{"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-2605-11666","title":"Evolutionary Task Discovery: Advancing Reasoning Frontiers via Skill Composition and Complexity Scaling","arxiv_id":"2605.11666","date":"2026-05-12","proceeding":null,"authors":["Liqin Ye","Yanbin Yin","Michael Galarnyk","Yuzhao Heng","Sudheer Chava","Chao Zhang"],"abstract":"The reasoning frontier of Large Language Models (LLMs) has advanced significantly through modern post-training paradigms (e.g., Reinforcement Learning from Verifiable Rewards (RLVR)). However, the efficacy of these methods remains fundamentally constrained by the diversity and complexity of the training data. One practical solution is data synthesis; yet, prevalent methods relying on unstructured mutation or exploration suffer from homogeneity collapse, failing to systematically expand the reasoning frontier. To overcome this, we propose Evoutionary Task Discovery (EvoTD), a framework that treats data synthesis as a directed search over a dual-axis manifold of Algorithmic Skills and Complexity Attributes. We introduce structured evolutionary operators to navigate this space: a Crossover operator that synthesizes novel skill compositions to enhance diversity, and a Parametric Mutation operator that scales structural constraints (e.g., input size, tree depth) to drive robust generalization. Crucially, we integrate a dynamic Zone of Proximal Development filter, ensuring tasks lie within the learnable region of the model. Empirically, EvoTD delivers substantial reasoning gains that generalize consistently across model architectures, pretraining regimes, and scales, demonstrating that structured evolutionary curricula can effectively support reasoning improvement. We release our code on https://github.com/liqinye/EvoTD.","url_abs":"https://arxiv.org/abs/2605.11666","url_pdf":"https://arxiv.org/pdf/2605.11666","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":null,"mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2605.11666"}},"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/liqinye/EvoTD","reach":null}],"summary":{"ran":9,"unverified":3},"by_repo_kind":{"found_in_text":{"samples":12,"ran":9,"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":"c891f12deed31373","entry":"check_determinism","repo":"liqinye/EvoTD","repo_kind":"found_in_text","path":"src/utils/checks.py","file_url":"https://github.com/liqinye/EvoTD/blob/HEAD/src/utils/checks.py","link_basis":"first_harvest_node","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":"c891f12deed31373"}},{"code_sha256_prefix":"7c5bfa5e64c2cf12","entry":"check_no_definitions","repo":"liqinye/EvoTD","repo_kind":"found_in_text","path":"src/utils/checks.py","file_url":"https://github.com/liqinye/EvoTD/blob/HEAD/src/utils/checks.py","link_basis":"first_harvest_node","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":"7c5bfa5e64c2cf12"}},{"code_sha256_prefix":"14ce7d41383636f4","entry":"contains_banned_imports","repo":"liqinye/EvoTD","repo_kind":"found_in_text","path":"src/utils/checks.py","file_url":"https://github.com/liqinye/EvoTD/blob/HEAD/src/utils/checks.py","link_basis":"first_harvest_node","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":"14ce7d41383636f4"}},{"code_sha256_prefix":"4a47f642a1a60100","entry":"generate_reflection_prompt","repo":"liqinye/EvoTD","repo_kind":"found_in_text","path":"src/prompts.py","file_url":"https://github.com/liqinye/EvoTD/blob/HEAD/src/prompts.py","link_basis":"first_harvest_node","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":"4a47f642a1a60100"}},{"code_sha256_prefix":"0359d2b5bb763b57","entry":"get_code_problem_predictor_prompt","repo":"liqinye/EvoTD","repo_kind":"found_in_text","path":"src/prompts.py","file_url":"https://github.com/liqinye/EvoTD/blob/HEAD/src/prompts.py","link_basis":"first_harvest_node","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":"0359d2b5bb763b57"}},{"code_sha256_prefix":"ab2b0c92bd727ae9","entry":"has_test_input","repo":"liqinye/EvoTD","repo_kind":"found_in_text","path":"src/utils/parsers.py","file_url":"https://github.com/liqinye/EvoTD/blob/HEAD/src/utils/parsers.py","link_basis":"first_harvest_node","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":"ab2b0c92bd727ae9"}},{"code_sha256_prefix":"a3af72d6d85018e3","entry":"strip_json","repo":"liqinye/EvoTD","repo_kind":"found_in_text","path":"src/utils/parsers.py","file_url":"https://github.com/liqinye/EvoTD/blob/HEAD/src/utils/parsers.py","link_basis":"first_harvest_node","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":"a3af72d6d85018e3"}},{"code_sha256_prefix":"13bf57e5913a0b39","entry":"strip_think_token","repo":"liqinye/EvoTD","repo_kind":"found_in_text","path":"src/utils/parsers.py","file_url":"https://github.com/liqinye/EvoTD/blob/HEAD/src/utils/parsers.py","link_basis":"first_harvest_node","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":"13bf57e5913a0b39"}},{"code_sha256_prefix":"1fc961efe6e4123c","entry":"to_bool","repo":"liqinye/EvoTD","repo_kind":"found_in_text","path":"src/utils/equality.py","file_url":"https://github.com/liqinye/EvoTD/blob/HEAD/src/utils/equality.py","link_basis":"first_harvest_node","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":"1fc961efe6e4123c"}},{"code_sha256_prefix":"ab00e2f134f1d2d4","entry":"deep_equal","repo":"liqinye/EvoTD","repo_kind":"found_in_text","path":"src/utils/equality.py","file_url":"https://github.com/liqinye/EvoTD/blob/HEAD/src/utils/equality.py","link_basis":"first_harvest_node","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":"ab00e2f134f1d2d4"}},{"code_sha256_prefix":"d67cad27811abe22","entry":"format_reward","repo":"liqinye/EvoTD","repo_kind":"found_in_text","path":"src/reward_function.py","file_url":"https://github.com/liqinye/EvoTD/blob/HEAD/src/reward_function.py","link_basis":"first_harvest_node","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":"d67cad27811abe22"}},{"code_sha256_prefix":"d1604f4eadcbe65a","entry":"run_huggingface_download","repo":"liqinye/EvoTD","repo_kind":"found_in_text","path":"src/utils/remove_think_qwen3_tokenizer.py","file_url":"https://github.com/liqinye/EvoTD/blob/HEAD/src/utils/remove_think_qwen3_tokenizer.py","link_basis":"first_harvest_node","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":"d1604f4eadcbe65a"}}]},"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}