# NT DNA Model

This is the DNA component of a jointly trained NT-ESM2 model pair for DNA-protein analysis.

## Model Details

- **Model Type**: Nucleotide Transformer (NT) for DNA sequences
- **Training**: Jointly trained with ESM2 protein model
- **Architecture**: Transformer-based language model for DNA

## Usage

```python
from transformers import AutoModelForMaskedLM, AutoTokenizer

# Load model and tokenizer - requires trust_remote_code
model = AutoModelForMaskedLM.from_pretrained("vsubasri/joint-nt-esm2-transcript-coding-dna", trust_remote_code=True)
tokenizer = AutoTokenizer.from_pretrained("vsubasri/joint-nt-esm2-transcript-coding-dna", trust_remote_code=True)

# Example usage
dna_sequence = "ATCGATCGATCG"
inputs = tokenizer(dna_sequence, return_tensors="pt")
outputs = model(**inputs)
```

## Training Details

- Jointly trained with protein sequences for cross-modal understanding
- Batch size: 8 (based on directory name)
- Context length: 4096 tokens
- Transcript-specific coding sequences

## Files

- `config.json`: Model configuration
- `model.safetensors`: Model weights
- `tokenizer_config.json`: Tokenizer configuration
- `vocab.txt`: Vocabulary file
- `special_tokens_map.json`: Special tokens mapping

## Citation

If you use this model, please cite the original NT paper and your joint training work.