This demo is a sequence model for assigning tags to each token in a sentence. The task is described at CONLL2000 Text Chunking task.
cd demo/sequence_tagging
./data/get_data.sh
cd demo/sequence_tagging
./train.sh
We provide two models. One is a linear CRF model (linear_crf.py) with is equivalent to the one at leon.bottou.org/projects/sgd. The second one is a stacked bidirectional RNN and CRF model (rnn_crf.py).
Model name | Number of parameters | F1 score |
---|---|---|
linear_crf | 1.8M | 0.937 |
rnn_crf | 960K | 0.941 |