Trained AE-model and data sets

This collection contains the datasets and the trained Autoencoder using the transient J-V simulations for the project that uses AE to estimate parameters of carbon-based PSCs.x_train, x_val and x_test are the respective input data in form of the simulated currents for the different scan speeds (100, 50, 5, 0.5, 0.05, 0.005) V/s, grouped in one array per simulation. Values for the voltage are not specifically given; they are 0V-1.2V-0V (forward scan and backward scan) in steps of 0.02V. y_train, y_val and y_test are the corresponding simulation parameters for the input data (surface recombination velocity, e & h mobilities in perovskite, e & h recombination lifetimes in perovskite, electron mobility in TiO2, anion & cation densities in perovskite, cation mobility in perovskite).The fully trained AE is given, as well as its individual encoder (for parameter estimation) and decoder (for J-V simulation) parts are given as well.current_df_info.csv contains information to re-transform the currents from normalized values back to mA/cm2, log_param_df_info.csv is for the re-transformation of the log-normalized parameters (cm/s, cm2/Vs, ns, cm2/Vs, cm2/Vs, cm-3).