{"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/improving-data-efficiency-for-llm","title":"Improving Data Efficiency for LLM Reinforcement Fine-tuning Through Difficulty-targeted Online Data Selection and Rollout Replay","arxiv_id":"2506.05316","date":"2025-06-05","proceeding":null,"authors":["Yifan Sun","Jingyan Shen","Yibin Wang","Tianyu Chen","Zhendong Wang","Mingyuan Zhou","huan zhang"],"abstract":"Reinforcement learning (RL) has become an effective approach for fine-tuning large language models (LLMs), particularly to enhance their reasoning capabilities. However, RL fine-tuning remains highly resource-intensive, and existing work has largely overlooked the problem of data efficiency. In this paper, we propose two techniques to improve data efficiency in LLM RL fine-tuning: difficulty-targeted online data selection and rollout replay. We introduce the notion of adaptive difficulty to guide online data selection, prioritizing questions of moderate difficulty that are more likely to yield informative learning signals. To estimate adaptive difficulty efficiently, we develop an attention-based framework that requires rollouts for only a small reference set of questions. The adaptive difficulty of the remaining questions is then estimated based on their similarity to this set. To further reduce rollout cost, we introduce a rollout replay mechanism that reuses recent rollouts, lowering per-step computation while maintaining stable updates. Extensive experiments across 6 LLM-dataset combinations show that our method reduces RL fine-tuning time by 25% to 65% to reach the same level of performance as the original GRPO algorithm.","url_abs":"https://arxiv.org/abs/2506.05316v1","url_pdf":"https://arxiv.org/pdf/2506.05316v1.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":"improving-data-efficiency-for-llm","repo_url":"https://github.com/astral-group/data-efficient-llm-rl","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"}],"methods":[{"method_slug":"set","method_name":"SET"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2506.05316","atlas_url":"https://app.syntology.ai/?focus=2506.05316","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2506.05316"}},"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":"deterministic:regex_extraction","url":"https://github.com/ASTRAL-Group/data-efficient-llm-rl","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/astral-group/data-efficient-llm-rl","reach":{"status":"ok"}}],"summary":{"ran":1,"ran_draft_wrong":1,"unverified":4},"by_repo_kind":{"official":{"samples":6,"ran":2,"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":6,"samples":[{"code_sha256_prefix":"20ba9212ecfbbbbc","entry":"DataProto","repo":"astral-group/data-efficient-llm-rl","repo_kind":"official","path":"rl_training/verl/verl/trainer/ppo/ray_trainer_teacher_replay.py","file_url":"https://github.com/astral-group/data-efficient-llm-rl/blob/HEAD/rl_training/verl/verl/trainer/ppo/ray_trainer_teacher_replay.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"20ba9212ecfbbbbc"}},{"code_sha256_prefix":"182d4d22f782691b","entry":"union_two_dict","repo":"astral-group/data-efficient-llm-rl","repo_kind":"official","path":"rl_training/verl/verl/trainer/ppo/ray_trainer_teacher_replay.py","file_url":"https://github.com/astral-group/data-efficient-llm-rl/blob/HEAD/rl_training/verl/verl/trainer/ppo/ray_trainer_teacher_replay.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"182d4d22f782691b"}},{"code_sha256_prefix":"3673e3d46aca5c5e","entry":"DataProtoItem","repo":"astral-group/data-efficient-llm-rl","repo_kind":"official","path":"rl_training/verl/verl/trainer/ppo/ray_trainer_teacher_replay.py","file_url":"https://github.com/astral-group/data-efficient-llm-rl/blob/HEAD/rl_training/verl/verl/trainer/ppo/ray_trainer_teacher_replay.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"3673e3d46aca5c5e"}},{"code_sha256_prefix":"d11b448b5d76e167","entry":"collate_fn","repo":"astral-group/data-efficient-llm-rl","repo_kind":"official","path":"rl_training/verl/verl/trainer/ppo/ray_trainer_teacher_replay.py","file_url":"https://github.com/astral-group/data-efficient-llm-rl/blob/HEAD/rl_training/verl/verl/trainer/ppo/ray_trainer_teacher_replay.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"d11b448b5d76e167"}},{"code_sha256_prefix":"483d81ba8924261d","entry":"dataprotoitem_to_dataproto","repo":"astral-group/data-efficient-llm-rl","repo_kind":"official","path":"rl_training/verl/verl/trainer/ppo/ray_trainer_teacher_replay.py","file_url":"https://github.com/astral-group/data-efficient-llm-rl/blob/HEAD/rl_training/verl/verl/trainer/ppo/ray_trainer_teacher_replay.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"483d81ba8924261d"}},{"code_sha256_prefix":"8dff9b7e4a9fed7e","entry":"select_buffer","repo":"astral-group/data-efficient-llm-rl","repo_kind":"official","path":"rl_training/verl/verl/trainer/ppo/ray_trainer_teacher_replay.py","file_url":"https://github.com/astral-group/data-efficient-llm-rl/blob/HEAD/rl_training/verl/verl/trainer/ppo/ray_trainer_teacher_replay.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"8dff9b7e4a9fed7e"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}