EGGROLL, Unrolled: Understanding and Improving Low-Rank Evolution Strategies at Scale
This story is from 2026-09-11. It is preserved in the archive; the latest stories are on the live feed.
arXiv:2609.10980v1 Announce Type: new Abstract: EGGROLL makes evolution strategies (ES) practical for LLMs by replacing dense Gaussian weight perturbations with low-rank Gaussian products, often of rank one. This choice is computationally attractive but geometrically severe: each rank-one perturbat…
Read the full story at arXiv cs.LG ↗
Timeline · 1 report
- 2026-09-11 04:00 · arXiv cs.LG
EGGROLL, Unrolled: Understanding and Improving Low-Rank Evolution Strategies at Scale