🧬 NTv3: A Foundation Model for Genomics
NTv3 is a series of foundational models designed to understand and generate genomic sequences. It unifies representation learning, functional prediction, and controllable sequence generation within a single, efficient U-Net-like architecture. It also enables the modeling of long-range dependencies, up to 1 Mb of context, at nucleotide resolution. Pretrained on 9 trillion base pairs, NTv3 excels at functional-track prediction and genome annotation across 24 animal and plant species. It can also be fine-tuned into a controllable generative model for genomic sequence design. This is the generative model based on NTv3, capable of context-aware DNA sequence generation with desired activity levels.It builds on the post-trained NTv3 model with MDLM based fine-tuning.For more details, please refer to the NTv3 paper.
⚖️ License Summary
- The Licensed Models are only available under this License for Non-Commercial Purposes.
- You are permitted to reproduce, publish, share and adapt the Output generated by the Licensed Model only for Non-Commercial Purposes and in accordance with this License.
- You may not use the Licensed Models or any of its Outputs in connection with:
- any Commercial Purposes, unless agreed by Us under a separate licence;
- to train, improve or otherwise influence the functionality or performance of any other third-party derivative model that is commercial or intended for a Commercial Purpose and is similar to the Licensed Models;
- to create models distilled or derived from the Outputs of the Licensed Models, unless such models are for Non-Commercial Purposes and open-sourced under the same license as the Licensed Models; or
- in violation of any applicable laws and regulations.
📋 Model Summary
- Architecture: Conditioned U-Net with adaptive layer norms + Transformer stack
- Training: Masked Discrete Language Modeling (MDLM)
- Conditioning: Species + Activity levels (0-4)
- Tokenizer: Character-level over A T C G N + special tokens
- Dependencies: transformers >= 4.55.0
- Input size: Model trained on 4096bp sequences with 249bp generation length
- Note: Custom code → use
trust_remote_code=True