---
dataset_info:
  features:
  - name: labels
    dtype: string
  - name: seqs
    dtype: string
  splits:
  - name: train
    num_bytes: 8259340114
    num_examples: 17070828
  - name: test
    num_bytes: 138308
    num_examples: 474
  - name: valid
    num_bytes: 142890
    num_examples: 474
  download_size: 7022292725
  dataset_size: 8259621312
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
  - split: valid
    path: data/valid-*
---

This dataset is from the ProstT5 project ([paper](https://huggingface.co/datasets/Rostlab/ProstT5Dataset) | [code](https://github.com/mheinzinger/ProstT5)).

The only modification made is that it was decoded with their tokenizer from [here](https://huggingface.co/datasets/Rostlab/ProstT5Dataset) to yield the raw sequences.