AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

DNACHUNKER: Learnable Tokenization for DNA Language Models

Taewon Kim, Jihwan Shin, Hyomin Kim, Youngmok Jung, Jonghoon Lee, Won-Chul Lee, Sungsoo Ahn, Insu Han

ICML 2026 · Large language models › Language, safety and interpretability · approach: Text generation and evaluation metrics · 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 mask protection (forced boundaries around masked bases) (ablation)DNAChunker 50M · Revised NT (hg38, linear probe)overall MCC: 0.332 -> 0.390Improved
add masked residual gating in decoder (ablation)DNAChunker 50M · Revised NT (hg38, linear probe)overall MCC: 0.353 -> 0.390Improved
add ratio loss controlling compression (ablation)DNAChunker 50M · Revised NT (hg38, linear probe)overall MCC: 0.348 -> 0.390Improved
add ratio loss controlling compression (ablation)DNAChunker 50M · Revised NT splice siteMCC: 0.123 -> 0.290Improved
replace 6-mer tokenization with learned chunking (ablation)DNAChunker 50M · Revised NT (hg38, linear probe)overall MCC: 0.347 -> 0.390Improved
replace BPE tokenization with learned chunking (ablation)DNAChunker 50M · Revised NT (hg38, linear probe)overall MCC: 0.375 -> 0.390Improved
pretrain on multispecies instead of human genome only (ablation)DNAChunker 50M · Revised NT (linear probe)overall MCC: 0.390 -> 0.448Improved
learned chunking vs BPE under multispecies pretraining (ablation)DNAChunker 50M · Revised NT (linear probe)overall MCC: +0.015 -> +0.073 lead over BPEImproved
use learned chunking instead of BPEDNAChunker · Motif fragmentationmean tokens per motif: 2.47 -> 1.22Improved
use learned chunking instead of BPEDNAChunker · Mutation robustness (ClinVar InDel benign)tokenization similarity: 0.751 -> 0.851Improved
use learned chunking instead of BPEDNAChunker · Mutation robustness (ClinVar SNV)tokenization similarity: 0.999 -> 0.999No clear effect
use learned chunking instead of BPE or single-nucleotide tokensDNAChunker 170M · 16k-bp forward FLOPsFLOPs: BPE 1.58x vs DNAChunker; single-bp 17.6xImproved

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