{"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/advantage-guided-distillation-for-preference","title":"Advantage-Guided Distillation for Preference Alignment in Small Language Models","arxiv_id":"2502.17927","date":"2025-02-25","proceeding":null,"authors":["Shiping Gao","Fanqi Wan","Jiajian Guo","Xiaojun Quan","Qifan Wang"],"abstract":"Alignment techniques enable Large Language Models (LLMs) to generate outputs that align with human preferences and play a crucial role in their effectiveness. However, their impact often diminishes when applied to Small Language Models (SLMs), likely due to the limited capacity of these models. Instead of directly applying existing alignment techniques to SLMs, we propose to utilize a well-aligned teacher LLM to guide the alignment process for these models, thereby facilitating the transfer of the teacher's knowledge of human preferences to the student model. To achieve this, we first explore a straightforward approach, Dual-Constrained Knowledge Distillation (DCKD), that employs knowledge distillation with two KL-divergence constraints from the aligned teacher to the unaligned student. To further enhance the student's ability to distinguish between preferred and dispreferred responses, we then propose Advantage-Guided Distillation for Preference Alignment (ADPA), which leverages an advantage function from the aligned teacher to deliver more nuanced, distribution-level reward signals for the student's alignment. Our experimental results show that these two approaches appreciably improve the alignment of SLMs and narrow the performance gap with larger counterparts. Among them, ADPA demonstrates superior performance and achieves even greater effectiveness when integrated with DCKD. Our code is available at https://github.com/SLIT-AI/ADPA.","url_abs":"https://arxiv.org/abs/2502.17927v1","url_pdf":"https://arxiv.org/pdf/2502.17927v1.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":"advantage-guided-distillation-for-preference","repo_url":"https://github.com/slit-ai/adpa","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"knowledge-distillation","task_name":"Knowledge Distillation"}],"methods":[{"method_slug":"align","method_name":"ALIGN"},{"method_slug":"knowledge-distillation","method_name":"Knowledge Distillation"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2502.17927","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2502.17927"}},"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":"deterministic:regex_extraction","url":"https://github.com/SLIT-AI/ADPA","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/slit-ai/adpa","reach":null}],"summary":{"ran_draft_wrong":3},"by_repo_kind":{"official":{"samples":3,"ran":3,"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":3,"samples":[{"code_sha256_prefix":"907eece5744b1b04","entry":"compress_probs","repo":"SLIT-AI/ADPA","repo_kind":"official","path":"utils/precompute_logits.py","file_url":"https://github.com/SLIT-AI/ADPA/blob/HEAD/utils/precompute_logits.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"907eece5744b1b04"}},{"code_sha256_prefix":"2275f0a56a14dbe8","entry":"custom_collate_fn","repo":"SLIT-AI/ADPA","repo_kind":"official","path":"utils/precompute_logits.py","file_url":"https://github.com/SLIT-AI/ADPA/blob/HEAD/utils/precompute_logits.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"2275f0a56a14dbe8"}},{"code_sha256_prefix":"10598df62dee077f","entry":"get_margin_logp_token","repo":"slit-ai/adpa","repo_kind":"official","path":"utils/merge_logits_adpa_dataset.py","file_url":"https://github.com/slit-ai/adpa/blob/HEAD/utils/merge_logits_adpa_dataset.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"10598df62dee077f"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}