AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

GENEB: Why Genomic Models Are Hard to Compare

Daria Ledneva, Mikhail Nuridinov, Denis Kuznetsov

ICML 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Dense features and foundation priors · 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
add multi-species instead of human-only pretraining (6 matched pairs)Transformer genomic models · GENEB macro-MCCmacro-MCC: +0.018 macro-MCC (5/6 pairs)Improved
multi-species vs human pretraining on chromatin accessibilitymatched pairs · GENEB chromatin accessibilitymacro-MCC: +0.076 (6/6 pairs)Improved
multi-species vs human pretraining on virus/phagematched pairs · GENEB virus/phagemacro-MCC: -0.025Hurt
multi-species eukaryotic vs microbial-only pretrainingNT-v2-100M-MS, Genomics-FM vs DNABERT-S · GENEB macro-MCCmacro-MCC: +0.084 macro-MCCImproved
curated eukaryotic genes vs broad multi-species pretrainingGENERator-Eukaryote-3B vs DNA-GPT-3B-M · GENEB macro-MCCmacro-MCC: +0.063 macro-MCCImproved
Transformer-decoder vs Mamba, matched corpus and tokenizationGenomeOcean-500M vs eccDNAMamba · GENEB macro-MCCmacro-MCC: 0.419 -> 0.550Improved
Transformer-encoder vs decoder, matched multi-species/BPEGENA-LM-Large-T2T vs OmniNA-220M · GENEB macro-MCCmacro-MCC: 0.425 -> 0.552Improved
Mamba vs Transformer-decoder on chromatin accessibilityeccDNAMamba vs GenomeOcean-500M · GENEB chromatin accessibilitymacro-MCC: 0.475 -> 0.599Improved
BPE vs k-mer tokenization, Transformer-encoders multi-speciesmatched pairs · GENEB macro-MCCmacro-MCC: +0.006 across 5 pairsNo clear effect
BPE vs k-mer tokenization, Transformer-decoderGenomeOcean-4B vs DNA-GPT-3B-M · GENEB macro-MCCmacro-MCC: +0.032 macro-MCCImproved
single-nucleotide vs BPE tokenization, human pretrainingMutBERT · GENEB macro-MCCmacro-MCC: +0.033 over GENA-LM, +0.038 over GROVERImproved
scale up model size (log params)40 genomic models · GENEB macro-MCCSpearman rho: rho 0.573; 0.694 excluding Evo-1-131kImproved

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