
|
| 542 |
+
"step": 267
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.75,
|
| 546 |
+
"learning_rate": 5.706965136345439e-05,
|
| 547 |
+
"loss": 1.2825,
|
| 548 |
+
"step": 270
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.75,
|
| 552 |
+
"learning_rate": 5.658728169878033e-05,
|
| 553 |
+
"loss": 1.2655,
|
| 554 |
+
"step": 273
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.76,
|
| 558 |
+
"learning_rate": 5.6101981015269436e-05,
|
| 559 |
+
"loss": 1.1506,
|
| 560 |
+
"step": 276
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 0.77,
|
| 564 |
+
"learning_rate": 5.561383506690303e-05,
|
| 565 |
+
"loss": 1.2926,
|
| 566 |
+
"step": 279
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 0.78,
|
| 570 |
+
"learning_rate": 5.512293011042863e-05,
|
| 571 |
+
"loss": 1.1894,
|
| 572 |
+
"step": 282
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.79,
|
| 576 |
+
"learning_rate": 5.462935289011821e-05,
|
| 577 |
+
"loss": 1.2418,
|
| 578 |
+
"step": 285
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.8,
|
| 582 |
+
"learning_rate": 5.4133190622440153e-05,
|
| 583 |
+
"loss": 1.2954,
|
| 584 |
+
"step": 288
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.8,
|
| 588 |
+
"learning_rate": 5.363453098064792e-05,
|
| 589 |
+
"loss": 1.2612,
|
| 590 |
+
"step": 291
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.81,
|
| 594 |
+
"learning_rate": 5.313346207928795e-05,
|
| 595 |
+
"loss": 1.2406,
|
| 596 |
+
"step": 294
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 0.82,
|
| 600 |
+
"learning_rate": 5.2630072458629526e-05,
|
| 601 |
+
"loss": 1.1681,
|
| 602 |
+
"step": 297
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.83,
|
| 606 |
+
"learning_rate": 5.2124451069019495e-05,
|
| 607 |
+
"loss": 1.2025,
|
| 608 |
+
"step": 300
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 0.84,
|
| 612 |
+
"learning_rate": 5.161668725516451e-05,
|
| 613 |
+
"loss": 1.2296,
|
| 614 |
+
"step": 303
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 0.85,
|
| 618 |
+
"learning_rate": 5.110687074034351e-05,
|
| 619 |
+
"loss": 1.195,
|
| 620 |
+
"step": 306
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.85,
|
| 624 |
+
"learning_rate": 5.059509161055343e-05,
|
| 625 |
+
"loss": 1.2232,
|
| 626 |
+
"step": 309
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.86,
|
| 630 |
+
"learning_rate": 5.008144029859074e-05,
|
| 631 |
+
"loss": 1.1391,
|
| 632 |
+
"step": 312
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.87,
|
| 636 |
+
"learning_rate": 4.956600756807172e-05,
|
| 637 |
+
"loss": 1.2474,
|
| 638 |
+
"step": 315
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 0.88,
|
| 642 |
+
"learning_rate": 4.904888449739422e-05,
|
| 643 |
+
"loss": 1.2665,
|
| 644 |
+
"step": 318
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 0.89,
|
| 648 |
+
"learning_rate": 4.8530162463643935e-05,
|
| 649 |
+
"loss": 1.1458,
|
| 650 |
+
"step": 321
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.9,
|
| 654 |
+
"learning_rate": 4.800993312644778e-05,
|
| 655 |
+
"loss": 1.3478,
|
| 656 |
+
"step": 324
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 0.9,
|
| 660 |
+
"learning_rate": 4.748828841177738e-05,
|
| 661 |
+
"loss": 1.2419,
|
| 662 |
+
"step": 327
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 0.91,
