rcmalli/keras-vggface

How to finetune last layers in keras vggface?

ZER-0-NE opened this issue · 3 comments

print(tf.version)
1.8.0
print(keras.version)
2.1.6
print(keras_vggface.version)
0.5

There are really good blog posts about this.

Which of these blog posts shows how to fine-tune a pre-trained keras-vggface model? I don't see that in any of them.

you can set layer.trainable=False expect the last layers, and then re-train the last layers.