Commit ·
daf0177
1
Parent(s): 48317c6
Update README.md
Browse files
README.md
CHANGED
|
@@ -121,7 +121,8 @@ The training set was also augmented using:
|
|
| 121 |
The following are the hyperparameters used for training:
|
| 122 |
|
| 123 |
- **Training regime:** fp32
|
| 124 |
-
- **
|
|
|
|
| 125 |
- **Learning rate**: 1e-3
|
| 126 |
- **Learning rate scheduler**: Reduce on plateau which monitors validation loss with patience of 2 and decay rate of 0.1
|
| 127 |
- **Batch size**: 64
|
|
|
|
| 121 |
The following are the hyperparameters used for training:
|
| 122 |
|
| 123 |
- **Training regime:** fp32
|
| 124 |
+
- **Loss:** Cross entropy
|
| 125 |
+
- **Optimizer**: Adam with default betas (0.99, 0.999)
|
| 126 |
- **Learning rate**: 1e-3
|
| 127 |
- **Learning rate scheduler**: Reduce on plateau which monitors validation loss with patience of 2 and decay rate of 0.1
|
| 128 |
- **Batch size**: 64
|