{"paper":{"title":"S3Gym: Can LLMs Turn Self-Testing and Self-Judging into Self-Improvement?","arxiv_id":"2608.31100","primary_cat":"cs.CL","submitted_at":"2026-08-31T17:05:41+00:00","version_compiled":"v1"},"notice":"Every quoted sentence is the authors' text from the arXiv listing abstract. Labels and probabilities are the outputs of classifier-1.13.0 on the question bank abstract-p1-20260923, exactly as returned. No sentence in this record was written by a language model.","schema":"pith.compiled-record/v0","status":"compiled","abstract":{"source":"arXiv listing abstract v1","sentences":[{"i":0,"p":0.99,"id":"s0","own":false,"role":"background","text":"Large language models (LLMs) increasingly interact with external environments and accumulate substantial behavioral experience, yet existing agent benchmarks largely evaluate them as fixed policies.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.06,"main_claim_p":0.02,"quantitative":false,"quantitative_p":0.16,"role_probabilities":{"gap":0.01,"method":0.0,"result":0.0,"background":0.99,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":1,"p":0.99,"id":"s1","own":false,"role":"gap","text":"It therefore remains unclear whether an agent can actively test its behavior, judge the resulting experience, and use that experience to improve future decisions.","hedge":{"p":0.41,"label":"asserted"},"own_p":0.04,"main_claim_p":0.03,"quantitative":false,"quantitative_p":0.03,"role_probabilities":{"gap":0.99,"method":0.0,"result":0.0,"background":0.01,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":2,"p":0.99,"id":"s2","own":true,"role":"contribution","text":"We introduce \\textbf{S\\textsuperscript{3}Gym}, an interactive benchmark for evaluating LLM self-improvement through three coupled capabilities: \\textbf{Self-Testing}, \\textbf{Self-Judging}, and \\textbf{Self-Improvement}.","hedge":{"p":0.99,"label":"asserted"},"own_p":0.91,"main_claim_p":0.05,"quantitative":true,"quantitative_p":0.62,"role_probabilities":{"gap":0.0,"method":0.01,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.99,"interpretation":0.0}},{"i":3,"p":0.89,"id":"s3","own":true,"role":"method","text":"S$^3$Gym separates permissive exploration from strict held-out evaluation and instantiates this protocol in seven text-based games with executable environment verifiers.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.91,"main_claim_p":0.03,"quantitative":true,"quantitative_p":0.88,"role_probabilities":{"gap":0.0,"method":0.89,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.11,"interpretation":0.0}},{"i":4,"p":0.94,"id":"s4","own":true,"role":"method","text":"We evaluate three pathways for incorporating interaction experience: direct History ICL, score-conditioned Summary Memory, and parameter Training.","hedge":{"p":0.77,"label":"asserted"},"own_p":0.85,"main_claim_p":0.03,"quantitative":true,"quantitative_p":0.79,"role_probabilities":{"gap":0.0,"method":0.94,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.06,"interpretation":0.0}},{"i":5,"p":1.0,"id":"s5","own":true,"role":"result","text":"Our experiments reveal that self-improvement is neither automatic nor uniform.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.92,"main_claim_p":0.26,"quantitative":false,"quantitative_p":0.09,"role_probabilities":{"gap":0.0,"method":0.0,"result":1.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":6,"p":0.87,"id":"s6","own":true,"role":"result","text":"Context-level experience improves performance for several model--game pairs, but the most effective pathway depends strongly on the task structure: summaries are beneficial when experience can be compressed into reusable strategic rules, yet often underperform raw history when success depends on precise, state-contingent information.","hedge":{"p":0.99,"label":"asserted"},"own_p":0.95,"main_claim_p":0.25,"quantitative":true,"quantitative_p":0.84,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.87,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.13}},{"i":7,"p":1.0,"id":"s7","own":true,"role":"result","text":"Parameter training produces substantial gains on some tasks, but also exhibits unstable improvement and severe negative transfer on others.