I need help at a implementation of Transfer Learning in TensorFlow - one time project.
I've got a code from github: [login to view URL]
All neural networks/solvers are implementated and are tested to approximate some kind of equations (called PDE). The approximation is good. The main objective of this job is to reduce training time with Transfer Learning. Mathbackground is a bonus, but in my eyes not a must as long as you can understand the code.
Furthermore good documentation is needed, so I can do TransferLearning for different models with different pretrained models in the existing code from GitHub on my own. So it is sufficient to implement TransferLearning only between 2 models. As you can see in the code, the networks used for this are custom made.
I look forward to your proposal. For more details message me.