{"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/clear-conv-like-linearization-revs-pre","title":"CLEAR: Conv-Like Linearization Revs Pre-Trained Diffusion Transformers Up","arxiv_id":"2412.16112","date":"2024-12-20","proceeding":null,"authors":["Songhua Liu","Zhenxiong Tan","Xinchao Wang"],"abstract":"Diffusion Transformers (DiT) have become a leading architecture in image generation. However, the quadratic complexity of attention mechanisms, which are responsible for modeling token-wise relationships, results in significant latency when generating high-resolution images. To address this issue, we aim at a linear attention mechanism in this paper that reduces the complexity of pre-trained DiTs to linear. We begin our exploration with a comprehensive summary of existing efficient attention mechanisms and identify four key factors crucial for successful linearization of pre-trained DiTs: locality, formulation consistency, high-rank attention maps, and feature integrity. Based on these insights, we introduce a convolution-like local attention strategy termed CLEAR, which limits feature interactions to a local window around each query token, and thus achieves linear complexity. Our experiments indicate that, by fine-tuning the attention layer on merely 10K self-generated samples for 10K iterations, we can effectively transfer knowledge from a pre-trained DiT to a student model with linear complexity, yielding results comparable to the teacher model. Simultaneously, it reduces attention computations by 99.5% and accelerates generation by 6.3 times for generating 8K-resolution images. Furthermore, we investigate favorable properties in the distilled attention layers, such as zero-shot generalization cross various models and plugins, and improved support for multi-GPU parallel inference. Models and codes are available here: https://github.com/Huage001/CLEAR.","url_abs":"https://arxiv.org/abs/2412.16112v1","url_pdf":"https://arxiv.org/pdf/2412.16112v1.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":"clear-conv-like-linearization-revs-pre","repo_url":"https://github.com/huage001/clear","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"clear-conv-like-linearization-revs-pre","repo_url":"https://github.com/huage001/linfusion","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":null,"task_name":"8k"},{"task_slug":null,"task_name":"GPU"},{"task_slug":"image-generation","task_name":"Image Generation"},{"task_slug":"zero-shot-generalization","task_name":"Zero-shot Generalization"}],"methods":[{"method_slug":"attention","method_name":"Attention"},{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2412.16112","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2412.16112"}},"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/huage001/clear","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/huage001/linfusion","reach":null}],"summary":{"ran_violates":1,"ran_draft_wrong":1,"unverified":11},"by_repo_kind":{"official":{"samples":13,"ran":2,"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":"4e3ad3679e5c37b2","entry":"calculate_shift","repo":"huage001/clear","repo_kind":"official","path":"pipeline_flux_img2img.py","file_url":"https://github.com/huage001/clear/blob/HEAD/pipeline_flux_img2img.py","link_basis":"harvester_set","language":"python","status":"ran_violates","verification_level":1,"contract_check":"VIOLATES","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"4e3ad3679e5c37b2"}},{"code_sha256_prefix":"0b56680ee1708fee","entry":"retrieve_timesteps","repo":"huage001/clear","repo_kind":"official","path":"pipeline_flux_img2img.py","file_url":"https://github.com/huage001/clear/blob/HEAD/pipeline_flux_img2img.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"0b56680ee1708fee"}},{"code_sha256_prefix":"667cd02575574903","entry":"c_crop","repo":"huage001/clear","repo_kind":"official","path":"dataset.py","file_url":"https://github.com/huage001/clear/blob/HEAD/dataset.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"667cd02575574903"}},{"code_sha256_prefix":"66a1a388e37620ac","entry":"crop_to_aspect_ratio","repo":"huage001/clear","repo_kind":"official","path":"dataset.py","file_url":"https://github.com/huage001/clear/blob/HEAD/dataset.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"66a1a388e37620ac"}},{"code_sha256_prefix":"67daa643d2f19a49","entry":"encode_prompt","repo":"huage001/clear","repo_kind":"official","path":"cache_prompt_embeds.py","file_url":"https://github.com/huage001/clear/blob/HEAD/cache_prompt_embeds.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"67daa643d2f19a49"}},{"code_sha256_prefix":"3a2a628bb69065bf","entry":"image_resize","repo":"huage001/clear","repo_kind":"official","path":"dataset.py","file_url":"https://github.com/huage001/clear/blob/HEAD/dataset.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"3a2a628bb69065bf"}},{"code_sha256_prefix":"212f670149d23d38","entry":"import_model_class_from_model_name_or_path","repo":"huage001/clear","repo_kind":"official","path":"distill.py","file_url":"https://github.com/huage001/clear/blob/HEAD/distill.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"212f670149d23d38"}},{"code_sha256_prefix":"28a7c6dea6157b37","entry":"load_text_encoders","repo":"huage001/clear","repo_kind":"official","path":"distill.py","file_url":"https://github.com/huage001/clear/blob/HEAD/distill.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"28a7c6dea6157b37"}},{"code_sha256_prefix":"8affca81b78ce6f7","entry":"log_validation","repo":"huage001/clear","repo_kind":"official","path":"distill.py","file_url":"https://github.com/huage001/clear/blob/HEAD/distill.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"8affca81b78ce6f7"}},{"code_sha256_prefix":"509fc25c579c767d","entry":"parse_args","repo":"huage001/clear","repo_kind":"official","path":"cache_latent_codes.py","file_url":"https://github.com/huage001/clear/blob/HEAD/cache_latent_codes.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"509fc25c579c767d"}},{"code_sha256_prefix":"9e59d8e6626d076b","entry":"parse_args","repo":"huage001/clear","repo_kind":"official","path":"cache_prompt_embeds.py","file_url":"https://github.com/huage001/clear/blob/HEAD/cache_prompt_embeds.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"9e59d8e6626d076b"}},{"code_sha256_prefix":"98bcd15cf273e3b4","entry":"retrieve_latents","repo":"huage001/clear","repo_kind":"official","path":"pipeline_flux_img2img.py","file_url":"https://github.com/huage001/clear/blob/HEAD/pipeline_flux_img2img.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"98bcd15cf273e3b4"}},{"code_sha256_prefix":"eb4f8a321d97a458","entry":"tokenize_prompt","repo":"huage001/clear","repo_kind":"official","path":"cache_prompt_embeds.py","file_url":"https://github.com/huage001/clear/blob/HEAD/cache_prompt_embeds.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"eb4f8a321d97a458"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}