Data and models for IMS and GR network PhaseNet and EarthquakeTransformer

Part of the publication "Effects of Waveform Denoising and Network-Specific Training on ML-Based Seismic Phase Picking" submitted to Pageoph. The files named "adversary" use the PhaseNet model. The models and GR dataset can be loaded with seisbench (https://github.com/seisbench/seisbench), using e.g. the commands on cuda ready devices: model = sbm.PhaseNet.load("model_IMS_adversary") model = sbm.EQTransformer.load("model_eqt_bgr") The models were trained with cuda, so on cpu only devices the map_location needs to be explicitly defined: model = sbm.PhaseNet.load("model_IMS_adversary", map_location=torch.device('cpu')) model = sbm.EQTransformer.load("model_eqt_bgr", map_location=torch.device('cpu'))