|
| 666 |
+
"learning_rate": 4.6965320495705504e-05,
|
| 667 |
+
"loss": 1.2474,
|
| 668 |
+
"step": 330
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.92,
|
| 672 |
+
"learning_rate": 4.644112178811828e-05,
|
| 673 |
+
"loss": 1.2526,
|
| 674 |
+
"step": 333
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.93,
|
| 678 |
+
"learning_rate": 4.591578491638613e-05,
|
| 679 |
+
"loss": 1.2924,
|
| 680 |
+
"step": 336
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 0.94,
|
| 684 |
+
"learning_rate": 4.538940270899625e-05,
|
| 685 |
+
"loss": 1.2351,
|
| 686 |
+
"step": 339
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 0.94,
|
| 690 |
+
"learning_rate": 4.4862068179149546e-05,
|
| 691 |
+
"loss": 1.2399,
|
| 692 |
+
"step": 342
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.95,
|
| 696 |
+
"learning_rate": 4.4333874508324964e-05,
|
| 697 |
+
"loss": 1.2213,
|
| 698 |
+
"step": 345
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.96,
|
| 702 |
+
"learning_rate": 4.3804915029814054e-05,
|
| 703 |
+
"loss": 1.2421,
|
| 704 |
+
"step": 348
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.97,
|
| 708 |
+
"learning_rate": 4.327528321222869e-05,
|
| 709 |
+
"loss": 1.4176,
|
| 710 |
+
"step": 351
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.98,
|
| 714 |
+
"learning_rate": 4.274507264298496e-05,
|
| 715 |
+
"loss": 1.112,
|
| 716 |
+
"step": 354
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.99,
|
| 720 |
+
"learning_rate": 4.2214377011765956e-05,
|
| 721 |
+
"loss": 1.2284,
|
| 722 |
+
"step": 357
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.99,
|
| 726 |
+
"learning_rate": 4.1683290093966603e-05,
|
| 727 |
+
"loss": 1.1939,
|
| 728 |
+
"step": 360
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.0,
|
| 732 |
+
"eval_gen_len": 13.259246336357293,
|
| 733 |
+
"eval_loss": 1.0821669101715088,
|
| 734 |
+
"eval_rouge1": 58.1758,
|
| 735 |
+
"eval_rouge2": 40.9388,
|
| 736 |
+
"eval_rougeL": 56.1219,
|
| 737 |
+
"eval_rougeLsum": 56.2464,
|
| 738 |
+
"eval_runtime": 238.7086,
|
| 739 |
+
"eval_samples_per_second": 12.006,
|
| 740 |
+
"eval_steps_per_second": 3.004,
|
| 741 |
+
"step": 362
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 1.0,
|
| 745 |
+
"learning_rate": 4.115190573412321e-05,
|
| 746 |
+
"loss": 1.2689,
|
| 747 |
+
"step": 363
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.01,
|
| 751 |
+
"learning_rate": 4.062031782933099e-05,
|
| 752 |
+
"loss": 1.1309,
|
| 753 |
+
"step": 366
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 1.02,
|
| 757 |
+
"learning_rate": 4.008862031265205e-05,
|
| 758 |
+
"loss": 1.1703,
|
| 759 |
+
"step": 369
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 1.03,
|
| 763 |
+
"learning_rate": 3.955690713651723e-05,
|
| 764 |
+
"loss": 1.2228,
|
| 765 |
+
"step": 372
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 1.04,
|
| 769 |
+
"learning_rate": 3.902527225612447e-05,
|
| 770 |
+
"loss": 1.242,
|
| 771 |
+
"step": 375
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 1.04,
|
| 775 |
+
"learning_rate": 3.849380961283661e-05,
|
| 776 |
+
"loss": 1.1599,
|
| 777 |
+
"step": 378
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 1.05,
|
| 781 |
+
"learning_rate": 3.796261311758174e-05,
|
| 782 |
+
"loss": 1.275,
|
| 783 |
