{"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-2509-15188","title":"Fast and Fluent Diffusion Language Models via Convolutional Decoding and Rejective Fine-tuning","arxiv_id":"2509.15188","date":"2025-09-18","proceeding":"NeurIPS","authors":["Yeongbin Seo","Dongha Lee","Jaehyung Kim","Jinyoung Yeo"],"abstract":"Autoregressive (AR) language models generate text one token at a time, which limits their inference speed. Diffusion-based language models offer a promising alternative, as they can decode multiple tokens in parallel. However, we identify a key bottleneck in current diffusion LMs: the long decoding-window problem, where tokens generated far from the input context often become irrelevant or repetitive. Previous solutions like semi-autoregressive address this issue by splitting windows into blocks (sacrificing bidirectionality), but we find that this also leads to time-interval expansion problem, sacrificing the speed. Therefore, semi-AR eliminates the main advantages of diffusion models. To overcome this, we propose Convolutional decoding (Conv), a normalization-based method that narrows the decoding window without hard segmentation, leading to better fluency and flexibility. Additionally, we introduce Rejecting Rule-based Fine-Tuning (R2FT), a post-hoc training scheme that better aligns tokens at positions far from context. Our methods achieve state-of-the-art results on open-ended generation benchmarks (e.g., AlpacaEval) among diffusion LM baselines, with significantly lower step size than previous works, demonstrating both speed and quality improvements.","url_abs":"https://arxiv.org/abs/2509.15188","url_pdf":"https://arxiv.org/pdf/2509.15188","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=2509.15188","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2509.15188"}},"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/ybseo-ac/Conv","reach":{"status":"ok"}}],"summary":{"ran_fixture":3,"ran":5,"ran_draft_wrong":1,"unverified":8},"by_repo_kind":{"found_in_text":{"samples":17,"ran":9,"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":17,"samples":[{"code_sha256_prefix":"a5334721ab55f48f","entry":"add_gumbel_noise","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"generate_yb.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/generate_yb.py","link_basis":"plan_row","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"a5334721ab55f48f"}},{"code_sha256_prefix":"fbbc6722b74bda27","entry":"bias_dropout_add_scale","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"models/autoregressive.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/models/autoregressive.py","link_basis":"plan_row","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"fbbc6722b74bda27"}},{"code_sha256_prefix":"ee894c881614c3ec","entry":"bias_dropout_add_scale_fused_train","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"models/autoregressive.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/models/autoregressive.py","link_basis":"plan_row","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"ee894c881614c3ec"}},{"code_sha256_prefix":"06d2d8e50f91481f","entry":"get_bias_dropout_add_scale","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"models/autoregressive.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/models/autoregressive.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"06d2d8e50f91481f"}},{"code_sha256_prefix":"6f22da9cc766ffbe","entry":"get_num_transfer_tokens","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"generate_yb.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/generate_yb.py","link_basis":"plan_row","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":"6f22da9cc766ffbe"}},{"code_sha256_prefix":"befea8ff69c3c190","entry":"get_num_transfer_tokens","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"gen1_2_answer_generation_llada.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/gen1_2_answer_generation_llada.py","link_basis":"plan_row","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"befea8ff69c3c190"}},{"code_sha256_prefix":"576e1b9b9bcae8ff","entry":"lm1b_detokenizer","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"dataloader.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/dataloader.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"576e1b9b9bcae8ff"}},{"code_sha256_prefix":"9f7fbdbe7ca280e0","entry":"ptb_detokenizer","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"dataloader.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/dataloader.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"9f7fbdbe7ca280e0"}},{"code_sha256_prefix":"46c10627206a7795","entry":"wt_detokenizer","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"dataloader.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/dataloader.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"46c10627206a7795"}},{"code_sha256_prefix":"529aefed4b15eb2b","entry":"add_gumbel_noise","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"gen1_2_answer_generation_llada.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/gen1_2_answer_generation_llada.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":"529aefed4b15eb2b"}},{"code_sha256_prefix":"38b9682eb110eda2","entry":"cross_entropy","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"models/dimamba.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/models/dimamba.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":"38b9682eb110eda2"}},{"code_sha256_prefix":"5e8a159c6df1dcb1","entry":"generate","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"gen1_2_answer_generation_llada.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/gen1_2_answer_generation_llada.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":"5e8a159c6df1dcb1"}},{"code_sha256_prefix":"6e4c11b91e906864","entry":"generate","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"generate_yb.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/generate_yb.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":"6e4c11b91e906864"}},{"code_sha256_prefix":"371859ac56b363f1","entry":"get_dataloaders_finetune","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"finetune_tools.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/finetune_tools.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":"371859ac56b363f1"}},{"code_sha256_prefix":"935b6e84a478b512","entry":"prior_std_mean_calcul_conditional","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"finetune_tools.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/finetune_tools.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":"935b6e84a478b512"}},{"code_sha256_prefix":"4e433264265cd4c7","entry":"q_xt_finetune","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"finetune_tools.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/finetune_tools.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":"4e433264265cd4c7"}},{"code_sha256_prefix":"fdbb5a9a25cff18f","entry":"weighted_cross_entropy","repo":"ybseo-ac/Conv","repo_kind":"found_in_text","path":"models/dimamba.py","file_url":"https://github.com/ybseo-ac/Conv/blob/HEAD/models/dimamba.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":"fdbb5a9a25cff18f"}}]},"arxiv_metadata":{"licence":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)","fields":["title","abstract","authors","date"],"primary_category":"cs.CL","source":"arxiv_api"},"syntology_extracted_results":null}