Referências:
Use o seguinte comando para carregar este conjunto de dados no TFDS:
ds = tfds.load('huggingface:biosses')
- Descrição :
BIOSSES is a benchmark dataset for biomedical sentence similarity estimation. The dataset comprises 100 sentence pairs, in which each sentence was selected from the TAC (Text Analysis Conference) Biomedical Summarization Track Training Dataset containing articles from the biomedical domain. The sentence pairs were evaluated by five different human experts that judged their similarity and gave scores ranging from 0 (no relation) to 4 (equivalent).
Licença : BIOSSES é disponibilizado sob os termos da GNU Common Public License v.3.0.
Versão : 0.0.0
Divisões :
Dividir | Exemplos |
---|---|
'train' | 100 |
- Características :
{
"sentence1": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"sentence2": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"score": {
"dtype": "float32",
"id": null,
"_type": "Value"
}
}