+
"step": 381
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 1.06,
|
| 787 |
+
"learning_rate": 3.743177663425883e-05,
|
| 788 |
+
"loss": 1.1893,
|
| 789 |
+
"step": 384
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 1.07,
|
| 793 |
+
"learning_rate": 3.690139396315174e-05,
|
| 794 |
+
"loss": 1.1084,
|
| 795 |
+
"step": 387
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 1.08,
|
| 799 |
+
"learning_rate": 3.637155882435446e-05,
|
| 800 |
+
"loss": 1.2845,
|
| 801 |
+
"step": 390
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 1.09,
|
| 805 |
+
"learning_rate": 3.5842364841210466e-05,
|
| 806 |
+
"loss": 1.2036,
|
| 807 |
+
"step": 393
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 1.09,
|
| 811 |
+
"learning_rate": 3.53139055237693e-05,
|
| 812 |
+
"loss": 1.1238,
|
| 813 |
+
"step": 396
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 1.1,
|
| 817 |
+
"learning_rate": 3.478627425226299e-05,
|
| 818 |
+
"loss": 1.1746,
|
| 819 |
+
"step": 399
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 1.11,
|
| 823 |
+
"learning_rate": 3.4259564260605564e-05,
|
| 824 |
+
"loss": 1.2253,
|
| 825 |
+
"step": 402
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 1.12,
|
| 829 |
+
"learning_rate": 3.373386861991832e-05,
|
| 830 |
+
"loss": 1.235,
|
| 831 |
+
"step": 405
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 1.13,
|
| 835 |
+
"learning_rate": 3.320928022208392e-05,
|
| 836 |
+
"loss": 1.143,
|
| 837 |
+
"step": 408
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 1.14,
|
| 841 |
+
"learning_rate": 3.268589176333213e-05,
|
| 842 |
+
"loss": 1.2727,
|
| 843 |
+
"step": 411
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 1.14,
|
| 847 |
+
"learning_rate": 3.216379572786015e-05,
|
| 848 |
+
"loss": 1.1688,
|
| 849 |
+
"step": 414
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 1.15,
|
| 853 |
+
"learning_rate": 3.1643084371490394e-05,
|
| 854 |
+
"loss": 1.162,
|
| 855 |
+
"step": 417
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 1.16,
|
| 859 |
+
"learning_rate": 3.112384970536862e-05,
|
| 860 |
+
"loss": 1.1652,
|
| 861 |
+
"step": 420
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 1.17,
|
| 865 |
+
"learning_rate": 3.060618347970529e-05,
|
| 866 |
+
"loss": 1.1569,
|
| 867 |
+
"step": 423
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1.18,
|
| 871 |
+
"learning_rate": 3.0090177167563106e-05,
|
| 872 |
+
"loss": 1.1664,
|
| 873 |
+
"step": 426
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 1.19,
|
| 877 |
+
"learning_rate": 2.9575921948693394e-05,
|
| 878 |
+
"loss": 1.202,
|
| 879 |
+
"step": 429
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 1.19,
|
| 883 |
+
"learning_rate": 2.906350869342447e-05,
|
| 884 |
+
"loss": 1.1277,
|
| 885 |
+
"step": 432
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 1.2,
|
| 889 |
+
"learning_rate": 2.8553027946604523e-05,
|
| 890 |
+
"loss": 1.1029,
|
| 891 |
+
"step": 435
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1.21,
|
| 895 |
+
"learning_rate": 2.8044569911602134e-05,
|
| 896 |
+
"loss": 1.1686,
|
| 897 |
+
"step": 438
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.22,
|
| 901 |
+
"learning_rate": 2.7538224434367063e-05,
|
| 902 |
+
"loss": 1.1803,
|
| 903 |
+
"step": 441
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 1.23,
