{"ID":2875890,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2509.01400","arxiv_id":"2509.01400","title":"Distillation of a tractable model from the VQ-VAE","abstract":"Deep generative models with discrete latent space, such as the Vector-Quantized Variational Autoencoder (VQ-VAE), offer excellent data generation capabilities, but, due to the large size of their latent space, their probabilistic inference is deemed intractable. We demonstrate that the VQ-VAE can be distilled into a tractable model by selecting a subset of latent variables with high probabilities. This simple strategy is particularly efficient, especially if the VQ-VAE underutilizes its latent space, which is, indeed, very often the case. We frame the distilled model as a probabilistic circuit, and show that it preserves expressiveness of the VQ-VAE while providing tractable probabilistic inference. Experiments illustrate competitive performance in density estimation and conditional generation tasks, challenging the view of the VQ-VAE as an inherently intractable model.","short_abstract":"Deep generative models with discrete latent space, such as the Vector-Quantized Variational Autoencoder (VQ-VAE), offer excellent data generation capabilities, but, due to the large size of their latent space, their probabilistic inference is deemed intractable. We demonstrate that the VQ-VAE can be distilled into a tr...","url_abs":"https://arxiv.org/abs/2509.01400","url_pdf":"https://arxiv.org/pdf/2509.01400v1","authors":"[\"Armin Hadžić\",\"Milan Papez\",\"Tomáš Pevný\"]","published":"2025-09-01T11:51:08Z","proceeding":"cs.LG","tasks":"[\"cs.LG\"]","methods":"[\"Variational Autoencoder\"]","has_code":false}