AI Science Tracker
EN

Accretion Labs

1 discovery from 2026

Discoveries

Accretion Labs

Chemistry helps flag stability disagreements in GNoME materials

Source check: Source checked. Not independent scientific confirmation.

Finding: Within a selected subset of GNoME materials, a fitted ranking using composition and hull distance found 45.8% of the cases where two density functional workflows disagreed, compared with 40.4% for hull distance alone. Both figures use a 20% screening budget across held-out prototype-cluster splits. Either workflow’s verdict can differ from experimental stability.

Technology: GNoME; fitted logistic regression

Family GNoME: profile · models page

Publication: Preprint · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

AI contribution
The audited population comes from GNoME candidate discovery. Fitted logistic regression used chemistry-cell disagreement rates and distance to the reference hull to rank candidates for review.
Test and result
The ranking comparison used 117,018 records initially labeled stable with 7,822 structural-prototype proxy clusters, repeated over 20 cluster splits. The target was a precomputed change in stability verdict.
Limitation / still unknown
The sample was selected for available recomputations. Relative ranking varied with chemistry and reference construction. Actual thermodynamic stability and out-of-catalog performance were untested.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://arxiv.org/abs/2610.33101v1 · https://arxiv.org/html/2610.33101v1

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 21 Nov 2026. Review scope: arXiv v1 front matter, data provenance, Sections 3 and 4.12 and limitations.

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Open the interactive profile for the timeline, or browse all organizations and model families.