{"id":"arxiv:2608.19197","title":"SPADE: Self-Play in Adaptive Synthetic Executable Environments","date":"2026-08-19","section":"Agents, open-endedness, AGI","quality":0.382295,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/","source_url":"https://arxiv.org/abs/2608.19197","text_available":true,"authors":["Bo Liu","Simon Yu","Yiding Jiang","Ao Qu"],"outgoing":[{"source":"arxiv:2608.19197","target":"arxiv:1901.01753","evidence":"id","id":"arxiv:1901.01753","title":"Paired Open-Ended Trailblazer (POET): Endlessly Generating Increasingly Complex and Diverse Learning Environments and Their Solutions","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/a9790a1ede488fbbb2d7b4cdb97b884a5a1e21357b72c5e3419a83ea629d5a4b/"},{"source":"arxiv:2608.19197","target":"arxiv:1905.10985","evidence":"id","id":"arxiv:1905.10985","title":"AI-GAs: AI-generating algorithms, an alternate paradigm for producing general artificial intelligence","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/089142f7297bed414d01b82c78dc888e4018fecff7efbe5cae66442e03f87c4b/"},{"source":"arxiv:2608.19197","target":"arxiv:2402.03300","evidence":"id","id":"arxiv:2402.03300","title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/636fb871904bc06f742cfdea069796f9e9665f95bd58dc63260eaff5154ef985/"},{"source":"arxiv:2608.19197","target":"arxiv:2406.04268","evidence":"id","id":"arxiv:2406.04268","title":"Open-Endedness is Essential for Artificial Superhuman Intelligence","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/13964b2db9a9f793d1fcae07ca83fbe09f65454214269cac56777cb8be423a4e/"},{"source":"arxiv:2608.19197","target":"arxiv:2503.14476","evidence":"id","id":"arxiv:2503.14476","title":"DAPO: An Open-Source LLM Reinforcement Learning System at Scale","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/548e934b912ef84691ace63bf1cfd670e7b0257d9d6986dde926f569e7b2f8b7/"},{"source":"arxiv:2608.19197","target":"arxiv:2505.21444","evidence":"id","id":"arxiv:2505.21444","title":"Can Large Reasoning Models Self-Train?","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/206224ca50098e0bdc230140722feec107cfc90c1cef2ed8d3ed52355ef56bca/"},{"source":"arxiv:2608.19197","target":"arxiv:2511.07317","evidence":"id","id":"arxiv:2511.07317","title":"RLVE: Scaling Up Reinforcement Learning for Language Models with Adaptive Verifiable Environments","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/"},{"source":"arxiv:2608.19197","target":"arxiv:2512.18552","evidence":"id","id":"arxiv:2512.18552","title":"Toward Training Superintelligent Software Agents through Self-Play SWE-RL","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/da99b397e01b725186243c38a2dc6072d3a5c81bd8f759e898f04ad09aa0f4b1/"},{"source":"arxiv:2608.19197","target":"arxiv:2601.03192","evidence":"id","id":"arxiv:2601.03192","title":"MemRL: Self-Evolving Agents via Runtime Reinforcement Learning on Episodic Memory","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/4f5bbf3318b1fe409e0ef5dad02157180b844de366d194e82ce74bc997912eff/"},{"source":"arxiv:2608.19197","target":"arxiv:2602.07755","evidence":"id","id":"arxiv:2602.07755","title":"Learning to Continually Learn via Meta-learning Agentic Memory Designs","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/4d0a0ea160495ea50583e0e24027b6085e2f430a44195e3053ba69cda8cc82ea/"},{"source":"arxiv:2608.19197","target":"arxiv:2603.19461","evidence":"id","id":"arxiv:2603.19461","title":"Hyperagents","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/17eec8a15debc212b5530f117dd8ba8bad339590729cbbc4f75a47755ebcb544/"}],"incoming":[],"text_parts":[{"start":0,"end":14309,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-1.html"},{"start":14309,"end":29877,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-2.html"},{"start":29877,"end":45254,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-3.html"},{"start":45254,"end":60799,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html"},{"start":60799,"end":70489,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-5.html"},{"start":70489,"end":84238,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-6.html"},{"start":84238,"end":99159,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-7.html"},{"start":99159,"end":111752,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html"}],"contents":[{"title":"SPADE ♠ \\spadesuit : Self-Play in Adaptive Synthetic Executable Environments","offset":455,"level":1,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-1.html#text"},{"title":"Abstract","offset":535,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-1.html#text"},{"title":"1 Introduction","offset":2342,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-1.html#text"},{"title":"2 Related Work","offset":9378,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-1.html#text"},{"title":"3 Preliminaries","offset":16550,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-2.html#text"},{"title":"3.1 Markov Decision Processes and Gym-Style Environments","offset":16861,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-2.html#text"},{"title":"3.2 RLVR and GRPO","offset":17893,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-2.html#text"},{"title":"4 SPADE : Self-Play in Adaptive Synthetic Executable Environments","offset":19100,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-2.html#text"},{"title":"4.1 Dual-Role Self-Play and Code-as-Environment","offset":19756,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-2.html#text"},{"title":"4.2 Hint-Based Regret Reward","offset":23929,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-2.html#text"},{"title":"4.3 Environment Design","offset":26346,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-2.html#text"},{"title":"5 Experimental Setup","offset":28575,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-2.html#text"},{"title":"5.1 SPADE Training Recipe","offset":28600,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-2.html#text"},{"title":"5.2 Domain: Games","offset":30317,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-3.html#text"},{"title":"5.3 Domain: Tool Use","offset":32033,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-3.html#text"},{"title":"6 Experimental Results","offset":33939,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-3.html#text"},{"title":"6.1 Quantitative Analysis","offset":34280,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-3.html#text"},{"title":"6.2 Qualitative Analysis","offset":35766,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-3.html#text"},{"title":"7 Ablations","offset":38828,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-3.html#text"},{"title":"7.1 Ablation: Environment Designer Adaptation","offset":39735,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-3.html#text"},{"title":"7.2 Ablation: Environment Designer Reward","offset":40849,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-3.html#text"},{"title":"8 Scaling