Open Rukhmini opened 4 years ago
By default, no. It is possible to replace the output layer of the model with a binary output layer, which would let you use the remaining pre-trained weights when training the new output layer.
Hi, it seems that this issue ticket is just posted as issue template without modifying it. Do you have any reason not to close it?
Issue Template
Before posting, have you looked at the FAQ page?
Description
Please include a summary of the issue. Please include the steps to reproduce. List any additional libraries that are affected.
Steps to Reproduce
Expected behavior
A description of what you expected to happen.
Actual behavior
A description of what happens instead.
Environment