|
| 907 |
+
"learning_rate": 2.703408098755424e-05,
|
| 908 |
+
"loss": 1.1585,
|
| 909 |
+
"step": 444
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 1.23,
|
| 913 |
+
"learning_rate": 2.6532228654713706e-05,
|
| 914 |
+
"loss": 1.1992,
|
| 915 |
+
"step": 447
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 1.24,
|
| 919 |
+
"learning_rate": 2.603275611454928e-05,
|
| 920 |
+
"loss": 1.1172,
|
| 921 |
+
"step": 450
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 1.25,
|
| 925 |
+
"learning_rate": 2.5535751625248784e-05,
|
| 926 |
+
"loss": 1.0731,
|
| 927 |
+
"step": 453
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 1.26,
|
| 931 |
+
"learning_rate": 2.5041303008888593e-05,
|
| 932 |
+
"loss": 1.0831,
|
| 933 |
+
"step": 456
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 1.27,
|
| 937 |
+
"learning_rate": 2.454949763591521e-05,
|
| 938 |
+
"loss": 1.1137,
|
| 939 |
+
"step": 459
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 1.28,
|
| 943 |
+
"learning_rate": 2.406042240970668e-05,
|
| 944 |
+
"loss": 1.2103,
|
| 945 |
+
"step": 462
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 1.28,
|
| 949 |
+
"learning_rate": 2.3574163751216513e-05,
|
| 950 |
+
"loss": 1.2651,
|
| 951 |
+
"step": 465
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 1.29,
|
| 955 |
+
"learning_rate": 2.30908075837029e-05,
|
| 956 |
+
"loss": 1.1251,
|
| 957 |
+
"step": 468
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 1.3,
|
| 961 |
+
"learning_rate": 2.2610439317545723e-05,
|
| 962 |
+
"loss": 1.0784,
|
| 963 |
+
"step": 471
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 1.31,
|
| 967 |
+
"learning_rate": 2.213314383515447e-05,
|
| 968 |
+
"loss": 1.1539,
|
| 969 |
+
"step": 474
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 1.32,
|
| 973 |
+
"learning_rate": 2.1659005475969125e-05,
|
| 974 |
+
"loss": 1.1628,
|
| 975 |
+
"step": 477
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 1.33,
|
| 979 |
+
"learning_rate": 2.1188108021557236e-05,
|
| 980 |
+
"loss": 1.1177,
|
| 981 |
+
"step": 480
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 1.33,
|
| 985 |
+
"learning_rate": 2.0720534680809452e-05,
|
| 986 |
+
"loss": 1.1,
|
| 987 |
+
"step": 483
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 1.34,
|
| 991 |
+
"learning_rate": 2.0256368075236296e-05,
|
| 992 |
+
"loss": 1.2544,
|
| 993 |
+
"step": 486
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 1.35,
|
| 997 |
+
"learning_rate": 1.979569022436869e-05,
|
| 998 |
+
"loss": 1.1787,
|
| 999 |
+
"step": 489
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 1.36,
|
| 1003 |
+
"learning_rate": 1.9338582531264908e-05,
|
| 1004 |
+
"loss": 1.1325,
|
| 1005 |
+
"step": 492
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 1.37,
|
| 1009 |
+
"learning_rate": 1.8885125768126405e-05,
|
| 1010 |
+
"loss": 1.1357,
|
| 1011 |
+
"step": 495
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 1.38,
|
| 1015 |
+
"learning_rate": 1.843540006202513e-05,
|
| 1016 |
+
"loss": 1.0284,
|
| 1017 |
+
"step": 498
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 1.38,
|
| 1021 |
+
"learning_rate": 1.7989484880744917e-05,
|
| 1022 |
+
"loss": 1.2133,
|
| 1023 |
+
"step": 501
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 1.39,
|
