ZhejiangLab/OneGenomeRice

Model

OneGenome-Rice (OGR)

5

9 commits

1 linked in READMEs

updated Apr 23, 2026

See the code

README

OneGenome-Rice (OGR)

OGR is a foundational model for AI-driven precision breeding and functional genomics in rice. It is a generative genomic foundation model trained to process DNA sequences up to 1 million base pairs in length, with 1.25B total parameters and a Mixture-of-Experts (MoE) architecture. It was pre-trained on a curated corpus of 422 rice genomes spanning cultivated and wild Oryza diversity.

For instructions, details, and examples, see the project repository OGR GitHub.

The table below summarizes training scale and key hyperparameters.

Model SpecificationOneGenomeRice (OGR)
Model Scale
Total Parameters1.25B
Activated Parameters0.33B
Architecture
ArchitectureMoE
Number of Experts8
Selected Experts per Token2
Number of Layers12
Attention Hidden Dimension1024
Number of Attention Heads16 (GQA, 8 KV groups)
MoE Hidden Dimension (per Expert)4096
Vocabulary Size128 (padded)
Context Lengthup to 1Mb
biology
mixtral
safetensors

ZhejiangLab/OneGenomeRice

Model

OneGenome-Rice (OGR)

5

9 commits

1 linked in READMEs

updated Apr 23, 2026

See the code

README

OneGenome-Rice (OGR)

OGR is a foundational model for AI-driven precision breeding and functional genomics in rice. It is a generative genomic foundation model trained to process DNA sequences up to 1 million base pairs in length, with 1.25B total parameters and a Mixture-of-Experts (MoE) architecture. It was pre-trained on a curated corpus of 422 rice genomes spanning cultivated and wild Oryza diversity.

For instructions, details, and examples, see the project repository OGR GitHub.

The table below summarizes training scale and key hyperparameters.

Model SpecificationOneGenomeRice (OGR)
Model Scale
Total Parameters1.25B
Activated Parameters0.33B
Architecture
ArchitectureMoE
Number of Experts8
Selected Experts per Token2
Number of Layers12
Attention Hidden Dimension1024
Number of Attention Heads16 (GQA, 8 KV groups)
MoE Hidden Dimension (per Expert)4096
Vocabulary Size128 (padded)
Context Lengthup to 1Mb
biology
mixtral
safetensors