{"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/block-diffusion-interpolating-between","title":"Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Models","arxiv_id":"2503.09573","date":"2025-03-12","proceeding":null,"authors":["Marianne Arriola","Aaron Gokaslan","Justin T. Chiu","Zhihan Yang","Zhixuan Qi","Jiaqi Han","Subham Sekhar Sahoo","Volodymyr Kuleshov"],"abstract":"Diffusion language models offer unique benefits over autoregressive models due to their potential for parallelized generation and controllability, yet they lag in likelihood modeling and are limited to fixed-length generation. In this work, we introduce a class of block diffusion language models that interpolate between discrete denoising diffusion and autoregressive models. Block diffusion overcomes key limitations of both approaches by supporting flexible-length generation and improving inference efficiency with KV caching and parallel token sampling. We propose a recipe for building effective block diffusion models that includes an efficient training algorithm, estimators of gradient variance, and data-driven noise schedules to minimize the variance. Block diffusion sets a new state-of-the-art performance among diffusion models on language modeling benchmarks and enables generation of arbitrary-length sequences. We provide the code, along with the model weights and blog post on the project page: https://m-arriola.com/bd3lms","url_abs":"https://arxiv.org/abs/2503.09573v3","url_pdf":"https://arxiv.org/pdf/2503.09573v3.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":"block-diffusion-interpolating-between","repo_url":"https://github.com/kuleshov-group/bd3lms","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"block-diffusion-interpolating-between","repo_url":"https://github.com/MindSpore-scientific/code-12/tree/main/Block_Model","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null}],"tasks":[{"task_slug":"denoising","task_name":"Denoising"},{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"}],"methods":[{"method_slug":"diffusion","method_name":"Diffusion"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/language-modelling-on-openwebtext","task":"Language Modelling","dataset":"OpenWebText","model":"BD3-LMs","rank_in_archive_order":8,"of":12,"metrics":{"eval_perplexity":"20.73","parameters":"131M"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2503.09573","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2503.09573"}},"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/MindSpore-scientific/code-12/tree/main/Block_Model","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/kuleshov-group/bd3lms","reach":{"status":"ok","spdx":"Apache-2.0"}}],"summary":{"ran_fixture":2,"ran_draft_wrong":1,"ran":4,"unverified":8},"by_repo_kind":{"official":{"samples":11,"ran":7,"repositories":1},"listed":{"samples":4,"ran":0,"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":"fbbc6722b74bda27","entry":"bias_dropout_add_scale","repo":"kuleshov-group/bd3lms","repo_kind":"official","path":"models/autoregressive.py","file_url":"https://github.com/kuleshov-group/bd3lms/blob/HEAD/models/autoregressive.py","link_basis":"harvester_set","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"fbbc6722b74bda27"}},{"code_sha256_prefix":"ee894c881614c3ec","entry":"bias_dropout_add_scale_fused_train","repo":"kuleshov-group/bd3lms","repo_kind":"official","path":"models/autoregressive.py","file_url":"https://github.com/kuleshov-group/bd3lms/blob/HEAD/models/autoregressive.py","link_basis":"harvester_set","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"ee894c881614c3ec"}},{"code_sha256_prefix":"14959ccf1e594504","entry":"block_diff_mask","repo":"kuleshov-group/bd3lms","repo_kind":"official","path":"models/dit.py","file_url":"https://github.com/kuleshov-group/bd3lms/blob/HEAD/models/dit.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"14959ccf1e594504"}},{"code_sha256_prefix":"06d2d8e50f91481f","entry":"get_bias_dropout_add_scale","repo":"kuleshov-group/bd3lms","repo_kind":"official","path":"models/autoregressive.py","file_url":"https://github.com/kuleshov-group/bd3lms/blob/HEAD/models/autoregressive.py","link_basis":"harvester_set","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":"06d2d8e50f91481f"}},{"code_sha256_prefix":"576e1b9b9bcae8ff","entry":"lm1b_detokenizer","repo":"kuleshov-group/bd3lms","repo_kind":"official","path":"dataloader.py","file_url":"https://github.com/kuleshov-group/bd3lms/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":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"576e1b9b9bcae8ff"}},{"code_sha256_prefix":"9f7fbdbe7ca280e0","entry":"ptb_detokenizer","repo":"kuleshov-group/bd3lms","repo_kind":"official","path":"dataloader.py","file_url":"https://github.com/kuleshov-group/bd3lms/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":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"9f7fbdbe7ca280e0"}},{"code_sha256_prefix":"46c10627206a7795","entry":"wt_detokenizer","repo":"kuleshov-group/bd3lms","repo_kind":"official","path":"dataloader.py","file_url":"https://github.com/kuleshov-group/bd3lms/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":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"46c10627206a7795"}},{"code_sha256_prefix":"72814d7de33e5aef","entry":"decode","repo":"MindSpore-scientific/code-12","repo_kind":"listed","path":"AECRNet/data.py","file_url":"https://github.com/MindSpore-scientific/code-12/blob/HEAD/AECRNet/data.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":"72814d7de33e5aef"}},{"code_sha256_prefix":"b4f2749f771988c2","entry":"fused_flex_attention","repo":"kuleshov-group/bd3lms","repo_kind":"official","path":"models/dit.py","file_url":"https://github.com/kuleshov-group/bd3lms/blob/HEAD/models/dit.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":"b4f2749f771988c2"}},{"code_sha256_prefix":"e1fdbc4d0bebba92","entry":"get_noise","repo":"kuleshov-group/bd3lms","repo_kind":"official","path":"noise_schedule.py","file_url":"https://github.com/kuleshov-group/bd3lms/blob/HEAD/noise_schedule.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":"e1fdbc4d0bebba92"}},{"code_sha256_prefix":"0e5797e82608fcdc","entry":"get_time_variables","repo":"kuleshov-group/bd3lms","repo_kind":"official","path":"ssd-lm/ssd_model_decode.py","file_url":"https://github.com/kuleshov-group/bd3lms/blob/HEAD/ssd-lm/ssd_model_decode.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":"0e5797e82608fcdc"}},{"code_sha256_prefix":"ea40ca66b8b3349e","entry":"load_variavle","repo":"MindSpore-scientific/code-12","repo_kind":"listed","path":"Alexnet-ABeffect/feature_predict_abm.py","file_url":"https://github.com/MindSpore-scientific/code-12/blob/HEAD/Alexnet-ABeffect/feature_predict_abm.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":"ea40ca66b8b3349e"}},{"code_sha256_prefix":"5f25d11e3f7d8a07","entry":"logits_projection","repo":"kuleshov-group/bd3lms","repo_kind":"official","path":"ssd-lm/ssd_model_decode.py","file_url":"https://github.com/kuleshov-group/bd3lms/blob/HEAD/ssd-lm/ssd_model_decode.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":"5f25d11e3f7d8a07"}},{"code_sha256_prefix":"f34618312c4fa9c9","entry":"save_variable","repo":"MindSpore-scientific/code-12","repo_kind":"listed","path":"Alexnet-ABeffect/feature_predict_abm.py","file_url":"https://github.com/MindSpore-scientific/code-12/blob/HEAD/Alexnet-ABeffect/feature_predict_abm.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":"f34618312c4fa9c9"}},{"code_sha256_prefix":"14c181fabe865cbf","entry":"set_random_seed","repo":"MindSpore-scientific/code-12","repo_kind":"listed","path":"AECRNet/data.py","file_url":"https://github.com/MindSpore-scientific/code-12/blob/HEAD/AECRNet/data.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":"14c181fabe865cbf"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}