| 1027 |
+
"learning_rate": 1.754745901873923e-05,
|
| 1028 |
+
"loss": 1.2311,
|
| 1029 |
+
"step": 504
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 1.4,
|
| 1033 |
+
"learning_rate": 1.7109400583207977e-05,
|
| 1034 |
+
"loss": 1.2002,
|
| 1035 |
+
"step": 507
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 1.41,
|
| 1039 |
+
"learning_rate": 1.667538698029581e-05,
|
| 1040 |
+
"loss": 1.1465,
|
| 1041 |
+
"step": 510
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 1.42,
|
| 1045 |
+
"learning_rate": 1.624549490141417e-05,
|
| 1046 |
+
"loss": 1.1434,
|
| 1047 |
+
"step": 513
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 1.43,
|
| 1051 |
+
"learning_rate": 1.581980030968974e-05,
|
| 1052 |
+
"loss": 1.2555,
|
| 1053 |
+
"step": 516
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 1.43,
|
| 1057 |
+
"learning_rate": 1.5398378426541535e-05,
|
| 1058 |
+
"loss": 1.1692,
|
| 1059 |
+
"step": 519
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 1.44,
|
| 1063 |
+
"learning_rate": 1.4981303718389088e-05,
|
| 1064 |
+
"loss": 1.1189,
|
| 1065 |
+
"step": 522
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 1.45,
|
| 1069 |
+
"learning_rate": 1.4568649883494001e-05,
|
| 1070 |
+
"loss": 1.1591,
|
| 1071 |
+
"step": 525
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 1.46,
|
| 1075 |
+
"learning_rate": 1.416048983893727e-05,
|
| 1076 |
+
"loss": 1.2663,
|
| 1077 |
+
"step": 528
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 1.47,
|
| 1081 |
+
"learning_rate": 1.3756895707734637e-05,
|
| 1082 |
+
"loss": 1.2414,
|
| 1083 |
+
"step": 531
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 1.48,
|
| 1087 |
+
"learning_rate": 1.3357938806092245e-05,
|
| 1088 |
+
"loss": 1.0607,
|
| 1089 |
+
"step": 534
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 1.48,
|
| 1093 |
+
"learning_rate": 1.2963689630804854e-05,
|
| 1094 |
+
"loss": 1.159,
|
| 1095 |
+
"step": 537
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 1.49,
|
| 1099 |
+
"learning_rate": 1.2574217846798921e-05,
|
| 1100 |
+
"loss": 1.1072,
|
| 1101 |
+
"step": 540
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 1.5,
|
| 1105 |
+
"learning_rate": 1.2189592274822526e-05,
|
| 1106 |
+
"loss": 1.0534,
|
| 1107 |
+
"step": 543
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 1.51,
|
| 1111 |
+
"learning_rate": 1.1809880879284608e-05,
|
| 1112 |
+
"loss": 1.1124,
|
| 1113 |
+
"step": 546
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 1.52,
|
| 1117 |
+
"learning_rate": 1.1435150756245439e-05,
|
| 1118 |
+
"loss": 1.208,
|
| 1119 |
+
"step": 549
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 1.52,
|
| 1123 |
+
"learning_rate": 1.1065468121560627e-05,
|
| 1124 |
+
"loss": 1.1458,
|
| 1125 |
+
"step": 552
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 1.53,
|
| 1129 |
+
"learning_rate": 1.0700898299180493e-05,
|
| 1130 |
+
"loss": 1.1531,
|
| 1131 |
+
"step": 555
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 1.54,
|
| 1135 |
+
"learning_rate": 1.034150570960721e-05,
|
| 1136 |
+
"loss": 1.0989,
|
| 1137 |
+
"step": 558
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 1.55,
|
| 1141 |
+
"learning_rate": 9.987353858511506e-06,
|
| 1142 |
+
"loss": 1.0973,
|
| 1143 |
+
"step": 561
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.56,
