HCSpec: Two-Tier Horizontal Cascade Speculative Decoding for High-Efficiency Large Language Model Inference.

Yizhou Zhang, Siming Chen, Hao Ye, Erhu Feng

Browse the full ACL paper archive.