{"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-2605-27102","title":"JLT: Clean-Latent Prediction in Latent Diffusion Transformers","arxiv_id":"2605.27102","date":"2026-05-26","proceeding":null,"authors":["Funing Fu","Tenghui Wang","Guanyu Zhou","Junyong Cen","Qichao Zhu"],"abstract":"Flow matching with clean-data prediction has shown that regressing the clean point can exploit low-dimensional structure more effectively than predicting an ambient noised quantity. We ask whether this principle remains useful after images are mapped into a learned latent space, where compression has already removed much of the raw pixel variability. We introduce JLT, a 130M latent diffusion Transformer over frozen FLUX.2 VAE codes, and compare clean-latent prediction with a matched velocity-prediction DiT under the same representation, backbone, and training settings. Although the three variables x, epsilon, and v are linearly convertible for a fixed corruption time, a local Gaussian analysis shows that velocity regression inherits an isotropic target-covariance floor and amplifies low-variance latent directions, while clean prediction damps them. On ImageNet 256 x 256, JLT-B/1 obtains FID-50K 2.50 with classifier-free guidance, with a large matched-target gap over velocity prediction. These results suggest that prediction targets in latent diffusion are representation-dependent geometric choices, rather than interchangeable algebraic parameterizations.","url_abs":"https://arxiv.org/abs/2605.27102","url_pdf":"https://arxiv.org/pdf/2605.27102","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":{"atlas_url":"https://app.syntology.ai/?focus=2605.27102","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2605.27102"}},"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":[{"mentioned_in_github":null,"is_official":null,"provenance":"deterministic:regex_extraction","mentioned_in_paper":null,"url":"https://github.com/akatsuki-neo/JLT","reach":null}],"summary":{"ran":5,"ran_honours":1,"ran_draft_wrong":2,"unverified":1},"by_repo_kind":{"found_in_text":{"samples":9,"ran":8,"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":"3e2b8a000ecc20ca","entry":"add_weight_decay","repo":"akatsuki-neo/JLT","repo_kind":"found_in_text","path":"misc.py","file_url":"https://github.com/akatsuki-neo/JLT/blob/HEAD/misc.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":"3e2b8a000ecc20ca"}},{"code_sha256_prefix":"3a774664bfc712ec","entry":"adjust_learning_rate","repo":"akatsuki-neo/JLT","repo_kind":"found_in_text","path":"util/lr_sched.py","file_url":"https://github.com/akatsuki-neo/JLT/blob/HEAD/util/lr_sched.py","link_basis":"harvester_set","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":"3a774664bfc712ec"}},{"code_sha256_prefix":"3dc19396537db789","entry":"all_reduce_mean","repo":"akatsuki-neo/JLT","repo_kind":"found_in_text","path":"misc.py","file_url":"https://github.com/akatsuki-neo/JLT/blob/HEAD/misc.py","link_basis":"harvester_set","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":"3dc19396537db789"}},{"code_sha256_prefix":"f50a8d20efb35c68","entry":"broadcat","repo":"akatsuki-neo/JLT","repo_kind":"found_in_text","path":"util/model_util.py","file_url":"https://github.com/akatsuki-neo/JLT/blob/HEAD/util/model_util.py","link_basis":"plan_row","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"f50a8d20efb35c68"}},{"code_sha256_prefix":"8bad335789e204f6","entry":"build_shard_index","repo":"akatsuki-neo/JLT","repo_kind":"found_in_text","path":"latent_dataset.py","file_url":"https://github.com/akatsuki-neo/JLT/blob/HEAD/latent_dataset.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":"8bad335789e204f6"}},{"code_sha256_prefix":"dadc1fccf5031ecc","entry":"build_vae","repo":"akatsuki-neo/JLT","repo_kind":"found_in_text","path":"vae.py","file_url":"https://github.com/akatsuki-neo/JLT/blob/HEAD/vae.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":"dadc1fccf5031ecc"}},{"code_sha256_prefix":"1712a07966b542ee","entry":"center_crop_arr","repo":"akatsuki-neo/JLT","repo_kind":"found_in_text","path":"util/crop.py","file_url":"https://github.com/akatsuki-neo/JLT/blob/HEAD/util/crop.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"1712a07966b542ee"}},{"code_sha256_prefix":"58823d9435a8751b","entry":"rotate_half","repo":"akatsuki-neo/JLT","repo_kind":"found_in_text","path":"util/model_util.py","file_url":"https://github.com/akatsuki-neo/JLT/blob/HEAD/util/model_util.py","link_basis":"plan_row","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"58823d9435a8751b"}},{"code_sha256_prefix":"5c1a59ed3feed147","entry":"modulate","repo":"akatsuki-neo/JLT","repo_kind":"found_in_text","path":"model_jit.py","file_url":"https://github.com/akatsuki-neo/JLT/blob/HEAD/model_jit.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":"5c1a59ed3feed147"}}]},"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}