|
| 1147 |
+
"learning_rate": 9.638505325511041e-06,
|
| 1148 |
+
"loss": 1.177,
|
| 1149 |
+
"step": 564
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 1.57,
|
| 1153 |
+
"learning_rate": 9.295021753112402e-06,
|
| 1154 |
+
"loss": 1.0232,
|
| 1155 |
+
"step": 567
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 1.57,
|
| 1159 |
+
"learning_rate": 8.956963835818708e-06,
|
| 1160 |
+
"loss": 1.136,
|
| 1161 |
+
"step": 570
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 1.58,
|
| 1165 |
+
"learning_rate": 8.62439130940472e-06,
|
| 1166 |
+
"loss": 1.1989,
|
| 1167 |
+
"step": 573
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 1.59,
|
| 1171 |
+
"learning_rate": 8.297362940361386e-06,
|
| 1172 |
+
"loss": 1.0936,
|
| 1173 |
+
"step": 576
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 1.6,
|
| 1177 |
+
"learning_rate": 7.975936515511598e-06,
|
| 1178 |
+
"loss": 1.1918,
|
| 1179 |
+
"step": 579
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 1.61,
|
| 1183 |
+
"learning_rate": 7.660168831799115e-06,
|
| 1184 |
+
"loss": 1.1603,
|
| 1185 |
+
"step": 582
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.62,
|
| 1189 |
+
"learning_rate": 7.350115686252399e-06,
|
| 1190 |
+
"loss": 1.1377,
|
| 1191 |
+
"step": 585
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 1.62,
|
| 1195 |
+
"learning_rate": 7.045831866125117e-06,
|
| 1196 |
+
"loss": 1.104,
|
| 1197 |
+
"step": 588
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 1.63,
|
| 1201 |
+
"learning_rate": 6.747371139215069e-06,
|
| 1202 |
+
"loss": 1.2495,
|
| 1203 |
+
"step": 591
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 1.64,
|
| 1207 |
+
"learning_rate": 6.454786244363292e-06,
|
| 1208 |
+
"loss": 1.2585,
|
| 1209 |
+
"step": 594
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 1.65,
|
| 1213 |
+
"learning_rate": 6.168128882134934e-06,
|
| 1214 |
+
"loss": 1.1646,
|
| 1215 |
+
"step": 597
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 1.66,
|
| 1219 |
+
"learning_rate": 5.887449705683632e-06,
|
| 1220 |
+
"loss": 1.0605,
|
| 1221 |
+
"step": 600
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 1.67,
|
| 1225 |
+
"learning_rate": 5.61279831180098e-06,
|
| 1226 |
+
"loss": 1.1383,
|
| 1227 |
+
"step": 603
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 1.67,
|
| 1231 |
+
"learning_rate": 5.344223232152596e-06,
|
| 1232 |
+
"loss": 1.15,
|
| 1233 |
+
"step": 606
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 1.68,
|
| 1237 |
+
"learning_rate": 5.081771924702468e-06,
|
| 1238 |
+
"loss": 1.1565,
|
| 1239 |
+
"step": 609
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 1.69,
|
| 1243 |
+
"learning_rate": 4.825490765327003e-06,
|
| 1244 |
+
"loss": 1.2282,
|
| 1245 |
+
"step": 612
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 1.7,
|
| 1249 |
+
"learning_rate": 4.575425039620265e-06,
|
| 1250 |
+
"loss": 1.1086,
|
| 1251 |
+
"step": 615
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 1.71,
|
| 1255 |
+
"learning_rate": 4.3316189348918855e-06,
|
| 1256 |
+
"loss": 1.1149,
|
| 1257 |
+
"step": 618
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 1.72,
|
| 1261 |
+
"learning_rate": 4.094115532359064e-06,
|
| 1262 |
+
"loss": 1.1229,
|
| 1263 |
+
"step": 621
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 1.72,
