AI Paper Atlas

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

XDomainBench: Diagnosing Reasoning Collapse in High-Dimensional Scientific Knowledge Composition

Gong Zhiren, Tiantong Wu, Jiaming Zhang, Fuyao Zhang, CHE WANG, Yurong Hao, Yikun Hou, Foo Ping, Yilei Zhao, Fei Huang, Chau Yuen, Wei Yang Bryan Lim

ICML 2026 · Large language models › Agents, code and math · approach: LLMs for scientific discovery · 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
increase composition order from 1 to 4 domains (ablation)LLMs (large models mean) · XDomainBench deterministicSessionSuccess@tau: 38.7 -> 27.1Hurt
increase composition order from 1 to 4 domains (ablation)LLMs (large models mean) · XDomainBench deterministicRecall: 28.8 -> 18.2Hurt
increase composition order from 1 to 4 domains (ablation)Small models (mean) · XDomainBench deterministicRecall: 26.2 -> 17.4Hurt
increase composition order from 1 to 4 domains (ablation)MoE models (mean) · XDomainBench deterministicRecall: 60.1 -> 35.2Hurt
increase composition order from 1 to 4 domains (ablation)MoE models (mean) · XDomainBench deterministicSessionSuccess@tau: 82.1 -> 42.0Hurt
score with semantic judge instead of token recall (ablation)evaluation metric · XDomainBench, k=1 to 4accuracy: Judge 37.1 -> 26.8; Recall 31.3 -> 21.9No clear effect
use MoE models instead of dense large or small modelsMoE models · XDomainBench deterministic, k=1SessionSuccess@tau: 82.1 vs 38.7 (large mean)Improved
compose domains in turn 1 versus single-domain baselines (ablation)LLMs · XDomainBench turn-1turn-1 correctnessHurt
volatile difficulty/mixture patterns (fluctuate, spike) versus stable (ablation)LLMs · XDomainBench sessionssession collapse rateHurt

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