tner/deberta-v3-large-ontonotes5

This model is a fine-tuned version of microsoft/deberta-v3-large on the tner/ontonotes5 dataset. Model fine-tuning is done via T-NER 's hyper-parameter search (see the repository for more detail). It achieves the following results on the test set:

• F1 (micro): 0.9069623608411381
• Precision (micro): 0.902100360312857
• Recall (micro): 0.9118770542773386
• F1 (macro): 0.834586960779896
• Precision (macro): 0.8237351069457466
• Recall (macro): 0.8475169311172334

The per-entity breakdown of the F1 score on the test set are below:

• cardinal_number: 0.853475935828877
• date: 0.8815545959284392
• event: 0.8030303030303031
• facility: 0.7896678966789669
• geopolitical_area: 0.9650033867690223
• group: 0.9337209302325581
• language: 0.8372093023255814
• law: 0.6756756756756757
• location: 0.7624020887728459
• money: 0.8818897637795275
• ordinal_number: 0.8635235732009926
• organization: 0.914952751528627
• percent: 0.9
• person: 0.9609866599546942
• product: 0.7901234567901234
• quantity: 0.8161434977578474
• time: 0.674364896073903
• work_of_art: 0.7188405797101449

For F1 scores, the confidence interval is obtained by bootstrap as below:

• F1 (micro):
  • 90%: [0.9019409960743083, 0.911751130722053]
  • 95%: [0.9010822890967028, 0.9125611412371442]
• F1 (macro):
  • 90%: [0.9019409960743083, 0.911751130722053]
  • 95%: [0.9010822890967028, 0.9125611412371442]

Full evaluation can be found at metric file of NER and metric file of entity span .

Usage

This model can be used through the tner library . Install the library via pip

pip install tner

and activate model as below.

from tner import TransformersNER
model = TransformersNER("tner/deberta-v3-large-ontonotes5")
model.predict(["Jacob Collier is a Grammy awarded English artist from London"])

It can be used via transformers library but it is not recommended as CRF layer is not supported at the moment.

Training hyperparameters

The following hyperparameters were used during training:

• dataset: ['tner/ontonotes5']
• dataset_split: train
• dataset_name: None
• local_dataset: None
• model: microsoft/deberta-v3-large
• crf: True
• max_length: 128
• epoch: 15
• batch_size: 16
• lr: 1e-05
• random_seed: 42
• gradient_accumulation_steps: 4
• weight_decay: 1e-07
• lr_warmup_step_ratio: 0.1
• max_grad_norm: 10.0

The full configuration can be found at fine-tuning parameter file .

Reference

If you use any resource from T-NER, please consider to cite our 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.",
}