This github contains the state-of-the-art complex word identification model SEQ from Gooding and Kochmar (2019).
The dataset used to train these models was collected by Yimam et al., (2018) and is available here.
The system output for both SEQ and the winning shared task submission CAMB is available in Sytem Output.