tner/roberta-large-wnut2017

这个模型是在数据集 tner/wnut2017 上对 roberta-large 经过微调的版本。模型的微调是通过 T-NER 的超参数搜索完成的（详见代码库中的细节）。它在测试集上取得了以下结果：

• F1（微平均）：0.5375139977603584
• 准确率（微平均）：0.6789250353606789
• 召回率（微平均）：0.4448563484708063
• F1（宏平均）：0.4734480458244917
• 准确率（宏平均）：0.59471614080646
• 召回率（宏平均）：0.4020936892146829

在测试集上的每个实体的F1分数细分如下：

• 公司：0.4065040650406504
• 群体：0.33913043478260874
• 地点：0.6715867158671587
• 个人：0.6657342657342658
• 产品：0.27999999999999997
• 艺术作品：0.4777327935222672

F1分数的置信区间通过自助法得出，如下所示：

• F1（微平均）：
  • 90%：[0.5084441265818846, 0.5659035599952082]
  • 95%：[0.5009032784561068, 0.5708361009044657]
• F1（宏平均）：
  • 90%：[0.5084441265818846, 0.5659035599952082]
  • 95%：[0.5009032784561068, 0.5708361009044657]

完整的评估结果可以在 metric file of NER 和 metric file of entity span 找到。

使用方法

可以通过 tner library 使用此模型。通过pip安装库

pip install tner

并按以下方式激活模型。

from tner import TransformersNER
model = TransformersNER("tner/roberta-large-wnut2017")
model.predict(["Jacob Collier is a Grammy awarded English artist from London"])

可以通过transformers库使用它，但目前不建议使用，因为不支持CRF层。

训练超参数

训练过程中使用了以下超参数：

• 数据集：['tner/wnut2017']
• 数据集分割：训练集
• 数据集名称：无
• 本地数据集：无
• 模型：roberta-large
• CRF：True
• 最大长度：128
• 训练轮数：15
• 批次大小：64
• 学习率：1e-05
• 随机种子：42
• 梯度累积步数：1
• 权重衰减：无
• 学习率预热步数比率：0.1
• 最大梯度范数：10.0

完整的配置信息可以在 fine-tuning parameter file 找到。

参考资料

如果您使用了 T-NER 的任何资源，请考虑引用我们的 paper 。

@inproceedings{ushio-camacho-collados-2021-ner,
    title = "{T}-{NER}: An All-Round Python Library for Transformer-based Named Entity Recognition",
    author = "Ushio, Asahi  and
      Camacho-Collados, Jose",
    booktitle = "Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations",
    month = apr,
    year = "2021",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2021.eacl-demos.7",
    doi = "10.18653/v1/2021.eacl-demos.7",
    pages = "53--62",
    abstract = "Language model (LM) pretraining has led to consistent improvements in many NLP downstream tasks, including named entity recognition (NER). In this paper, we present T-NER (Transformer-based Named Entity Recognition), a Python library for NER LM finetuning. In addition to its practical utility, T-NER facilitates the study and investigation of the cross-domain and cross-lingual generalization ability of LMs finetuned on NER. Our library also provides a web app where users can get model predictions interactively for arbitrary text, which facilitates qualitative model evaluation for non-expert programmers. We show the potential of the library by compiling nine public NER datasets into a unified format and evaluating the cross-domain and cross- lingual performance across the datasets. The results from our initial experiments show that in-domain performance is generally competitive across datasets. However, cross-domain generalization is challenging even with a large pretrained LM, which has nevertheless capacity to learn domain-specific features if fine- tuned on a combined dataset. To facilitate future research, we also release all our LM checkpoints via the Hugging Face model hub.",
}