{"id":"arxiv:2504.20571","title":"Reinforcement Learning for Reasoning in Large Language Models with One Training Example","date":"2025-04-29","section":"Reasoning and the \"physics\" of language models","quality":0.214668,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/","source_url":"https://arxiv.org/abs/2504.20571","text_available":true,"authors":["Yiping Wang","Qing Yang","Zhiyuan Zeng"],"outgoing":[{"source":"arxiv:2504.20571","target":"arxiv:2301.05217","evidence":"id","id":"arxiv:2301.05217","title":"Progress measures for grokking via mechanistic interpretability","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/f71e3b65a8b3f35beb171d17f6c65878e0914227a0042ccba59b88952554baef/"},{"source":"arxiv:2504.20571","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:2504.20571","target":"arxiv:2407.21783","evidence":"id","id":"arxiv:2407.21783","title":"The Llama 3 Herd of Models","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/de3cebb4c28bc5d41a843279f219dca14476284991ea5c7c0aeab7068a40e306/"},{"source":"arxiv:2504.20571","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:2504.20571","target":"arxiv:2503.20783","evidence":"id","id":"arxiv:2503.20783","title":"Understanding R1-Zero-Like Training: A Critical Perspective","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b49a056f3ce8e269dd65e3b11605927f3c4d0ef074bc68e0b1d7f84e6c9c6c6a/"},{"source":"arxiv:2504.20571","target":"arxiv:2504.13837","evidence":"id","id":"arxiv:2504.13837","title":"Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b90f70269aeb02311d74ffaadf47e139683424a47118dc18619000ba72174d17/"},{"source":"arxiv:2504.20571","target":"arxiv:2504.16084","evidence":"id","id":"arxiv:2504.16084","title":"TTRL: Test-Time Reinforcement Learning","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/ad54dd58d24afdfefc8aa5087d07c4aab2143e90ef06f6f7db8d3f3333452c48/"}],"incoming":[{"source":"arxiv:2504.16084","target":"arxiv:2504.20571","evidence":"id","id":"arxiv:2504.16084","title":"TTRL: Test-Time Reinforcement Learning","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/ad54dd58d24afdfefc8aa5087d07c4aab2143e90ef06f6f7db8d3f3333452c48/"},{"source":"arxiv:2505.15134","target":"arxiv:2504.20571","evidence":"id","id":"arxiv:2505.15134","title":"The Unreasonable Effectiveness of Entropy Minimization in LLM Reasoning","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/12b01b403e280a32d41fd991a27fa30f33315e09f0f42a4f43e9bece5ad9c53d/"},{"source":"arxiv:2505.21444","target":"arxiv:2504.20571","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:2506.01939","target":"arxiv:2504.20571","evidence":"id","id":"arxiv:2506.01939","title":"Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM Reasoning","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/e0f4a55bca612bce5895b6b98f89b0efc14cbd8cf45946a7743347cf6daee1a1/"},{"source":"arxiv:2506.08007","target":"arxiv:2504.20571","evidence":"title","id":"arxiv:2506.08007","title":"Reinforcement Pre-Training","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/a5ae77d4a2dcd6843e856083ecc6c30a72563e6896503d6efdfae117b8b007e6/"},{"source":"arxiv:2506.10947","target":"arxiv:2504.20571","evidence":"id","id":"arxiv:2506.10947","title":"Spurious Rewards: Rethinking Training Signals in RLVR","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/f48e13a9656d8eff1882eb80988e07a820c4799401298947deac735c004e8592/"},{"source":"arxiv:2506.21734","target":"arxiv:2504.20571","evidence":"id","id":"arxiv:2506.21734","title":"Hierarchical Reasoning Model","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3604312101ff09ab9903e7d6a32c7b2b8a9a7ece6a9d145d04ea01c9d13ffa3a/"},{"source":"arxiv:2507.12856","target":"arxiv:2504.20571","evidence":"id","id":"arxiv:2507.12856","title":"Supervised Fine Tuning on Curated Data is Reinforcement Learning (and can be improved)","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/5f6c1906eaa67369d9ff15b974756b89141d1b4cf9d7c0f6554e0d56f26e79be/"},{"source":"arxiv:2509.25123","target":"arxiv:2504.20571","evidence":"id","id":"arxiv:2509.25123","title":"From $f(x)$ and $g(x)$ to $f(g(x))$: LLMs Learn New Skills in RL by Composing Old Ones","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/7fdfe7abe26ee324fe7abad3ce802148163129a7c157d77bdec859c990868e03/"},{"source":"arxiv:2601.07055","target":"arxiv:2504.20571","evidence":"id","id":"arxiv:2601.07055","title":"Dr. Zero: Self-Evolving Search Agents without Training Data","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3baa3e7fc5768b75e4ef3e64c3cc5297fa42ef19b03bc918ef5352e437e1c168/"},{"source":"arxiv:2601.14525","target":"arxiv:2504.20571","evidence":"title","id":"arxiv:2601.14525","title":"Towards Execution-Grounded Automated AI Research","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/895006e818735c9a565dbe99ae31404366eefd6bdc2278612271c65a8493e4a5/"},{"source":"arxiv:2601.16175","target":"arxiv:2504.20571","evidence":"id","id":"arxiv:2601.16175","title":"Learning to Discover at Test Time","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/5d0b45cbbc72945765e689cbfb7ac7e4e92dfbb764861b27682ac0744ef05b70/"},{"source":"arxiv:2602.09000","target":"arxiv:2504.20571","evidence":"id","id":"arxiv:2602.09000","title":"iGRPO: Self-Feedback-Driven LLM Reasoning","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1cd6a66823fe24e8e2d83e817177d3a1e8cda1d7d5b59a4397f8669bee6306ae/"},{"source":"arxiv:2605.06241","target":"arxiv:2504.20571","evidence":"id","id":"arxiv:2605.06241","title":"Rethinking RL for LLM Reasoning: It's Sparse Policy Selection, Not Capability Learning","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c9e4fdd6e0c24f5a635ff70c2bca19da19c4012a05f69ceb20fdc0a86fd8b775/"}],"text_parts":[{"start":0,"end":14979,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-1.html"},{"start":14979,"end":30692,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-2.html"},{"start":30692,"end":46485,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html"},{"start":46485,"end":61937,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html"},{"start":61937,"end":67126,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-5.html"}],"contents":[{"title":"Reinforcement Learning for Reasoning in Large Language Models with One Training Example","offset":0,"level":1,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-1.html#text"},{"title":"Abstract","offset":91,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-1.html#text"},{"title":"1 Introduction","offset":2054,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-1.html#text"},{"title":"2 Preliminary","offset":7636,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-1.html#text"},{"title":"RL Loss Function.","offset":7654,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-1.html#text"},{"title":"Data Selection: Historical Variance Score.","offset":9035,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-1.html#text"},{"title":"3 