Fetching the paper…

CATE: Computation-aware Neural Architecture Encoding with Transformers · Around