microsoft / Deep3DFaceReconstruction

Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set (CVPRW 2019)
MIT License
2.18k stars 443 forks source link

About the learning rate #137

Closed hanssssssss closed 3 years ago

hanssssssss commented 3 years ago

Hi ! I tried to overfit a single image and I found that in my training pieline the output of the model trained under the the learning rate 1e-5 performed much better than that trained under the learning rate 1e-4. Should I used the learning rate of 1e-5 when I train on a bigger dataset(over 10k images),or 1e-4 is more suitable when training on a large amount of data? Thanks !