{"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/input-perturbation-reduces-exposure-bias-in","title":"Input Perturbation Reduces Exposure Bias in Diffusion Models","arxiv_id":"2301.11706","date":"2023-01-27","proceeding":null,"authors":["Mang Ning","Enver Sangineto","Angelo Porrello","Simone Calderara","Rita Cucchiara"],"abstract":"Denoising Diffusion Probabilistic Models have shown an impressive generation quality, although their long sampling chain leads to high computational costs. In this paper, we observe that a long sampling chain also leads to an error accumulation phenomenon, which is similar to the exposure bias problem in autoregressive text generation. Specifically, we note that there is a discrepancy between training and testing, since the former is conditioned on the ground truth samples, while the latter is conditioned on the previously generated results. To alleviate this problem, we propose a very simple but effective training regularization, consisting in perturbing the ground truth samples to simulate the inference time prediction errors. We empirically show that, without affecting the recall and precision, the proposed input perturbation leads to a significant improvement in the sample quality while reducing both the training and the inference times. For instance, on CelebA 64$\\times$64, we achieve a new state-of-the-art FID score of 1.27, while saving 37.5% of the training time. The code is publicly available at https://github.com/forever208/DDPM-IP","url_abs":"https://arxiv.org/abs/2301.11706v3","url_pdf":"https://arxiv.org/pdf/2301.11706v3.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":"input-perturbation-reduces-exposure-bias-in","repo_url":"https://github.com/forever208/ddpm-ip","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"denoising","task_name":"Denoising"},{"task_slug":"image-generation","task_name":"Image Generation"},{"task_slug":"text-generation","task_name":"Text Generation"}],"methods":[{"method_slug":"diffusion","method_name":"Diffusion"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/image-generation-on-celeba-64x64","task":"Image Generation","dataset":"CelebA 64x64","model":"DDPM-IP","rank_in_archive_order":1,"of":39,"metrics":{"FID":"1.27"},"uses_additional_data":false},{"leaderboard":"/sota/image-generation-on-ffhq-128-x-128","task":"Image Generation","dataset":"FFHQ 128 x 128","model":"DDPM-IP","rank_in_archive_order":1,"of":3,"metrics":{"FID":"2.98"},"uses_additional_data":false},{"leaderboard":"/sota/image-generation-on-imagenet-32x32","task":"Image Generation","dataset":"ImageNet 32x32","model":"DDPM-IP","rank_in_archive_order":4,"of":35,"metrics":{"FID":"2.66"},"uses_additional_data":false},{"leaderboard":"/sota/image-generation-on-lsun-tower-64x64","task":"Image Generation","dataset":"LSUN tower 64x64","model":"DDPM-IP","rank_in_archive_order":1,"of":1,"metrics":{"FID":"2.60"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=2301.11706","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2301.11706"}},"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/forever208/DDPM-IP","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/forever208/ddpm-ip","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran_honours":2,"unverified":1},"by_repo_kind":{"official":{"samples":1,"ran":0,"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":2,"samples":[{"code_sha256_prefix":"2ab2316ac6fdd869","entry":"betas_for_alpha_bar","repo":null,"repo_kind":null,"path":null,"file_url":null,"link_basis":"identical_code_first_harvested_elsewhere","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":null,"inline_ok":false,"mcp_get_code":{"code_sha256":"2ab2316ac6fdd869"}},{"code_sha256_prefix":"a086d6286a40b889","entry":"get_named_beta_schedule","repo":null,"repo_kind":null,"path":null,"file_url":null,"link_basis":"identical_code_first_harvested_elsewhere","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":null,"inline_ok":false,"mcp_get_code":{"code_sha256":"a086d6286a40b889"}},{"code_sha256_prefix":"4377cbccf9f10667","entry":"GaussianDiffusion","repo":"forever208/ddpm-ip","repo_kind":"official","path":"guided_diffusion/gaussian_diffusion.py","file_url":"https://github.com/forever208/ddpm-ip/blob/HEAD/guided_diffusion/gaussian_diffusion.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":"4377cbccf9f10667"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}