Results","offset":44739,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-3.html#text"},{"title":"9 Discussion","offset":45965,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"10 Conclusion","offset":48495,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Acknowledgments","offset":49681,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Appendix E Theoretical Analysis [back to contents]","offset":50365,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"E.1 Setup and Assumptions","offset":51318,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Environment class.","offset":51349,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Distribution notation.","offset":51840,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Inference protocol.","offset":52452,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Fine-tuning budget.","offset":52906,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Returns with and without hints.","offset":53260,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Game payoffs.","offset":54694,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Pure Nash equilibrium.","offset":55248,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Optimal values.","offset":55747,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Assumption E.1 (Sound generation) .","offset":55910,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Assumption E.2 (Articulated hints) .","offset":56299,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Assumption E.3 (Internalizability of hinted behavior) .","offset":56618,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"E.2 Main Results","offset":56938,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Lemma E.4 (Hint-regret equals hint-free regret) .","offset":57863,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Proof.","offset":58280,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Theorem E.5 (Nash equilibria imply hint-free optimality on every environment) .","offset":58521,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Proof.","offset":59326,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-4.html#text"},{"title":"Appendix F Method and Experiment Details [back to contents]","offset":61636,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-5.html#text"},{"title":"F.1 System Prompts and Templates","offset":61700,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-5.html#text"},{"title":"Environment-generation prompt.","offset":61738,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-5.html#text"},{"title":"Multi-turn environment-generation prompt.","offset":62474,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-5.html#text"},{"title":"Corpus-grounded interactive generation prompt.","offset":62770,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-5.html#text"},{"title":"Tool-use generation prompt.","offset":63102,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-5.html#text"},{"title":"Reasoning Agent gameplay prompt.","offset":63540,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-5.html#text"},{"title":"F.2 Implementation Details","offset":63694,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-5.html#text"},{"title":"F.3 Reproducibility","offset":65471,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-5.html#text"},{"title":"F.3.1 Comparability Notes for Table 2","offset":65727,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-5.html#text"},{"title":"Appendix G Extended Ablations [back to contents]","offset":67131,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-5.html#text"},{"title":"Appendix H Extended Related Work [back to contents]","offset":68126,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-5.html#text"},{"title":"H.1 Self-Play for LLMs","offset":68182,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-5.html#text"},{"title":"H.2 Unsupervised Environment Design and Open-Endedness","offset":77848,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-6.html#text"},{"title":"H.3 Synthetic Environment Generation","offset":84196,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-6.html#text"},{"title":"H.4 Environment Scaling","offset":93185,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-7.html#text"},{"title":"H.5 Agentic Memory Design and Self-Improving Code Systems","offset":96950,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-7.html#text"},{"title":"Appendix I Extended Quantitative Analysis [back to contents]","offset":98387,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-7.html#text"},{"title":"I.1 Games","offset":98996,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-7.html#text"},{"title":"I.1.1 Environment Quality","offset":99127,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-7.html#text"},{"title":"I.1.2 Environment Diversity","offset":100773,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"Metric and calibration.","offset":100807,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"Diversity is maintained at the ceiling.","offset":101459,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"Corpus grounding drives diversity.","offset":102211,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"Scope and limitations.","offset":103384,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"Appendix J Extended Qualitative Analysis [back to contents]","offset":103877,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"J.1 Games","offset":104590,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"J.1.1 Additional Privileged-Hint Examples","offset":104744,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"J.1.2 Environment Evolution over Training","offset":105208,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"J.1.3 Matched-Step Generation Contrast","offset":106226,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"J.2 Tool Use","offset":106784,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"J.3 Generated Environment Gallery","offset":108739,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"J.3.1 A Minimal Generated Environment","offset":108925,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"J.3.2 Full Source of Two Exemplar Generated Environments","offset":109454,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"Training step 0: CarOwnershipDisputeEnv (370 lines).","offset":110098,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"Training step 384: ThermodynamicCycleManipulationLabEnv (376 lines).","offset":110161,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"J.3.3 Full Source of a Generated Tool-Use Environment","offset":110240,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"Training step 392: CustomerSupportTicketWorkflowEnv (200 lines).","offset":110713,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"},{"title":"Instructions for reporting errors","offset":110788,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/part-8.html#text"}],"text_code_points":111752,"source_markdown":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/source.md"}