Update README.md
Browse files
README.md
CHANGED
|
@@ -78,7 +78,20 @@ The following hyperparameters were used during training:
|
|
| 78 |
- training precision: Mixed Precision
|
| 79 |
|
| 80 |
### Training results
|
| 81 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 82 |
|
| 83 |
|
| 84 |
### Framework versions
|
|
|
|
| 78 |
- training precision: Mixed Precision
|
| 79 |
|
| 80 |
### Training results
|
| 81 |
+
```
|
| 82 |
+
***** train metrics *****
|
| 83 |
+
"epoch": 10.0,
|
| 84 |
+
"train_loss": 2.8070910754504506,
|
| 85 |
+
"train_runtime": 11217.8167,
|
| 86 |
+
"train_samples": 114248,
|
| 87 |
+
"train_samples_per_second": 101.845,
|
| 88 |
+
"train_steps_per_second": 0.099
|
| 89 |
+
|
| 90 |
+
***** eval metrics *****
|
| 91 |
+
"eval_loss": 2.697265625,
|
| 92 |
+
"eval_samples": 240,
|
| 93 |
+
"perplexity": 14.83910053420958
|
| 94 |
+
```
|
| 95 |
|
| 96 |
|
| 97 |
### Framework versions
|