|
| 1267 |
+
"learning_rate": 3.862956799533977e-06,
|
| 1268 |
+
"loss": 1.1787,
|
| 1269 |
+
"step": 624
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 1.73,
|
| 1273 |
+
"learning_rate": 3.6381835828079946e-06,
|
| 1274 |
+
"loss": 1.1632,
|
| 1275 |
+
"step": 627
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 1.74,
|
| 1279 |
+
"learning_rate": 3.4198356002340405e-06,
|
| 1280 |
+
"loss": 1.1479,
|
| 1281 |
+
"step": 630
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 1.75,
|
| 1285 |
+
"learning_rate": 3.2079514345082764e-06,
|
| 1286 |
+
"loss": 1.1127,
|
| 1287 |
+
"step": 633
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 1.76,
|
| 1291 |
+
"learning_rate": 3.0025685261524297e-06,
|
| 1292 |
+
"loss": 1.155,
|
| 1293 |
+
"step": 636
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 1.77,
|
| 1297 |
+
"learning_rate": 2.803723166897965e-06,
|
| 1298 |
+
"loss": 1.1399,
|
| 1299 |
+
"step": 639
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 1.77,
|
| 1303 |
+
"learning_rate": 2.611450493273244e-06,
|
| 1304 |
+
"loss": 1.1701,
|
| 1305 |
+
"step": 642
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 1.78,
|
| 1309 |
+
"learning_rate": 2.4257844803947573e-06,
|
| 1310 |
+
"loss": 1.1976,
|
| 1311 |
+
"step": 645
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.79,
|
| 1315 |
+
"learning_rate": 2.2467579359636726e-06,
|
| 1316 |
+
"loss": 1.2275,
|
| 1317 |
+
"step": 648
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 1.8,
|
| 1321 |
+
"learning_rate": 2.0744024944685968e-06,
|
| 1322 |
+
"loss": 1.1749,
|
| 1323 |
+
"step": 651
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 1.81,
|
| 1327 |
+
"learning_rate": 1.9087486115956987e-06,
|
| 1328 |
+
"loss": 1.1776,
|
| 1329 |
+
"step": 654
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 1.81,
|
| 1333 |
+
"learning_rate": 1.7498255588470803e-06,
|
| 1334 |
+
"loss": 1.2092,
|
| 1335 |
+
"step": 657
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 1.82,
|
| 1339 |
+
"learning_rate": 1.5976614183684214e-06,
|
| 1340 |
+
"loss": 1.1771,
|
| 1341 |
+
"step": 660
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 1.83,
|
| 1345 |
+
"learning_rate": 1.452283077986807e-06,
|
| 1346 |
+
"loss": 1.2077,
|
| 1347 |
+
"step": 663
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 1.84,
|
| 1351 |
+
"learning_rate": 1.3137162264595493e-06,
|
| 1352 |
+
"loss": 1.1609,
|
| 1353 |
+
"step": 666
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 1.85,
|
| 1357 |
+
"learning_rate": 1.181985348934931e-06,
|
| 1358 |
+
"loss": 1.1285,
|
| 1359 |
+
"step": 669
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.86,
|
| 1363 |
+
"learning_rate": 1.0571137226256067e-06,
|
| 1364 |
+
"loss": 1.1526,
|
| 1365 |
+
"step": 672
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 1.86,
|
| 1369 |
+
"learning_rate": 9.391234126954463e-07,
|
| 1370 |
+
"loss": 1.1976,
|
| 1371 |
+
"step": 675
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 1.87,
|
| 1375 |
+
"learning_rate": 8.280352683605764e-07,
|
| 1376 |
+
"loss": 1.199,
|
| 1377 |
+
"step": 678
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 1.88,
|
| 1381 |
+
"learning_rate": 7.238689192052439e-07,
|
| 1382 |
+
"loss": 1.1274,
|
| 1383 |
+
"step": 681
