Petrov, N.; Andersson, S. Sparse Experts Scale Better in Efficient Mixture Architectures for Trillion Parameter Models. CPL 2026, 14 (2), 16-22. https://doi.org/10.54097/baczzj49.