Predictive Pipelined Decoding: A Compute-Latency Trade-off for Exact LLM Decoding

Add code
Jul 12, 2023

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: