{"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-2605-08083","title":"LLMs Improving LLMs: Agentic Discovery for Test-Time Scaling","arxiv_id":"2605.08083","date":"2026-05-08","proceeding":null,"authors":["Tong Zheng","Haolin Liu","Chengsong Huang","Huiwen Bao","Sheng Zhang","Rui Liu","Runpeng Dai","Ruibo Chen","Chenxi Liu","Tianyi Xiong","Xidong Wu","Hongming Zhang","Heng Huang"],"abstract":"Test-time scaling (TTS) has become an effective approach for improving large language model performance by allocating additional computation during inference. However, existing TTS strategies are largely hand-crafted: researchers manually design reasoning patterns and tune heuristics by intuition, leaving much of the computation-allocation space unexplored. We propose an environment-driven framework, AutoTTS, that changes what researchers design: from individual TTS heuristics to environments where TTS strategies can be discovered automatically. The key to AutoTTS lies in environment construction: the discovery environment must make the control space tractable and provide cheap, frequent feedback for TTS search. As a concrete instantiation, we formulate width--depth TTS as controller synthesis over pre-collected reasoning trajectories and probe signals, where controllers decide when to branch, continue, probe, prune, or stop and can be evaluated cheaply without repeated LLM calls. We further introduce beta parameterization to make the search tractable and fine-grained execution trace feedback to improve discovery efficiency by helping the agent diagnose why a TTS program fails. Experiments on mathematical reasoning benchmarks show that the discovered strategies improve the overall accuracy--cost tradeoff over strong manually designed baselines. The discovered strategies generalize to held-out benchmarks and model scales, while the entire discovery costs only $39.9 and 160 minutes. Our data, and code will be open-source at https://github.com/zhengkid/AutoTTS.","url_abs":"https://arxiv.org/abs/2605.08083","url_pdf":"https://arxiv.org/pdf/2605.08083","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":"https://app.syntology.ai/?focus=2605.08083","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2605.08083"}},"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/zhengkid/AutoTTS","reach":null}],"summary":{"ran":4},"by_repo_kind":{"found_in_text":{"samples":4,"ran":4,"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":4,"samples":[{"code_sha256_prefix":"c66918620c0773fb","entry":"load_prompt","repo":"zhengkid/AutoTTS","repo_kind":"found_in_text","path":"efficient_reasoning_controller/workspace/controller_search/claude_proposer.py","file_url":"https://github.com/zhengkid/AutoTTS/blob/HEAD/efficient_reasoning_controller/workspace/controller_search/claude_proposer.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":"c66918620c0773fb"}},{"code_sha256_prefix":"fbc864cd6fceb2d4","entry":"make_problem_trace","repo":"zhengkid/AutoTTS","repo_kind":"found_in_text","path":"efficient_reasoning_controller/eval/trace_schema.py","file_url":"https://github.com/zhengkid/AutoTTS/blob/HEAD/efficient_reasoning_controller/eval/trace_schema.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":"fbc864cd6fceb2d4"}},{"code_sha256_prefix":"99ea9f901626ea5c","entry":"make_step_trace","repo":"zhengkid/AutoTTS","repo_kind":"found_in_text","path":"efficient_reasoning_controller/eval/trace_schema.py","file_url":"https://github.com/zhengkid/AutoTTS/blob/HEAD/efficient_reasoning_controller/eval/trace_schema.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":"99ea9f901626ea5c"}},{"code_sha256_prefix":"1ae73d61385c10fe","entry":"message_to_jsonable","repo":"zhengkid/AutoTTS","repo_kind":"found_in_text","path":"efficient_reasoning_controller/workspace/controller_search/claude_proposer.py","file_url":"https://github.com/zhengkid/AutoTTS/blob/HEAD/efficient_reasoning_controller/workspace/controller_search/claude_proposer.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":"1ae73d61385c10fe"}}]},"arxiv_metadata":{"licence":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)","fields":["title","abstract","authors","date"],"primary_category":"cs.CL","source":"arxiv_2026.jsonl"},"syntology_extracted_results":null}