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 1.89,
|
| 1387 |
+
"learning_rate": 6.266427717132218e-07,
|
| 1388 |
+
"loss": 1.1299,
|
| 1389 |
+
"step": 684
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 1.9,
|
| 1393 |
+
"learning_rate": 5.363740060153522e-07,
|
| 1394 |
+
"loss": 1.1242,
|
| 1395 |
+
"step": 687
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 1.91,
|
| 1399 |
+
"learning_rate": 4.530785728537401e-07,
|
| 1400 |
+
"loss": 1.209,
|
| 1401 |
+
"step": 690
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 1.91,
|
| 1405 |
+
"learning_rate": 3.76771190763221e-07,
|
| 1406 |
+
"loss": 1.1538,
|
| 1407 |
+
"step": 693
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 1.92,
|
| 1411 |
+
"learning_rate": 3.074653434705699e-07,
|
| 1412 |
+
"loss": 1.1249,
|
| 1413 |
+
"step": 696
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 1.93,
|
| 1417 |
+
"learning_rate": 2.4517327751187423e-07,
|
| 1418 |
+
"loss": 1.1414,
|
| 1419 |
+
"step": 699
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 1.94,
|
| 1423 |
+
"learning_rate": 1.8990600006854488e-07,
|
| 1424 |
+
"loss": 1.1384,
|
| 1425 |
+
"step": 702
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 1.95,
|
| 1429 |
+
"learning_rate": 1.4167327702230283e-07,
|
| 1430 |
+
"loss": 1.2001,
|
| 1431 |
+
"step": 705
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 1.96,
|
| 1435 |
+
"learning_rate": 1.0048363122954208e-07,
|
| 1436 |
+
"loss": 1.2172,
|
| 1437 |
+
"step": 708
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 1.96,
|
| 1441 |
+
"learning_rate": 6.634434101529863e-08,
|
| 1442 |
+
"loss": 1.1307,
|
| 1443 |
+
"step": 711
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 1.97,
|
| 1447 |
+
"learning_rate": 3.926143888715484e-08,
|
| 1448 |
+
"loss": 1.1053,
|
| 1449 |
+
"step": 714
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 1.98,
|
| 1453 |
+
"learning_rate": 1.9239710469296512e-08,
|
| 1454 |
+
"loss": 1.2246,
|
| 1455 |
+
"step": 717
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 1.99,
|
| 1459 |
+
"learning_rate": 6.282693656842753e-09,
|
| 1460 |
+
"loss": 1.2048,
|
| 1461 |
+
"step": 720
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 2.0,
|
| 1465 |
+
"learning_rate": 3.9267799072817415e-10,
|
| 1466 |
+
"loss": 1.1667,
|
| 1467 |
+
"step": 723
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 2.0,
|
| 1471 |
+
"eval_gen_len": 13.1493370551291,
|
| 1472 |
+
"eval_loss": 1.0642070770263672,
|
| 1473 |
+
"eval_rouge1": 58.9516,
|
| 1474 |
+
"eval_rouge2": 41.8006,
|
| 1475 |
+
"eval_rougeL": 56.8249,
|
| 1476 |
+
"eval_rougeLsum": 56.9171,
|
| 1477 |
+
"eval_runtime": 235.5914,
|
| 1478 |
+
"eval_samples_per_second": 12.165,
|
| 1479 |
+
"eval_steps_per_second": 3.043,
|
| 1480 |
+
"step": 724
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 2.0,
|
| 1484 |
+
"step": 724,
|
| 1485 |
+
"total_flos": 4.759132075510989e+16,
|
| 1486 |
+
"train_loss": 1.2681677364512702,
|
| 1487 |
+
"train_runtime": 3900.1166,
|
| 1488 |
+
"train_samples_per_second": 11.88,
|
| 1489 |
+
"train_steps_per_second": 0.186
|
| 1490 |
+
}
|
| 1491 |
+
],
|
| 1492 |
+
"max_steps": 724,
|
| 1493 |
+
"num_train_epochs": 2,
|
| 1494 |
+
"total_flos": 4.759132075510989e+16,
|
| 1495 |
+
"trial_name": null,
|
| 1496 |
+
"trial_params": null
|
| 1497 |
+
}
|