{"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/improved-training-technique-for-latent","title":"Improved Training Technique for Latent Consistency Models","arxiv_id":"2502.01441","date":"2025-02-03","proceeding":null,"authors":["Quan Dao","Khanh Doan","Di Liu","Trung Le","Dimitris Metaxas"],"abstract":"Consistency models are a new family of generative models capable of producing high-quality samples in either a single step or multiple steps. Recently, consistency models have demonstrated impressive performance, achieving results on par with diffusion models in the pixel space. However, the success of scaling consistency training to large-scale datasets, particularly for text-to-image and video generation tasks, is determined by performance in the latent space. In this work, we analyze the statistical differences between pixel and latent spaces, discovering that latent data often contains highly impulsive outliers, which significantly degrade the performance of iCT in the latent space. To address this, we replace Pseudo-Huber losses with Cauchy losses, effectively mitigating the impact of outliers. Additionally, we introduce a diffusion loss at early timesteps and employ optimal transport (OT) coupling to further enhance performance. Lastly, we introduce the adaptive scaling-$c$ scheduler to manage the robust training process and adopt Non-scaling LayerNorm in the architecture to better capture the statistics of the features and reduce outlier impact. With these strategies, we successfully train latent consistency models capable of high-quality sampling with one or two steps, significantly narrowing the performance gap between latent consistency and diffusion models. The implementation is released here: https://github.com/quandao10/sLCT/","url_abs":"https://arxiv.org/abs/2502.01441v2","url_pdf":"https://arxiv.org/pdf/2502.01441v2.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":"improved-training-technique-for-latent","repo_url":"https://github.com/quandao10/slct","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"video-generation","task_name":"Video Generation"}],"methods":[{"method_slug":"adopt","method_name":"ADOPT"},{"method_slug":"cm","method_name":"Consistency Models"},{"method_slug":"diffusion","method_name":"Diffusion"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2502.01441","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2502.01441"}},"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":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/quandao10/slct","reach":null}],"summary":{"ran_fixture":2,"ran_honours":1,"ran_draft_wrong":4,"unverified":1},"by_repo_kind":{"official":{"samples":7,"ran":6,"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":8,"samples":[{"code_sha256_prefix":"75cca24835775136","entry":"normal_kl","repo":"quandao10/slct","repo_kind":"official","path":"models/losses.py","file_url":"https://github.com/quandao10/slct/blob/HEAD/models/losses.py","link_basis":"first_harvest_node","language":"python","status":"ran_fixture","verification_level":2,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"75cca24835775136"}},{"code_sha256_prefix":"bd3578806ecf5ac2","entry":"adjust_learning_rate","repo":"quandao10/slct","repo_kind":"official","path":"train_cm_latent_repa.py","file_url":"https://github.com/quandao10/slct/blob/HEAD/train_cm_latent_repa.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"bd3578806ecf5ac2"}},{"code_sha256_prefix":"cfd76fd0d89574a4","entry":"approx_standard_normal_cdf","repo":"quandao10/slct","repo_kind":"official","path":"models/losses.py","file_url":"https://github.com/quandao10/slct/blob/HEAD/models/losses.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"cfd76fd0d89574a4"}},{"code_sha256_prefix":"1712a07966b542ee","entry":"center_crop_arr","repo":null,"repo_kind":null,"path":null,"file_url":null,"link_basis":"identical_code_first_harvested_elsewhere","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":null,"inline_ok":false,"mcp_get_code":{"code_sha256":"1712a07966b542ee"}},{"code_sha256_prefix":"cd33283d615fb3d7","entry":"discretized_gaussian_log_likelihood","repo":"quandao10/slct","repo_kind":"official","path":"models/losses.py","file_url":"https://github.com/quandao10/slct/blob/HEAD/models/losses.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"cd33283d615fb3d7"}},{"code_sha256_prefix":"3d32915eed1b1d44","entry":"exp_decay_weight","repo":"quandao10/slct","repo_kind":"official","path":"models/karras_diffusion.py","file_url":"https://github.com/quandao10/slct/blob/HEAD/models/karras_diffusion.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"3d32915eed1b1d44"}},{"code_sha256_prefix":"b99b8334fa5b1bc7","entry":"get_weightings","repo":"quandao10/slct","repo_kind":"official","path":"models/karras_diffusion.py","file_url":"https://github.com/quandao10/slct/blob/HEAD/models/karras_diffusion.py","link_basis":"first_harvest_node","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"b99b8334fa5b1bc7"}},{"code_sha256_prefix":"8382f7e48cc8a309","entry":"create_logger","repo":"quandao10/slct","repo_kind":"official","path":"train_cm_latent_repa.py","file_url":"https://github.com/quandao10/slct/blob/HEAD/train_cm_latent_repa.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"8382f7e48cc8a309"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}