metadata

license: mit
language:
  - ja
tags:
  - generated_from_trainer
metrics:
  - f1
model-index:
  - name: xlm-roberta-ner-ja
    results: []
widget:
  - text: 鈴木は4月の陽気の良い日に、鈴をつけて熊本県の阿蘇山に登った
  - text: 中国では、中国共産党による一党統治が続く

xlm-roberta-ner-ja

(Japanese caption : 日本語の固有表現抽出のモデル)

This model is a fine-tuned NER (named entity recognition) token classification model of xlm-roberta-base (pre-trained cross-lingual RobertaModel) on Wikipedia Japanese NER dataset by Stockmark Inc.
See here for the license of this dataset.

Intended uses & limitations

from transformers import AutoModelForTokenClassification
from transformers import pipeline

model_name = "tsmatz/xlm-roberta-ner-ja"
model = AutoModelForTokenClassification.from_pretrained(model_name)

classifier = pipeline("token-classification", model=model_name)
classifier("鈴木は4月の陽気の良い日に、鈴をつけて熊本県の阿蘇山に登った")

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 5e-05
train_batch_size: 12
eval_batch_size: 12
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	F1
No log	1.0	446	0.1510	0.8457
No log	2.0	892	0.0626	0.9261
No log	3.0	1338	0.0366	0.9580
No log	4.0	1784	0.0196	0.9792
No log	5.0	2230	0.0173	0.9864

Framework versions

Transformers 4.23.1
Pytorch 1.12.1+cu102
Datasets 2.6.1
Tokenizers 0.13.1