Experiments","offset":11306,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-1.html#text"},{"title":"3.1 Setup","offset":11324,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-1.html#text"},{"title":"3.2 Observation of 1/Few-Shot RLVR","offset":14414,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-1.html#text"},{"title":"3.2.1 Dissection of π 1 \\pi_{1} : A Not-So-Difficult Problem","offset":14912,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-1.html#text"},{"title":"3.2.2 Post-saturation Generalization: Generalization After Training Accuracy Saturation","offset":16440,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-2.html#text"},{"title":"3.2.3 1-shot RLVR is Effective for Many Examples & Brings Improvements across Categories","offset":18198,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-2.html#text"},{"title":"3.2.4 More Frequent Self-Reflection on Test Data","offset":20476,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-2.html#text"},{"title":"3.3 1/Few-shot RLVR on Other Models/Algorithms","offset":21828,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-2.html#text"},{"title":"4 Analysis","offset":23004,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-2.html#text"},{"title":"4.1 Ablation Study: Policy Gradient Loss is the Main Contributor, and Entropy Loss Further Improve Post-Saturation Generalization","offset":23446,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-2.html#text"},{"title":"4.2 Entropy-Loss-Only Training & Label Correctness","offset":26003,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-2.html#text"},{"title":"5 Conclusion","offset":27912,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-2.html#text"},{"title":"6 Acknoledgements","offset":28978,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-2.html#text"},{"title":"Appendix A Related Work","offset":29470,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-2.html#text"},{"title":"Appendix B Experiment Setup","offset":31857,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"B.1 Details of Loss Function","offset":31889,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"B.2 Training Dataset","offset":36780,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"DeepScaleR-sub.","offset":36806,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"MATH.","offset":37385,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"B.3 Evaulation Dataset","offset":38069,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"MATH500.","offset":38350,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"AIME 2024/2025.","offset":38616,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"AMC 2023.","offset":38850,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"Minerva Math.","offset":39279,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"OlympiadBench.","offset":39960,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"ARC-Challenge/Easy.","offset":40476,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"B.4 More Training Details","offset":41309,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"B.5 More Evaluation Details","offset":42058,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"B.6 Performance Difference on Initial Model","offset":43437,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"Appendix C Evaluation Result","offset":45065,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"C.1 Main Experiments","offset":45098,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"C.1.1 Detailed performance on Qwen2.5-Math-1.5B.","offset":45124,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"C.1.2 Detailed Performance on More Models and Training Examples.","offset":45328,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"C.1.3 Detailed performance with best per-benchmark results","offset":46420,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-3.html#text"},{"title":"C.1.4 Detailed Test curves on MATH500 for 1-shot RLVR on Qwen2.5-Math-1.5B.","offset":47419,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"C.1.5 Detailed RLVR results on eacn benchmark over training process.","offset":47829,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"C.1.6 More Evaluation on DeepSeek-R1-Distill-Qwen-1.5B","offset":48589,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"C.2 Analysis","offset":48802,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"C.2.1 Test Curves for Ablation Study","offset":48820,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"C.2.2 Entropy loss","offset":49074,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"Detailed results of entropy-loss-only training.","offset":49099,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"Discussion of entropy loss and its function in 1-shot RLVR.","offset":49713,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"C.2.3 (Only) Format Correction?","offset":50559,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"(a). Applying format reward in RLVR.","offset":50935,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"(b) Observe the change of format in 1-shot RLVR.","offset":52667,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"C.2.4 Influence of Random Wrong Labels","offset":55551,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"C.2.5 Change the Prompt of π 1 \\pi_{1}","offset":56636,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"C.3 Response Length","offset":57645,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"C.4 Pass@8 Results","offset":57850,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"Appendix D Discussions","offset":58273,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"D.1 Limitations of Our Work","offset":58300,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"D.2 Reasoning Capability of Base Models","offset":60041,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"D.3 Why Model Continues Improving After the Training Accuracy Reaches Near 100%?","offset":61048,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"D.4 Future Works","offset":61802,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"Data Selection and Curation.","offset":61901,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-4.html#text"},{"title":"Understanding 1-shot RLVR and Post-saturation Generalization","offset":62622,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-5.html#text"},{"title":"Importance of Exploration.","offset":63603,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-5.html#text"},{"title":"Other Applications.","offset":64793,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-5.html#text"},{"title":"Appendix E Example Details","offset":65488,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-5.html#text"},{"title":"Instructions for reporting errors","offset":66162,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/part-5.html#text"}],"text_code_points":67126,"source_markdown":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b2b21c3df49619588c371411dba9467a41f81b3fd4da5441898b5145bb0de120/source.md"}