Speculative Streaming: Efficient and Scalable Speculative Decoding with Multi-Stream Attention.
Nikhil Bhendawade, Irina Belousova, Qichen Fu, Henry Mason, Antonie Lin, Mohammad Rastegari, Mahyar Najibi
Browse the full EMNLP paper archive.
Nikhil Bhendawade, Irina Belousova, Qichen Fu, Henry Mason, Antonie Lin, Mohammad Rastegari, Mahyar Najibi
Browse the full EMNLP paper archive.