few_rel

参考:

使用以下命令在 TFDS 中加载此数据集:

ds = tfds.load('huggingface:few_rel')
  • 说明
FewRel is a large-scale few-shot relation extraction dataset, which contains more than one hundred relations and tens of thousands of annotated instances cross different domains.
拆分 样本
'pubmed_unsupervised' 2500
'train_wiki' 44800
'val_nyt' 2500
'val_pubmed' 1000
'val_semeval' 8851
'val_wiki' 11200
  • 特征
{
    "relation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "tokens": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "head": {
        "text": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "type": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "indices": {
            "feature": {
                "feature": {
                    "dtype": "int64",
                    "id": null,
                    "_type": "Value"
                },
                "length": -1,
                "id": null,
                "_type": "Sequence"
            },
            "length": -1,
            "id": null,
            "_type": "Sequence"
        }
    },
    "tail": {
        "text": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "type": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "indices": {
            "feature": {
                "feature": {
                    "dtype": "int64",
                    "id": null,
                    "_type": "Value"
                },
                "length": -1,
                "id": null,
                "_type": "Sequence"
            },
            "length": -1,
            "id": null,
            "_type": "Sequence"
        }
    },
    "names": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    }
}

pid2name

使用以下命令在 TFDS 中加载此数据集:

ds = tfds.load('huggingface:few_rel/pid2name')
  • 说明
FewRel is a large-scale few-shot relation extraction dataset, which contains more than one hundred relations and tens of thousands of annotated instances cross different domains.
拆分 样本
'pid2name' 744
  • 特征
{
    "relation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "names": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    }
}