seq_labeling_ner

Sequential labeling based named entity recognition toolkit

Available Models

• BERT
• RoBERTa (only pretokenize=False)
• DeBERTa (only pretokenize=False)
• ModernBERT

Usage

Instllation

git clone [email protected]:YuSawan/seq_labeling_ner.git
cd seq_labeling_ner
python -m venv .venv
source .venv/bin/activate
pip install .

Dataset preparation

Dataset

{
  "id": "doc-001",
  "examples": [
    {
      "id": "doc-001-P1",
      "text": "She graduated from NAIST.",
      "entities": [
        {
          "start": 19,
          "end": 24,
          "label": "ORG"
        }
      ],
      # Optional
      "word_positions": [
        [0, 3], [4, 13], [14, 18], [19, 24], [24, 25]
      ]
    }
  ]
}

Finetuning

BERT enclosed in parentheses () has its parameters frozen.

BERT Token Classification

python src/cli/run.py \
    --do_train \
    --do_eval \
    --do_predict \
    --config_file configs/config.yaml
    --output_dir ./output/

BERT-CRF

python src/cli/run.py \
    --do_train \
    --do_eval \
    --do_predict \
    --config_file configs/config_bert-crf.yaml
    --output_dir ./output/

(BERT)-CRF

python src/cli/run.py \
    --do_train \
    --do_eval \
    --do_predict \
    --config_file configs/config_crf.yaml
    --output_dir ./output/

(BERT)-BiLSTM

python src/cli/run.py \
    --do_train \
    --do_eval \
    --do_predict \
    --config_file configs/config_bert-lstm.yaml
    --output_dir ./output/

(BERT)-BiLSTM-CRF

python src/cli/run.py \
    --do_train \
    --do_eval \
    --do_predict \
    --config_file configs/config_bert-lstm-crf.yaml
    --output_dir ./output/