ML Anthology

Guo, Gabe

1 publications

ICLR 2026 Self-Speculative Decoding Accelerates Lossless Inference in Any-Order and Any-Subset Autoregressive Models Gabe Guo, Stefano Ermon