SpectraFormerRaman Spectra Unmixing
Transformer-based Machine Learning model for graphene buffer layer on SiC substrate
Skip to content
What is SpectraFormer? ​
SpectraFormer is a transformer-based model for spectral unmixing of Raman data from epitaxial graphene on SiC. It learns to predict and subtract the SiC substrate contribution, isolating the buffer layer signal.
The model masks the SiC spectral region, predicts the missing signal using self-attention, and subtracts it from the original spectrum to extract the graphene/buffer layer contribution.
Ready to try it? Check out the Installation guide.