CvT-Design-Photonics

Applying latest development in vision transformers (ViT, CvT...) to the design of nanophotonic structures.

Demo experiments include photonic crystals (L3 nanocavity, nanobeam...).

Results using CvT here have proven to be superior to those reported previously for CNNs. For a quick look of the CNN results, visit: https://github.com/Arcadianlee/Deep-Learning-Design-Photonic-Crystals

Provide feedback

Name		Name	Last commit message	Last commit date
Latest commit History 9 Commits
Dataset		Dataset
V train 0.854 V test 0.805 lr 0.0001		V train 0.854 V test 0.805 lr 0.0001
V train 0.871 V test 0.820 lr 0.001 (1 stage)		V train 0.871 V test 0.820 lr 0.001 (1 stage)
V train 0.909 V test 0.886 lr 0.01		V train 0.909 V test 0.886 lr 0.01
V train 0.911 V test 0.877 lr 0.01		V train 0.911 V test 0.877 lr 0.01
V train 0.915 V test 0.876 lr 0.0025		V train 0.915 V test 0.876 lr 0.0025
V train 0.915 V test 0.886 lr 0.01 (2 stages)		V train 0.915 V test 0.886 lr 0.01 (2 stages)
V train 0.919 V test 0.871 lr 0.001 (2 stages)		V train 0.919 V test 0.871 lr 0.001 (2 stages)
V train 0.924 V test 0.882 lr 0.01		V train 0.924 V test 0.882 lr 0.01
V train 0.929 V test 0.853 lr 0.00025		V train 0.929 V test 0.853 lr 0.00025
V train 0.971 V test 0.888 lr 0.001		V train 0.971 V test 0.888 lr 0.001
V train 0.990 V test 0.882 lr 0.01 fake		V train 0.990 V test 0.882 lr 0.01 fake
dependencies		dependencies
LICENSE		LICENSE
README.md		README.md