Fig. 2From: Putting hands to rest: efficient deep CNN-RNN architecture for chemical named entity recognition with no hand-crafted rulesModel architecture. The figure illustrates the topology and of the best-performing full-featured model. \(^{1}\hbox {CNN}\) convolutional neural network; \(^{2}\hbox {HS-biGRU}\) half-stateful bidirectional gated recurrent unit; \(^{3}\hbox {FCN}\) fully-connected networkBack to article page