---
library_name: transformers
tags:
- rna
- language-model
- splicing
license: cc-by-4.0
---

# SpliceBERT-510nt

Minimal HuggingFace port of the **vertebrate-510nt** variant of
[SpliceBERT](https://github.com/biomed-AI/SpliceBERT) -- a BERT-based RNA
language model trained with masked language modeling on fixed-length 510 nt
fragments from vertebrate primary RNA sequences.

**WARNING:** This model was trained on exactly 510 nt of input (excluding
[CLS] and [SEP]). Sequences of other lengths were not validated upstream and
may not work properly without fine-tuning.
For general-purpose RNA embedding, use [SpliceBERT-1024nt](https://huggingface.co/Taykhoom/SpliceBERT-1024nt) instead.

## Architecture

| Parameter | Value |
|---|---|
| Layers | 6 |
| Attention heads | 16 |
| Embedding dimension | 512 |
| FFN hidden dimension | 2048 (GELU) |
| Vocabulary size | 10 |
| Positional encoding | Learned absolute |
| Normalization | LayerNorm (post-residual, eps=1e-12) |
| Architecture | Post-LN BERT encoder |
| Max sequence length | 510 nt (512 tokens; fixed-length training) |
| Checkpoint size | ~19.5M parameters |

**Vocabulary:** `[PAD]`=0, `[UNK]`=1, `[CLS]`=2, `[SEP]`=3,
`[MASK]`=4, `N`=5, `A`=6, `C`=7, `G`=8, `T`=9. Input `U` is
normalized to `T`.

## Pretraining

- **Objective:** Masked language modeling (MLM)
- **Data:** >2 million vertebrate primary RNA sequences from 72 species
- **Sequence format:** Single-nucleotide tokenization with spaces; U converted to T; fixed 510 nt fragments
- **Source checkpoint:** `SpliceBERT.510nt/pytorch_model.bin` (from [zenodo:7995778](https://doi.org/10.5281/zenodo.7995778))

### Checkpoint selection

The 510nt vertebrate variant is intended for splice site prediction tasks where exact
510 nt windows are used (e.g., centered on a splice site). For variable-length sequences
use [SpliceBERT-1024nt](https://huggingface.co/Taykhoom/SpliceBERT-1024nt).

## Parity Verification

Hidden-state representations verified (max abs diff < 1e-5) against the original
checkpoint at all 7 representation levels (embedding + 6 transformer layers),
for both `eager` and `sdpa` attention backends.
Verified on GPU with PyTorch 2.7.1 / CUDA 12.9.

## Related Models

See the full [SpliceBERT collection](https://huggingface.co/collections/Taykhoom/splicebert-6a20b72e9bec05b79ce009aa).

| Model | Context | Training data | Notes |
|---|---|---|---|
| [SpliceBERT-1024nt](https://huggingface.co/Taykhoom/SpliceBERT-1024nt) | 1024 nt | 72 vertebrates | Variable-length; general purpose |
| **[SpliceBERT-510nt](https://huggingface.co/Taykhoom/SpliceBERT-510nt)** | 510 nt (fixed) | 72 vertebrates | This model |
| [SpliceBERT-human-510nt](https://huggingface.co/Taykhoom/SpliceBERT-human-510nt) | 510 nt (fixed) | Human only | Human-specific |

## Usage

```python
import torch
from transformers import AutoTokenizer, AutoModel

tokenizer = AutoTokenizer.from_pretrained("Taykhoom/SpliceBERT-510nt", trust_remote_code=True)
model = AutoModel.from_pretrained("Taykhoom/SpliceBERT-510nt", trust_remote_code=True)
model.eval()

# The model was trained on exactly 510 nt; tokenizer handles U->T automatically
seq = ("ATCGATCG" * 64)[:510]  # exactly 510 nt
enc = tokenizer(seq, return_tensors="pt")

with torch.no_grad():
    out = model(**enc, output_hidden_states=True)

hidden = out.last_hidden_state[0]  # (512, 512)
token_emb = hidden[1:-1]           # strip [CLS] and [SEP] -> (510, 512)
mean_emb = token_emb.mean(dim=0)   # (512,)
```

### Fine-tuning

Standard HF conventions. For splice site prediction, token-level classification
using all 510 token positions (excluding special tokens) is the typical setup.

## Implementation Notes

The original checkpoint was saved as `BertForMaskedLM` with `transformers==4.20.1`.
This port uses [BERT-updated](https://huggingface.co/Taykhoom/BERT-updated), which
adds `attn_implementation="sdpa"` and `attn_implementation="flash_attention_2"` support
not present in the original codebase.

The pooler weights (`pooler.dense`) are not present in the original checkpoint and are
not included in the saved `model.safetensors`. `add_pooling_layer=True` (the default)
allocates the pooler layer but its weights are randomly initialized -- do not use
`pooler_output` without fine-tuning.

## Citation

```bibtex
@article{chen2024_splicebert,
  title   = {Self-supervised learning on millions of primary {RNA} sequences from 72 vertebrates improves sequence-based {RNA} splicing prediction},
  author  = {Chen, Ken and Zhou, Yue and Ding, Maolin and Wang, Yu and Ren, Zhixiang and Yang, Yuedong},
  journal = {Briefings in Bioinformatics},
  volume  = {25},
  number  = {3},
  pages   = {bbae163},
  year    = {2024},
  doi     = {10.1093/bib/bbae163}
}
```

## Credits

Original model and code by Chen et al. Source:
[GitHub](https://github.com/biomed-AI/SpliceBERT).
The HF conversion code was authored primarily by [Claude Code](https://claude.ai/code)
and reviewed manually by Taykhoom Dalal.

## License

The checkpoint weights are distributed under
[CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) by the upstream
[Zenodo record](https://doi.org/10.5281/zenodo.7995778). The original
SpliceBERT source code is BSD 3-Clause licensed.