{"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/lead-iterative-data-selection-for-efficient","title":"LEAD: Iterative Data Selection for Efficient LLM Instruction Tuning","arxiv_id":"2505.07437","date":"2025-05-12","proceeding":null,"authors":["Xiaotian Lin","Yanlin Qi","Yizhang Zhu","Themis Palpanas","Chengliang Chai","Nan Tang","Yuyu Luo"],"abstract":"Instruction tuning has emerged as a critical paradigm for improving the capabilities and alignment of large language models (LLMs). However, existing iterative model-aware data selection methods incur significant computational overhead, as they rely on repeatedly performing full-dataset model inference to estimate sample utility for subsequent training iterations, creating a fundamental efficiency bottleneck. In this paper, we propose LEAD, an efficient iterative data selection framework that accurately estimates sample utility entirely within the standard training loop, eliminating the need for costly additional model inference. At its core, LEAD introduces Instance-Level Dynamic Uncertainty (IDU), a theoretically grounded utility function combining instantaneous training loss, gradient-based approximation of loss changes, and exponential smoothing of historical loss signals. To further scale efficiently to large datasets, LEAD employs a two-stage, coarse-to-fine selection strategy, adaptively prioritizing informative clusters through a multi-armed bandit mechanism, followed by precise fine-grained selection of high-utility samples using IDU. Extensive experiments across four diverse benchmarks show that LEAD significantly outperforms state-of-the-art methods, improving average model performance by 6.1%-10.8% while using only 2.5% of the training data and reducing overall training time by 5-10x.","url_abs":"https://arxiv.org/abs/2505.07437v1","url_pdf":"https://arxiv.org/pdf/2505.07437v1.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":"lead-iterative-data-selection-for-efficient","repo_url":"https://github.com/HKUSTDial/LEAD","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2505.07437","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2505.07437"}},"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/HKUSTDial/LEAD","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran":3,"unverified":18},"by_repo_kind":{"official":{"samples":21,"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":0,"samples":[{"code_sha256_prefix":"5637030b6f66a7c5","entry":"create_prompt_with_llama2_chat_format","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/eval/templates.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/eval/templates.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"5637030b6f66a7c5"}},{"code_sha256_prefix":"45e2cc1022c95249","entry":"create_prompt_with_olmo_chat_format","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/eval/templates.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/eval/templates.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"45e2cc1022c95249"}},{"code_sha256_prefix":"661a1d8a5c30c318","entry":"create_prompt_with_tulu_chat_format","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/eval/templates.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/eval/templates.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"661a1d8a5c30c318"}},{"code_sha256_prefix":"45c349e9550f0ef9","entry":"assign_dataset_task_ids","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/offline/clustering.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/offline/clustering.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":"45c349e9550f0ef9"}},{"code_sha256_prefix":"0228778589332787","entry":"calculate_cosine_similarity","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/online/utilityChange.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/online/utilityChange.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":"0228778589332787"}},{"code_sha256_prefix":"2f2560c89f620a6b","entry":"calculate_param_changes","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/online/utilityChange.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/online/utilityChange.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":"2f2560c89f620a6b"}},{"code_sha256_prefix":"6cf648b17d975e07","entry":"categorical_draw","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/online/exp3.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/online/exp3.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":"6cf648b17d975e07"}},{"code_sha256_prefix":"64d6138fb03f75c7","entry":"dispatch_openai_chat_requesets","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/eval/dispatch_openai_requests.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/eval/dispatch_openai_requests.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":"64d6138fb03f75c7"}},{"code_sha256_prefix":"5bbc896a72379511","entry":"dispatch_openai_prompt_requesets","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/eval/dispatch_openai_requests.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/eval/dispatch_openai_requests.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":"5bbc896a72379511"}},{"code_sha256_prefix":"1d1694af124350b8","entry":"encode_with_messages_format","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/online/template.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/online/template.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":"1d1694af124350b8"}},{"code_sha256_prefix":"7289ececedcd890f","entry":"encode_with_messages_format_wo_conv","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/online/template.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/online/template.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":"7289ececedcd890f"}},{"code_sha256_prefix":"57148174b9cdbbea","entry":"encode_with_prompt_completion_format","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/online/template.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/online/template.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":"57148174b9cdbbea"}},{"code_sha256_prefix":"cdbcc01805fac313","entry":"generate_completions","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/eval/utils.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/eval/utils.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":"cdbcc01805fac313"}},{"code_sha256_prefix":"298ba0aff249df55","entry":"get_model_params_vector","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/online/utilityChange.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/online/utilityChange.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":"298ba0aff249df55"}},{"code_sha256_prefix":"d134c88af0236263","entry":"get_next_word_predictions","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/eval/utils.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/eval/utils.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":"d134c88af0236263"}},{"code_sha256_prefix":"652eac296628dd5e","entry":"get_perplexity_and_embedding_part_text","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/offline/difficulty_score.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/offline/difficulty_score.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":"652eac296628dd5e"}},{"code_sha256_prefix":"e6e19cb6234cc995","entry":"get_perplexity_and_embedding_whole_text","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/offline/difficulty_score.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/offline/difficulty_score.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":"e6e19cb6234cc995"}},{"code_sha256_prefix":"30348ef4ac1d9f07","entry":"load_data","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/offline/clustering.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/offline/clustering.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":"30348ef4ac1d9f07"}},{"code_sha256_prefix":"157e4633d9e5ccce","entry":"load_data","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/offline/difficulty_score.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/offline/difficulty_score.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":"157e4633d9e5ccce"}},{"code_sha256_prefix":"ead339ee380ea819","entry":"perform_clustering","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/offline/clustering.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/offline/clustering.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":"ead339ee380ea819"}},{"code_sha256_prefix":"83f3d617be77562f","entry":"retry_on_exception","repo":"HKUSTDial/LEAD","repo_kind":"official","path":"src/eval/utils.py","file_url":"https://github.com/HKUSTDial/LEAD/blob/HEAD/src/eval/utils.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":"83f3d617be77562f"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}