Speculative Decoding via Early-exiting for Faster LLM Inference with Thompson Sampling Control Mechanism

Add code
Jun 06, 2024

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: