MULINLAB · POPULATION GENOMICS

Human genetic variation, modeled through haplotypes.

hEvo learns DNA representations from population-scale haplotype sequence variation to support research in variant effects, regulatory annotation, ancestry, and evolution.

MODELS · DNA REPRESENTATIONS

hEvo checkpoints

Both model cards include DNA embedding and single-nucleotide likelihood examples.

BASE · 87.1M PARAMETERS

hEVO-base-o

1,024-nt context · masked language modeling · public 1000 Genomes Project and HGDP sequences.

Read model card →

LARGE · 319.5M PARAMETERS

hEVO-large-o

1,024-nt context · masked language modeling · public 1000 Genomes Project and HGDP sequences.

Read model card →

DATA · 15 TASK BUNDLES

hEvo benchmarks

Frozen evaluation inputs or source downloaders, task-specific build scripts, and construction documentation.

Browse dataset card →