Amir Ziashahabi

Learned latent protein languages unlock autoregressive transformer scaling and drastically accelerate protein structure generation, outperforming tokenized amino acids and enabling faster, more diverse backbone generation.

This work proposes ASPIRE, a non-synchronized batched self-speculative decoding framework built on three components, which achieves speedup in decoding throughput over autoregressive baselines and improves average speedup by approximately $27\% over the strongest prior self-speculative baselines.