{"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/harmonica-harmonizing-training-and-inference","title":"HarmoniCa: Harmonizing Training and Inference for Better Feature Caching in Diffusion Transformer Acceleration","arxiv_id":"2410.01723","date":"2024-10-02","proceeding":null,"authors":["Yushi Huang","Zining Wang","Ruihao Gong","Jing Liu","Xinjie Zhang","Jinyang Guo","Xianglong Liu","Jun Zhang"],"abstract":"Diffusion Transformers (DiTs) excel in generative tasks but face practical deployment challenges due to high inference costs. Feature caching, which stores and retrieves redundant computations, offers the potential for acceleration. Existing learning-based caching, though adaptive, overlooks the impact of the prior timestep. It also suffers from misaligned objectives--aligned predicted noise vs. high-quality images--between training and inference. These two discrepancies compromise both performance and efficiency. To this end, we harmonize training and inference with a novel learning-based caching framework dubbed HarmoniCa. It first incorporates Step-Wise Denoising Training (SDT) to ensure the continuity of the denoising process, where prior steps can be leveraged. In addition, an Image Error Proxy-Guided Objective (IEPO) is applied to balance image quality against cache utilization through an efficient proxy to approximate the image error. Extensive experiments across $8$ models, $4$ samplers, and resolutions from $256\\times256$ to $2K$ demonstrate superior performance and speedup of our framework. For instance, it achieves over $40\\%$ latency reduction (i.e., $2.07\\times$ theoretical speedup) and improved performance on PixArt-$\\alpha$. Remarkably, our image-free approach reduces training time by $25\\%$ compared with the previous method. Our code is available at https://github.com/ModelTC/HarmoniCa.","url_abs":"https://arxiv.org/abs/2410.01723v5","url_pdf":"https://arxiv.org/pdf/2410.01723v5.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":"harmonica-harmonizing-training-and-inference","repo_url":"https://github.com/modeltc/harmonica","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"2k","task_name":"2k"},{"task_slug":"denoising","task_name":"Denoising"}],"methods":[{"method_slug":"diffusion","method_name":"Diffusion"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2410.01723","atlas_url":"https://app.syntology.ai/?focus=2410.01723","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2410.01723"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-25T09:33:49+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/modeltc/harmonica","reach":{"status":"ok","spdx":"Apache-2.0"}}],"summary":{"ran_draft_wrong":4,"ran_honours":4,"ran":1,"ran_fixture":1,"unverified":5},"by_repo_kind":{"official":{"samples":15,"ran":10,"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":"d6a68e210556f857","entry":"approx_standard_normal_cdf","repo":"modeltc/harmonica","repo_kind":"official","path":"diffusion/diffusion_utils.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/diffusion/diffusion_utils.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"d6a68e210556f857"}},{"code_sha256_prefix":"ab1c9568b4e13899","entry":"continuous_gaussian_log_likelihood","repo":"modeltc/harmonica","repo_kind":"official","path":"diffusion/diffusion_utils.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/diffusion/diffusion_utils.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"ab1c9568b4e13899"}},{"code_sha256_prefix":"c92c27c924b517e8","entry":"get_2d_sincos_pos_embed","repo":"modeltc/harmonica","repo_kind":"official","path":"models/dynamic_models.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/models/dynamic_models.py","link_basis":"harvester_set","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"c92c27c924b517e8"}},{"code_sha256_prefix":"3e0fa4efc22272d4","entry":"get_beta_schedule","repo":"modeltc/harmonica","repo_kind":"official","path":"diffusion/gaussian_diffusion.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/diffusion/gaussian_diffusion.py","link_basis":"harvester_set","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"3e0fa4efc22272d4"}},{"code_sha256_prefix":"495f938cb55dd81f","entry":"get_lambda","repo":"modeltc/harmonica","repo_kind":"official","path":"train_router.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/train_router.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"495f938cb55dd81f"}},{"code_sha256_prefix":"36e30c7fb679ec78","entry":"get_named_beta_schedule","repo":"modeltc/harmonica","repo_kind":"official","path":"diffusion/gaussian_diffusion.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/diffusion/gaussian_diffusion.py","link_basis":"harvester_set","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"36e30c7fb679ec78"}},{"code_sha256_prefix":"f6d7c009a8efb8b7","entry":"mean_flat","repo":"modeltc/harmonica","repo_kind":"official","path":"train_router.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/train_router.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"f6d7c009a8efb8b7"}},{"code_sha256_prefix":"03310bba324ae4fb","entry":"modulate","repo":"modeltc/harmonica","repo_kind":"official","path":"models/dynamic_models.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/models/dynamic_models.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"03310bba324ae4fb"}},{"code_sha256_prefix":"8afbfc42c6ea0448","entry":"normal_kl","repo":"modeltc/harmonica","repo_kind":"official","path":"diffusion/diffusion_utils.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/diffusion/diffusion_utils.py","link_basis":"harvester_set","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"8afbfc42c6ea0448"}},{"code_sha256_prefix":"ea9dbc131adf582e","entry":"space_timesteps","repo":"modeltc/harmonica","repo_kind":"official","path":"diffusion/respace.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/diffusion/respace.py","link_basis":"harvester_set","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"ea9dbc131adf582e"}},{"code_sha256_prefix":"8382f7e48cc8a309","entry":"create_logger","repo":"modeltc/harmonica","repo_kind":"official","path":"train_router.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/train_router.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":"8382f7e48cc8a309"}},{"code_sha256_prefix":"e48218d7d73db0b3","entry":"create_named_schedule_sampler","repo":"modeltc/harmonica","repo_kind":"official","path":"diffusion/timestep_sampler.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/diffusion/timestep_sampler.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":"e48218d7d73db0b3"}},{"code_sha256_prefix":"6a0d5ecd441905cc","entry":"download_model","repo":"modeltc/harmonica","repo_kind":"official","path":"download.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/download.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":"6a0d5ecd441905cc"}},{"code_sha256_prefix":"29947a0a94157558","entry":"find_model","repo":"modeltc/harmonica","repo_kind":"official","path":"download.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/download.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":"29947a0a94157558"}},{"code_sha256_prefix":"665d8a4e8f673a4c","entry":"get_2d_sincos_pos_embed_from_grid","repo":"modeltc/harmonica","repo_kind":"official","path":"models/dynamic_models.py","file_url":"https://github.com/modeltc/harmonica/blob/HEAD/models/dynamic_models.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":"665d8a4e8f673a4c"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}