1 paper
Taehyeon Kim, Ananda Theertha Suresh, Kishore Papineni +3
Despite the remarkable strides made by autoregressive language models, their potential is often hampered by the slow inference speeds inherent in sequential token generation. Block…