This is the official pre-trained model introduced in DNABERT-2: Efficient Foundation Model and Benchmark For Multi-Species Genome. We sincerely appreciate the MosaicML team for the MosaicBERT implementation, which serves as the base of DNABERT-2 development.
Model Card
This is the official pre-trained model introduced in DNABERT-2: Efficient Foundation Model and Benchmark For Multi-Species Genome. We sincerely appreciate the MosaicML team for the MosaicBERT implementation, which serves as the base of DNABERT-2 development. DNABERT-2 is a transformer-based genome foundation model trained on multi-species genome. To load the model from huggingface: To calculate the embedding of a dna sequence
Excerpt from the card by Zhihan Zhou.
Configuration
- Context length (tokens)
- 512
- Layers
- 12
- Hidden size
- 768
- Feed-forward size
- 3,072
- Attention heads
- 12
- Vocabulary size
- 4,096
- Stored precision
- float32
Identity and Version
- Repository
- zhihan1996/DNABERT-2-117M
- Publisher
- Zhihan Zhou
- Task
- Not stated by the source
- Modality
- Other
- Library
- transformers
- Parameters
- Not stated by the source
- Languages
- Not stated by the source
- Revision
- 7bce263b15377fc15361f52cfab88f8b586abda0
- First published
- 2023-06-26
- Last updated
- 2025-06-30
Files and Weights
12 files, 468.6 MB in total. The weights are 1 file totalling 468.4 MB in bin.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| pytorch_model.bin | Weights | 468.4 MB | 7ff39ec77a48 |
| bert_layers.py | Configuration | 40.7 KB | — |
| bert_padding.py | Configuration | 6.1 KB | — |
| config.json | Configuration | 904 B | — |
| configuration_bert.py | Configuration | 1.0 KB | — |
| flash_attn_triton.py | Configuration | 42.7 KB | — |
| generation_config.json | Configuration | 90 B | — |
| LICENSE | Documentation | 11.4 KB | — |
| README.md | Documentation | 1.3 KB | — |
| .gitattributes | Repository | 1.5 KB | — |
| tokenizer.json | Tokenizer | 167.9 KB | — |
| tokenizer_config.json | Tokenizer | 158 B | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 468.4 MB
Released by Zhihan Zhou through its official repository on Hugging Face.
Built From
- Described by arXiv:2306.15006
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 468.4 MB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About DNABERT-2-117M
What is DNABERT-2-117M's context length?
512 tokens, from the maximum position embeddings in its published configuration.