Article

Calibration-free compression brings Evo 2 to its full million-token context on a single GPU

2026-08-28

Abstract excerpt

Evo 2 is the largest openly available genomic foundation model, but its forty billion parameter configuration cannot be loaded onto a single 80 GB accelerator, placing genome-scale analysis beyond most laboratories. We present TurboQuant-Bio, an open toolkit that compresses Evo 2's weights and attention cache to four bits without calibration data, and serves both through fused kernels. Compression is near-lossless...

Topics

Open a Topic to create a Post that cites this publication.

Uncategorized literature

Identifiers and source

Literature Corpus work: 8c79567a-04b3-5e0f-9913-d1de3a080869
DOI: 10.64898/2026.08.28.747902

Open publication

Related research

Semantic proximity does not establish scientific evidence. Graph snapshot: 2026-09-29.

Click a neighbor to travelStep 1 · 12 closest

Calibration-free compression brings Evo 2 to its full million-token context on a single GPUDOI 10.64898/2026.08.28.747902Read

Genome modeling and design across all domains of life with Evo 2Semantic proximity Read

DNA Compression with Genomic Language Models: Tokenization, Benchmarking, and an Information-Content MapSemantic proximity Read

A Benchmark of Evo2 Genomic AI Models for Efficient and Practical DeploymentSemantic proximity Read

Mut-BPE: A Modified BPE Strategy Improves Variant Effect PredictionSemantic proximity Read

Genome modelling and design across all domains of life with Evo 2.Semantic proximity Read

Benchmarking DNA Foundation Models for Genomic and Genetic TasksSemantic proximity Read

Sequence modeling and design from molecular to genome scale with Evo.Semantic proximity Read

Benchmarking DNA Foundation Models for zero-shot variant effect prediction: the role of context, training, and architectureSemantic proximity Read

Benchmarking DNA Foundation Models for zero-shot variant effect prediction: the role of context, training, and architectureSemantic proximity Read

NucEL: Single-Nucleotide ELECTRA-Style Genomic Pre-training for Efficient and Interpretable RepresentationsSemantic proximity Read

A deterministic computational kernel encoded in the human genomeSemantic proximity Read

Introducing STRAND: A Foundational Sequence Transformer for Range Adaptive Nucleotide DecodingSemantic proximity Read