Research included in NeurIPS 2023 reveals that a Chinese team has successfully developed the large-scale model DeWave, which translates electroencephalogram (EEG) signals into text by learning brainwave data, without the need for large equipment. DeWave's non-invasive brainwave analysis outperforms state-of-the-art methods, offering communication assistance to patients with brain paralysis. Its core technology involves the introduction of the "discrete codebook" concept, enabling the large model to interpret brainwaves with robust and generalized capabilities.