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.94,"main_claim_p":0.11,"quantitative":false,"quantitative_p":0.42,"role_probabilities":{"gap":0.0,"method":0.0,"result":1.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":8,"p":1.0,"id":"s8","own":true,"role":"interpretation","text":"These findings show that recognizing successful actions is insufficient; agents must also transform feedback into executable and transferable policies.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.83,"main_claim_p":0.53,"quantitative":false,"quantitative_p":0.08,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":1.0}},{"i":9,"p":0.9,"id":"s9","own":true,"role":"contribution","text":"S$^3$Gym provides a unified framework for diagnosing this process and identifying the bottlenecks that prevent agents from translating interaction experience into reliable self-improvement.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.74,"main_claim_p":0.2,"quantitative":false,"quantitative_p":0.03,"role_probabilities":{"gap":0.0,"method":0.01,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.9,"interpretation":0.09}}]},"compiled":{"main_claim":{"p":0.88,"text":"These findings show that recognizing successful actions is insufficient; agents must also transform feedback into executable and transferable policies.","location":"abstract s8","hedge_strength":{"p":1.0,"label":"asserted"}},"provenance":{"route":{"via":"TypeSafe API","model":"classifier-1.13.0","endpoint":"https://api.typesafe.ai/v1/systemone","provider":"TypeSafe"},"packing":"one request per paper","requests":1,"input_tokens":7655,"cost_usd_reported":0.0},"compiled_at":"2026-09-23","bank_version":"abstract-p1-20260923","model_version":"classifier-1.13.0","abstract_states":{"claims_first":0.11,"states_limitation":0.7,"human_participants":0.04,"controlled_comparison":0.33,"randomized_assignment":0.02,"states_quantitative_result":0.16,"compares_with_prior_results":0.64,"mentions_code_or_data_availability":0.05},"evidential_basis":{"p":0.98,"value":"observation or experiment","probabilities":{"argument":0.0,"literature":0.0,"proof_or_derivation":0.0,"analysis_of_existing_data":0.0,"observation_or_experiment":0.98,"simulation_or_computation":0.02}},"contribution_type":{"p":1.0,"value":"dataset or benchmark","probabilities":{"review":0.0,"method_or_tool":0.0,"empirical_study":0.0,"simulation_study":0.0,"theoretical_result":0.0,"dataset_or_benchmark":1.0,"position_or_commentary":0.0}},"contribution_statement":{"p":1.0,"text":"We introduce \\textbf{S\\textsuperscript{3}Gym}, an interactive benchmark for evaluating LLM self-improvement through three coupled capabilities: \\textbf{Self-Testing}, \\textbf{Self-Judging}, and \\textbf{Self-Improvement}.","location":"abstract s2"}},"coverage":{"tier":"abstract","levels_run":["P abstract"],"not_examined":["full text","figures and tables","references and citing sentences","relations to other papers"],"sources_read":["arXiv listing abstract v1"],"levels_not_run":["S sections","T body sentences","C citations","R cross-paper relations"],"sentences_examined":10,"sentences_in_abstract":10},"measured_accuracy":{"url":"https://pith.science/accuracy","measured_on":"2026-10-02","method":"two independent model families labelled a fresh random sample blind; the figure is how often the published label matches their consensus","sentence_role_agreement":0.92,"main_claim_agreement":0.908,"typed_citation_agreement":0.933,"typed_citation_agreement_95_interval":[0.917,0.945],"review_comments_accurate":0.758,"review_comments_accurate_95_interval":[0.684,0.82],"review_caught_admitted_error_found_or_related":0.448,"review_comments_judged":{"judged_by":["gpt-5.6-luna","grok-4.7"],"judged_against":"Pith's own text copy of each paper: text extracted from the PDF, the first 90,000 characters","measured_on":"2026-10-04","accurate":113,"decided":149,"sampled_reviews":200,"split_by_lane":false,"by_lane":"https://pith.science/accuracy#by-lane"}}}