AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Low-Rank Curvature for Zeroth-Order Optimization in LLM Fine-tuning

Hyunseok Seung, Jaewoo Lee, Hyunsuk Ko

AAAI 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Adapter-based parameter-efficient tuning · 3 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add RLOO leave-one-out baseline to ZO-SGD (ablation)ZO-SGD · Sphere, Rastrigin, Rosenbrock (1,000-dim)gradient estimate MSEImproved
replace MeZO with LORENMeZO · DistilBERT GLUE avgaccuracy (%): 54.4 -> 65.4Improved
replace MeZO-Adam with LORENMeZO-Adam · RoBERTa-large GLUE avgaccuracy (%): 73.6 -> 70.1Hurt
replace MeZO-Adam with LORENMeZO-Adam · GPT-2-XL GLUE avgaccuracy (%): 71.5 -> 71.9Improved
replace MeZO with LORENMeZO · OPT-2.7B GLUE avgaccuracy (%): 70.7 -> 71.6Improved
replace MeZO with LORENMeZO · LLaMA-3-8B SuperGLUE avgaccuracy (%): 63.6 -> 63.7No clear effect
replace MeZO with LORENMeZO · OPT-13B SuperGLUE avgaccuracy (%): 63.4 -> 63.5No clear effect
use LOREN instead of HiZOOHiZOO · OPT-2.7B GLUE avgaccuracy (%): 70.0 -> 71.6Improved
use six forward passes per step instead of two or threeZO optimizers · GLUE and SuperGLUEaccuracyImproved
use LOREN instead of MeZO-Adam for memoryMeZO-Adam · OPT-13B peak memory (GB)memory: 76.0 -> 57.5Improved
use LOREN instead of MeZOMeZO · OPT-13B peak memory (GB)memory: 32.9 -> 57.5Hurt
use LOREN instead of MeZO for query efficiencyMeZO · GPT-2-XL SST-2 forward passes to 70%# queries: 16,752 -> 1,320Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy