AI Paper Atlas

Research map / Large language models / Agents, code and math

EvoGM: Learning to Merge LLMs via Evolutionary Generative Optimization

Tao Jiang, Yu Xinmeng, Chenhao Yi, Yiling Wu, Yan Li, Ran Cheng, Dongmei Jiang, Jianguo Zhang

ICML 2026 · Large language models › Agents, code and math · approach: LLMs for optimization and heuristics · 0 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
replace stochastic evolutionary operators with learned generative operatorEvoGM · GLUE 8 tasks (FLAN-T5-base)average accuracy: 81.2 -> 82.4Improved
replace stochastic evolutionary operators with learned generative operatorEvoGM · 8 unseen tasks multi-taskaverage score: 0.375 -> 0.380 (vs CMA)Improved
merge experts instead of choosing the single best expert (multi-task)EvoGM · 8 unseen tasks multi-taskaverage score: 0.405 -> 0.380Hurt
multi-task fine-tuning instead of base modelMTL · 8 unseen tasksaverage score: 0.352 -> 0.330Hurt
replace PSO-Merging/Model Swarm with EvoGM on per-task searchEvoGM · 8 unseen tasks single-tasktest accuracy: best on 5 of 8 benchmarksMixed
remove multi-round expert update (w/o Rounds) (ablation)EvoGM · MMLU, NLGraphaccuracy: decline reported, no numbers in textImproved
add cycle-consistency loss (ablation)EvoGM · NLGraph and reasoning tasksaccuracy: significant drop without it, no numbers in textImproved
use dual generators instead of a single generator (ablation)EvoGM · MMLU, NLGraph, TruthQA, HellaSwagaccuracyImproved
increase population size from 10 to 30 (ablation)EvoGM · 2-task average test accuracyaccuracy: +0.0305Improved
increase generator training epochs to 600 (ablation)EvoGM · 2-task average test accuracyaccuracy: -0.0035Hurt
increase iterations per round from 3 to 8 (ablation)EvoGM · 2-task average test accuracyaccuracyNo clear effect
increase evolution rounds from 1 to 5 (ablation)EvoGM · 2-task average test accuracyaccuracyImproved

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