Note: A notebook will be added soon detailing the usage

Bi-LSTM-CRF in PyTorch

• Compared with PyTorch BI-LSTM-CRF, following changes are made:
• In the original implementation tag token indices start from '0'. '0' is also used for padding token. This leads to '0' token for both padding and a dictionary token while tokenizing the tag sequences. This can make for erroneous training. This error is fixed by adding an appropriate unused token for padding tags sequences.

Installation

• dependencies
  • Python 3
  • PyTorch
• install

  $ pip install bi-lstm-crf

Training

corpus

• prepare your corpus in the specified structure and format
• there is also a sample corpus in bi_lstm_crf/app/sample_corpus

training

$ python -m bi_lstm_crf corpus_dir --model_dir "model_xxx"

• more options
• detail of model_dir

training curve

import pandas as pd
import matplotlib.pyplot as plt

# the training losses are saved in the model_dir
df = pd.read_csv(".../model_dir/loss.csv")
df[["train_loss", "val_loss"]].ffill().plot(grid=True)
plt.show()

CRF Module

The CRF module can be easily embeded into other models:

from bi_lstm_crf import CRF

# a BERT-CRF model for sequence tagging
class BertCrf(nn.Module):
    def __init__(self, ...):
        ...
        self.bert = BERT(...)
        self.crf = CRF(in_features, num_tags)

    def loss(self, xs, tags):
        features, = self.bert(xs)
        masks = xs.gt(0)
        loss = self.crf.loss(features, tags, masks)
        return loss

    def forward(self, xs):
        features, = self.bert(xs)
        masks = xs.gt(0)
        scores, tag_seq = self.crf(features, masks)
        return scores, tag_seq

References

1. Zhiheng Huang, Wei Xu, and Kai Yu. 2015. Bidirectional LSTM-CRF Models for Sequence Tagging. arXiv:1508.01991.
2. PyTorch tutorial ADVANCED: MAKING DYNAMIC DECISIONS AND THE BI-LSTM CRF