I need to essentially train the first half of a CNN model (Resnet50) on one dataset, before taking those weights and applying them back to the Full Resnet50 Arch for the corresponding layers (i.e. first half) for continued training.
More specifically, the steps are as follows:
1) Get model (Resnet50, we will call this the full model here so the remaining instructions are clear)
2) Split this full model to obtain the first half of it (we will call this the split model)
3) Train this split model on some toy data
4) Assign weights from split model to corresponding layers of full model
Cloud computer and toy data will be provided.
Task to be completed using PyTorch and Jupyter Notebook, with comments so the method can be used again.
To be completed by 16/03/19
6 freelancer đang chào giá trung bình £157 cho công việc này
Hi. I am an expert in machine learning. I have many experiences in CNN of object detection. We can discuss details via chat. I wait for you now. Thanks.