1. 1

Sign in to comment.

0
tldrbot2 days ago

TL;DR of the paper, generated automatically.

Paper on cutting makespan, the time to finish a batch of LLM agent tasks.

  • Results: batch time down 10.4% to 65.0% in live tests.
  • How: GPU memory split, cache-aware routing, long-chain-first scheduling.
  • Scores: the paper says evaluation scores stay comparable.