A keras implementation of Bidirectional-LSTM_CNNs for Named-Entity-Recoganition. The original paper can be found at https://arxiv.org/abs/1511.08308
The implementation differs from the original paper in the following ways :
- lexicons are not considered
- Bucketing is used to speed up the training
- nadam optimizer used instead of SGD
The model produces a test F1_score of 90.9 % with ~70 epochs. The results produced in the paper for the given architecture is 91.14 Architecture(BILSTM-CNN with emb + caps)
python3 nn.py
1) numpy
2) Keras
3) Tensorflow