Addressing Representation Collapse in Vector Quantized Models with One Linear Layer

Open in new window