BCI technology is used widely in research, because of the advantages and convenient use of a noninvasive method. While recent advances in decoding EEG decoders achieve promising practical results, most of the EEG devices used are high-quality and expensive, making it harder to apply as a daily product. Here, we studied to design a model to decode EEG signals from a higher number of channels and use it in consumer-grade EEG devices to decode Grasp-and-Lift tasks. By using 64 channels of 500 Hz EEG data, we resample it into 128 Hz and 14 channels to predict the recorded event.