{"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-2605-29398","title":"GDSD: Reinforcement Learning as Guided Denoiser Self-Distillation for Diffusion Language Models","arxiv_id":"2605.29398","date":"2026-05-28","proceeding":null,"authors":["Xiaohang Tang","Keyue Jiang","Che Liu","Qifang Zhao","Xiaoxiao Xu","Sangwoong Yoon","Ilija Bogunovic"],"abstract":"Reinforcement learning (RL) can be used to improve the policy (denoiser) of diffusion large language models (dLLMs), while being hindered by the intractability of the policy likelihood. A dominant and efficient family of methods replaces the likelihood in standard RL with its evidence lower bound (ELBO), estimated from randomly masked sequences. Despite being well aligned with pre-training, these approaches introduce bias through training--inference mismatch by using the ELBO as a likelihood surrogate, which can degrade performance. In this work, we propose Guided Denoiser Self-Distillation (GDSD) to directly distill the denoiser of dLLMs from an advantage-guided self-teacher, derived from the closed-form optimum of reverse-KL regularized RL. GDSD matches the dLLM's denoiser logits to the teacher's via a normalization-free objective, which reduces RL to likelihood-free self-distillation and thus bypasses the TIM biases. Recent ELBO-based methods emerge as instances of applying different distillation divergences, but with diagnosable pathologies that GDSD avoids. On planning, math, and coding benchmarks with LLaDA-8B and Dream-7B, GDSD consistently outperforms prior state-of-the-art ELBO-based methods with a more stable training reward dynamics, achieving test-accuracy improvements of up to $+19.6\\%$. These results suggest that direct denoiser self-distillation, without relying on an ELBO likelihood surrogate, can provide a more stable and effective RL procedure for dLLMs. Code is available at https://github.com/GaryBall/GDSD.","url_abs":"https://arxiv.org/abs/2605.29398","url_pdf":"https://arxiv.org/pdf/2605.29398","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=2605.29398","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2605.29398"}},"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/GaryBall/GDSD","reach":null}],"summary":{"ran":11,"unverified":2},"by_repo_kind":{"found_in_text":{"samples":13,"ran":11,"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":"5c23d8a8de0a68a7","entry":"boxed_in_answer","repo":"GaryBall/GDSD","repo_kind":"found_in_text","path":"gdsd/utils/math500_utils.py","file_url":"https://github.com/GaryBall/GDSD/blob/HEAD/gdsd/utils/math500_utils.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":"5c23d8a8de0a68a7"}},{"code_sha256_prefix":"773d9371a4bdce62","entry":"compute_approx_kl","repo":"GaryBall/GDSD","repo_kind":"found_in_text","path":"gdsd/trainers/espo_trainer.py","file_url":"https://github.com/GaryBall/GDSD/blob/HEAD/gdsd/trainers/espo_trainer.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":"773d9371a4bdce62"}},{"code_sha256_prefix":"5ac161f7ca1e4c6e","entry":"dummy_prepare_inputs_for_generation","repo":"GaryBall/GDSD","repo_kind":"found_in_text","path":"gdsd/utils/model_utils.py","file_url":"https://github.com/GaryBall/GDSD/blob/HEAD/gdsd/utils/model_utils.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":"5ac161f7ca1e4c6e"}},{"code_sha256_prefix":"0f07a2dc4671890b","entry":"extract_code","repo":"GaryBall/GDSD","repo_kind":"found_in_text","path":"gdsd/rewards.py","file_url":"https://github.com/GaryBall/GDSD/blob/HEAD/gdsd/rewards.py","link_basis":"first_harvest_node","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":"0f07a2dc4671890b"}},{"code_sha256_prefix":"96a43402b05ef0d4","entry":"get_code_format_reward","repo":"GaryBall/GDSD","repo_kind":"found_in_text","path":"gdsd/rewards.py","file_url":"https://github.com/GaryBall/GDSD/blob/HEAD/gdsd/rewards.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":"96a43402b05ef0d4"}},{"code_sha256_prefix":"b2845803d938cb26","entry":"is_equiv","repo":"GaryBall/GDSD","repo_kind":"found_in_text","path":"gdsd/utils/math500_utils.py","file_url":"https://github.com/GaryBall/GDSD/blob/HEAD/gdsd/utils/math500_utils.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":"b2845803d938cb26"}},{"code_sha256_prefix":"8268ac93c7808b81","entry":"make_code_conversation","repo":"GaryBall/GDSD","repo_kind":"found_in_text","path":"gdsd/data_utils.py","file_url":"https://github.com/GaryBall/GDSD/blob/HEAD/gdsd/data_utils.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":"8268ac93c7808b81"}},{"code_sha256_prefix":"e4d5a8bc57ae4499","entry":"remove_boxed","repo":"GaryBall/GDSD","repo_kind":"found_in_text","path":"gdsd/utils/math500_utils.py","file_url":"https://github.com/GaryBall/GDSD/blob/HEAD/gdsd/utils/math500_utils.py","link_basis":"first_harvest_node","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":"e4d5a8bc57ae4499"}},{"code_sha256_prefix":"2b84adae31d26dc5","entry":"run_single_test","repo":"GaryBall/GDSD","repo_kind":"found_in_text","path":"gdsd/utils/local_sandbox.py","file_url":"https://github.com/GaryBall/GDSD/blob/HEAD/gdsd/utils/local_sandbox.py","link_basis":"first_harvest_node","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":"2b84adae31d26dc5"}},{"code_sha256_prefix":"86bd416f8d3d9e07","entry":"run_single_test_against_program_helper","repo":"GaryBall/GDSD","repo_kind":"found_in_text","path":"gdsd/utils/local_sandbox.py","file_url":"https://github.com/GaryBall/GDSD/blob/HEAD/gdsd/utils/local_sandbox.py","link_basis":"first_harvest_node","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":"86bd416f8d3d9e07"}},{"code_sha256_prefix":"23e33cf487ecdf91","entry":"split_tensor_dict","repo":"GaryBall/GDSD","repo_kind":"found_in_text","path":"gdsd/trainers/espo_trainer.py","file_url":"https://github.com/GaryBall/GDSD/blob/HEAD/gdsd/trainers/espo_trainer.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":"23e33cf487ecdf91"}},{"code_sha256_prefix":"57c8964d40fed7a7","entry":"code_reward","repo":"GaryBall/GDSD","repo_kind":"found_in_text","path":"gdsd/rewards.py","file_url":"https://github.com/GaryBall/GDSD/blob/HEAD/gdsd/rewards.py","link_basis":"first_harvest_node","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":"57c8964d40fed7a7"}},{"code_sha256_prefix":"2dc0cf9cedea3958","entry":"local_execute","repo":"GaryBall/GDSD","repo_kind":"found_in_text","path":"gdsd/utils/local_sandbox.py","file_url":"https://github.com/GaryBall/GDSD/blob/HEAD/gdsd/utils/local_sandbox.py","link_basis":"first_harvest_node","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":"2dc0cf9cedea3958"}}]},"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}