{"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/stabilize-the-latent-space-for-image","title":"Stabilize the Latent Space for Image Autoregressive Modeling: A Unified Perspective","arxiv_id":"2410.12490","date":"2024-10-16","proceeding":null,"authors":["Yongxin Zhu","Bocheng Li","Hang Zhang","Xin Li","Linli Xu","Lidong Bing"],"abstract":"Latent-based image generative models, such as Latent Diffusion Models (LDMs) and Mask Image Models (MIMs), have achieved notable success in image generation tasks. These models typically leverage reconstructive autoencoders like VQGAN or VAE to encode pixels into a more compact latent space and learn the data distribution in the latent space instead of directly from pixels. However, this practice raises a pertinent question: Is it truly the optimal choice? In response, we begin with an intriguing observation: despite sharing the same latent space, autoregressive models significantly lag behind LDMs and MIMs in image generation. This finding contrasts sharply with the field of NLP, where the autoregressive model GPT has established a commanding presence. To address this discrepancy, we introduce a unified perspective on the relationship between latent space and generative models, emphasizing the stability of latent space in image generative modeling. Furthermore, we propose a simple but effective discrete image tokenizer to stabilize the latent space for image generative modeling by applying K-Means on the latent features of self-supervised learning models. Experimental results show that image autoregressive modeling with our tokenizer (DiGIT) benefits both image understanding and image generation with the next token prediction principle, which is inherently straightforward for GPT models but challenging for other generative models. Remarkably, for the first time, a GPT-style autoregressive model for images outperforms LDMs, which also exhibits substantial improvement akin to GPT when scaling up model size. Our findings underscore the potential of an optimized latent space and the integration of discrete tokenization in advancing the capabilities of image generative models. The code is available at \\url{https://github.com/DAMO-NLP-SG/DiGIT}.","url_abs":"https://arxiv.org/abs/2410.12490v2","url_pdf":"https://arxiv.org/pdf/2410.12490v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"stabilize-the-latent-space-for-image","repo_url":"https://github.com/DAMO-NLP-SG/DiGIT","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"conditional-image-generation","task_name":"Conditional Image Generation"},{"task_slug":"image-generation","task_name":"Image Generation"},{"task_slug":"linear-probe-classification","task_name":"Linear-Probe Classification"},{"task_slug":"self-supervised-image-classification","task_name":"Self-Supervised Image Classification"},{"task_slug":"self-supervised-learning","task_name":"Self-Supervised Learning"},{"task_slug":"unconditional-image-generation","task_name":"Unconditional Image Generation"}],"methods":[{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"attention-dropout","method_name":"Attention Dropout"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"cosine-annealing","method_name":"Cosine Annealing"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"diffusion","method_name":"Diffusion"},{"method_slug":"discriminative-fine-tuning","method_name":"Discriminative Fine-Tuning"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"gpt","method_name":"GPT"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"linear-warmup-with-cosine-annealing","method_name":"Linear Warmup With Cosine Annealing"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"weight-decay","method_name":"Weight Decay"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/conditional-image-generation-on-imagenet-2","task":"Conditional Image Generation","dataset":"ImageNet 256x256","model":"DiGIT","rank_in_archive_order":1,"of":5,"metrics":{"FID":"3.39","Inception score":"205.96"},"uses_additional_data":false},{"leaderboard":"/sota/image-generation-on-imagenet-256x256","task":"Image Generation","dataset":"ImageNet 256x256","model":"DiGIT-0.7B","rank_in_archive_order":68,"of":94,"metrics":{"FID":"3.39","Inception score":"205.96"},"uses_additional_data":false},{"leaderboard":"/sota/self-supervised-image-classification-on","task":"Self-Supervised Image Classification","dataset":"ImageNet","model":"DiGIT","rank_in_archive_order":23,"of":144,"metrics":{"Number of Params":"732M","Top 1 Accuracy":"80.3%"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2410.12490","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2410.12490"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+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":[{"provenance":"deterministic:regex_extraction","url":"https://github.com/DAMO-NLP-SG/DiGIT","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran":3,"unverified":2},"by_repo_kind":{"official":{"samples":5,"ran":3,"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":"da6613ada7726e69","entry":"build_mlp","repo":"DAMO-NLP-SG/DiGIT","repo_kind":"official","path":"fairseq_user/pretrained_enc/models_pretrained_enc.py","file_url":"https://github.com/DAMO-NLP-SG/DiGIT/blob/HEAD/fairseq_user/pretrained_enc/models_pretrained_enc.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":"da6613ada7726e69"}},{"code_sha256_prefix":"59ecadda9d80170f","entry":"init_distributed_context","repo":"DAMO-NLP-SG/DiGIT","repo_kind":"official","path":"fairseq_user/distributed.py","file_url":"https://github.com/DAMO-NLP-SG/DiGIT/blob/HEAD/fairseq_user/distributed.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":"59ecadda9d80170f"}},{"code_sha256_prefix":"68797ec8ea91ebba","entry":"load_pretrained_mae","repo":"DAMO-NLP-SG/DiGIT","repo_kind":"official","path":"fairseq_user/pretrained_enc/models_pretrained_enc.py","file_url":"https://github.com/DAMO-NLP-SG/DiGIT/blob/HEAD/fairseq_user/pretrained_enc/models_pretrained_enc.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":"68797ec8ea91ebba"}},{"code_sha256_prefix":"2628094d97691b3f","entry":"load_pretrained_moco","repo":"DAMO-NLP-SG/DiGIT","repo_kind":"official","path":"fairseq_user/pretrained_enc/models_pretrained_enc.py","file_url":"https://github.com/DAMO-NLP-SG/DiGIT/blob/HEAD/fairseq_user/pretrained_enc/models_pretrained_enc.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":"2628094d97691b3f"}},{"code_sha256_prefix":"92679b16417e54ec","entry":"mask_by_random_topk","repo":"DAMO-NLP-SG/DiGIT","repo_kind":"official","path":"fairseq_user/image_generation_stage2/model_nar.py","file_url":"https://github.com/DAMO-NLP-SG/DiGIT/blob/HEAD/fairseq_user/image_generation_stage2/model_nar.py","link_basis":"plan_row","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":"92679b16417e54ec"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}