{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/arxiv-2605-06870","title":"Continuous First, Discrete Later: VQ-VAEs Without Dimensional Collapse","arxiv_id":"2605.06870","date":"2026-05-07","proceeding":null,"authors":["Xinyu Zhao","Nikita Karagodin","Hamed Hassani","Sinan Hersek","Paul Pu Liang","Yury Polyanskiy"],"abstract":"While many approaches to improve VQ-VAE performance focus on codebook size and utilization, the effect of dimensional collapse, where trained VQ-VAE representations live in an extremely low-dimensional subspace (1-2% of full rank), remains unaddressed. We show theoretically and empirically that dimension collapse causes a hard loss lower bound that various codebook improvement techniques fail to surpass. Our analytic framework extends the sequential learning effect of Saxe et al. [2014] by introducing ideas from rate-distortion theory and explains how the latent collapse is caused by the VQ suppressing lower-variance directions. Our theory justifies a simple solution: a \"warm-up phase\" that trains the model as an (unquantized) autoencoder before introducing VQ. On both synthetic experiments and large-scale image (VQGAN) and audio (WavTokenizer) VQ-VAEs, we show that AE Warm-Up successfully restores representation dimension, leading to lower reconstruction and perceptual loss at the same training budget. Across codebook sizes $K \\in$ {$2^{10}, 2^{14}, 2^{16}$}, AE warm-up raises VQGAN codebook effective dimension from 3-5 to 17-19 and reduces rFID by 17-35%; on WavTokenizer at $K \\in$ {$2^{13}, 2^{14}$}, it raises codebook dimension from 4 to 17-19 and improves PESQ by 11-14%. We empirically characterize how warm-up duration governs the achievable final loss. In agreement with experiment, our theoretical analysis predicts downstream performance as a function of warm-up length, enabling an adaptive criterion for switching from AE Warm-up to VQ-VAE training.","url_abs":"https://arxiv.org/abs/2605.06870","url_pdf":"https://arxiv.org/pdf/2605.06870","source":{"archive":null,"snapshot":"2025-07-28","note":"not in the Papers with Code archive (frozen at the snapshot)","row_kind":"graph","title_abstract_authors_date":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)"},"code_links":[],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2605.06870","atlas_url":null,"mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2605.06870"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-25T09:33:49+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"mentioned_in_github":null,"is_official":null,"provenance":"deterministic:regex_extraction","mentioned_in_paper":null,"url":"https://github.com/xyz-zy/vqvae_latent_span_collapse","reach":null}],"summary":{"ran":13,"unverified":3},"by_repo_kind":{"found_in_text":{"samples":16,"ran":13,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"95e19c7fd52aa466","entry":"compute_waterfill","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/eval_vqgan_waterfilling.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/eval_vqgan_waterfilling.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"95e19c7fd52aa466"}},{"code_sha256_prefix":"ab48634212ca75af","entry":"compute_waterfill","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/plot_ae_waterfilling.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/plot_ae_waterfilling.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"ab48634212ca75af"}},{"code_sha256_prefix":"5c141b2be9354bae","entry":"find_ae_ckpt","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/plot_ae_waterfilling.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/plot_ae_waterfilling.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"5c141b2be9354bae"}},{"code_sha256_prefix":"b33525f44c0b0db9","entry":"find_ckpt","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/eval_val_std.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/eval_val_std.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"b33525f44c0b0db9"}},{"code_sha256_prefix":"06acef828d85d18a","entry":"find_config_for_run","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/eval_val_std.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/eval_val_std.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"06acef828d85d18a"}},{"code_sha256_prefix":"d54d3242c568bac6","entry":"get_final_metrics","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/plot_ae_warmup_scatter.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/plot_ae_warmup_scatter.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"d54d3242c568bac6"}},{"code_sha256_prefix":"70c57936d066c47a","entry":"get_final_metrics","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/plot_ae_warmup_scatter_cbdim.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/plot_ae_warmup_scatter_cbdim.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"70c57936d066c47a"}},{"code_sha256_prefix":"67195b3a6afd3017","entry":"get_final_std","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/plot_ae_warmup_scatter.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/plot_ae_warmup_scatter.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"67195b3a6afd3017"}},{"code_sha256_prefix":"7f4da51e0bc41db4","entry":"is_main","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/train_small2.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/train_small2.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"7f4da51e0bc41db4"}},{"code_sha256_prefix":"7b2110d938733a3f","entry":"kmeans_full","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/eval_ae_kmeans_step0_allsamples.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/eval_ae_kmeans_step0_allsamples.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"7b2110d938733a3f"}},{"code_sha256_prefix":"cda9542f1cb60640","entry":"load_metric","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/plot_cb16k_3panel.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/plot_cb16k_3panel.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"cda9542f1cb60640"}},{"code_sha256_prefix":"392ea4d7aa702e8c","entry":"to_uint8","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/eval_ae_kmeans_step0_allsamples.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/eval_ae_kmeans_step0_allsamples.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"392ea4d7aa702e8c"}},{"code_sha256_prefix":"3b024007f61d19cb","entry":"to_uint8_minus1_to_1","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/train_small2.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/train_small2.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"3b024007f61d19cb"}},{"code_sha256_prefix":"6a6a9dd34a779613","entry":"find_ae_run","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/plot_ae_waterfilling.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/plot_ae_waterfilling.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"6a6a9dd34a779613"}},{"code_sha256_prefix":"a9cade3c46a1d9b5","entry":"find_run","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/plot_ae_warmup_scatter.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/plot_ae_warmup_scatter.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"a9cade3c46a1d9b5"}},{"code_sha256_prefix":"a8836f887cef4509","entry":"load_model","repo":"xyz-zy/vqvae_latent_span_collapse","repo_kind":"found_in_text","path":"VQGAN/eval_val_std.py","file_url":"https://github.com/xyz-zy/vqvae_latent_span_collapse/blob/HEAD/VQGAN/eval_val_std.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"a8836f887cef4509"}}]},"arxiv_metadata":{"licence":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)","fields":["title","abstract","authors","date"],"primary_category":"cs.LG","source":"arxiv_2026.jsonl"},"syntology_extracted_results":null}