AWS and Booking.com Win ACL Best Paper by Making a 70B Model Run 20x Faster
· techtimes.com · popularity 21/100 · 0 similar stories today
Read the original at techtimes.com →
Summary
LLM inference optimization reached a validated milestone: AWS and Booking.com combined knowledge distillation and Medusa speculative decoding to run a 70B model 10-20x faster at over 90% quality in three live Booking.com A/B tests, winning the ACL 2025 Industry Track best paper from 421 competing submissions.
Found in: Tech Times. Rating basis: single outlet; outlet tier 2 (unlisted); 1 feeds; feed position 2.