---
license: cdla-permissive-2.0
task_categories:
- tabular-classification
pretty_name: avida_htnfa_onehot
tags:
- tabular
- proteomics
- binary
- Biomedical
- Benchmark
- antigenomics
configs:
- config_name: avida_htnfa_onehot
  data_files:
  - split: train
    path: data/train.parquet
  - split: validation
    path: data/val.parquet
  - split: test
    path: data/test.parquet
dataset_info:
  description: 'Binary classification: predict VHH nanobody binding to hTNFa. Single
    antigen — antigen features omitted. Features: padded integer AA token IDs of VHH
    sequence (vocab size 21, 0=PAD). Split: CD-HIT cluster-based 80/10/10 on VHH sequences
    (c=0.7, aS=0.8).'
---