Picture for Prakash Raghavendra

Prakash Raghavendra

Striking the Right Balance between Compute and Copy: Improving LLM Inferencing Under Speculative Decoding

Add code
Nov 15, 2025
Viaarxiv icon