{"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/understanding-the-performance-gap-in","title":"Understanding the Performance Gap in Preference Learning: A Dichotomy of RLHF and DPO","arxiv_id":"2505.19770","date":"2025-05-26","proceeding":null,"authors":["Ruizhe Shi","Minhak Song","Runlong Zhou","Zihan Zhang","Maryam Fazel","Simon S. Du"],"abstract":"We present a fine-grained theoretical analysis of the performance gap between reinforcement learning from human feedback (RLHF) and direct preference optimization (DPO) under a representation gap. Our study decomposes this gap into two sources: an explicit representation gap under exact optimization and an implicit representation gap under finite samples. In the exact optimization setting, we characterize how the relative capacities of the reward and policy model classes influence the final policy qualities. We show that RLHF, DPO, or online DPO can outperform one another depending on the type of model mis-specifications. Notably, online DPO can outperform both RLHF and standard DPO when the reward and policy model classes are isomorphic and both mis-specified. In the approximate optimization setting, we provide a concrete construction where the ground-truth reward is implicitly sparse and show that RLHF requires significantly fewer samples than DPO to recover an effective reward model -- highlighting a statistical advantage of two-stage learning. Together, these results provide a comprehensive understanding of the performance gap between RLHF and DPO under various settings, and offer practical insights into when each method is preferred.","url_abs":"https://arxiv.org/abs/2505.19770v1","url_pdf":"https://arxiv.org/pdf/2505.19770v1.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":"understanding-the-performance-gap-in","repo_url":"https://github.com/srzer/Gap-in-Preference-Learning","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[],"methods":[{"method_slug":"dpo","method_name":"DPO"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2505.19770","atlas_url":"https://app.syntology.ai/?focus=2505.19770","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2505.19770"}},"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/srzer/Gap-in-Preference-Learning","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran":1,"unverified":3},"by_repo_kind":{"official":{"samples":4,"ran":1,"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":"5100040c5aa125b4","entry":"stabilized_log1pexp","repo":"srzer/Gap-in-Preference-Learning","repo_kind":"official","path":"Exp-1/src/trainer/dpo_trainer.py","file_url":"https://github.com/srzer/Gap-in-Preference-Learning/blob/HEAD/Exp-1/src/trainer/dpo_trainer.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"5100040c5aa125b4"}},{"code_sha256_prefix":"0d39de6cae03583a","entry":"get_reward","repo":"srzer/Gap-in-Preference-Learning","repo_kind":"official","path":"Exp-0/annotate_data/get_rewards.py","file_url":"https://github.com/srzer/Gap-in-Preference-Learning/blob/HEAD/Exp-0/annotate_data/get_rewards.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"0d39de6cae03583a"}},{"code_sha256_prefix":"f4bd8b36e532b748","entry":"prepare_data","repo":"srzer/Gap-in-Preference-Learning","repo_kind":"official","path":"Exp-0/dpo_iteration/run_dpo.py","file_url":"https://github.com/srzer/Gap-in-Preference-Learning/blob/HEAD/Exp-0/dpo_iteration/run_dpo.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"f4bd8b36e532b748"}},{"code_sha256_prefix":"4981195135c64c27","entry":"relabel_dataset","repo":"srzer/Gap-in-Preference-Learning","repo_kind":"official","path":"Exp-1/src/utils/util_dataset.py","file_url":"https://github.com/srzer/Gap-in-Preference-Learning/blob/HEAD/Exp-1/src/utils/util_dataset.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"4981195135c64c27"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}