{"id":"arxiv:2004.12919","title":"First return, then explore","date":"2020-04-27","section":"Reinforcement learning","quality":0.582832,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/","source_url":"https://arxiv.org/abs/2004.12919","text_available":true,"authors":["Adrien Ecoffet","Joost Huizinga","Joel Lehman","Kenneth O. Stanley"],"outgoing":[{"source":"arxiv:2004.12919","target":"arxiv:1511.05952","evidence":"title","id":"arxiv:1511.05952","title":"Prioritized Experience Replay","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c88bf0c11c1817dcc457c001dfef5cd120a0946c8a009a0fe40648dae03eb7d4/"},{"source":"arxiv:2004.12919","target":"arxiv:1712.06567","evidence":"id","id":"arxiv:1712.06567","title":"Deep Neuroevolution: Genetic Algorithms Are a Competitive Alternative for Training Deep Neural Networks for Reinforcement Learning","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/7191dc11e224beb3dc9abb8cd4efc58c70b79939b61fd56e4416b4b5e1404312/"}],"incoming":[{"source":"arxiv:2206.13477","target":"arxiv:2004.12919","evidence":"title","id":"arxiv:2206.13477","title":"Parametrically Retargetable Decision-Makers Tend To Seek Power","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/e47c3ca813aa251e657f4fd4b126d747e91e7518bcab8f7e9139a9cf5c8d95fb/"},{"source":"arxiv:2505.22954","target":"arxiv:2004.12919","evidence":"title","id":"arxiv:2505.22954","title":"Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c29f716871f80314868b5647692de387f26e12166b93bd362915fd272b6e081c/"},{"source":"arxiv:2602.07755","target":"arxiv:2004.12919","evidence":"title","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:2608.23875","target":"arxiv:2004.12919","evidence":"title","id":"arxiv:2608.23875","title":"AI Finds A Way","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/afe34dd5ee58c2e9739f9651349c4c9180d95dabff51994a13339f48dfd0bb27/"}],"text_parts":[{"start":0,"end":15553,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-1.html"},{"start":15553,"end":29716,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-2.html"},{"start":29716,"end":45623,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-3.html"},{"start":45623,"end":60941,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-4.html"},{"start":60941,"end":69585,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-5.html"}],"contents":[{"title":"First return, then explore","offset":22,"level":1,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-1.html#text"},{"title":"Abstract","offset":52,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-1.html#text"},{"title":"The Go-Explore family of algorithms","offset":6328,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-1.html#text"},{"title":"Learning to play Atari when returning without a policy","offset":9754,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-1.html#text"},{"title":"A hard-exploration robotics environment","offset":17027,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-2.html#text"},{"title":"Policy-based Go-Explore","offset":20698,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-2.html#text"},{"title":"Conclusion","offset":25703,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-2.html#text"},{"title":"Supplementary Information","offset":27357,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-2.html#text"},{"title":"0.10 Detachment and derailment in reinforcement learning","offset":27387,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-2.html#text"},{"title":"0.11 Derailment in robotics","offset":34139,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-3.html#text"},{"title":"0.12 Policy-based Go-Explore and Stochastic Environments","offset":36451,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-3.html#text"},{"title":"0.13 Policy-based Go-Explore vs. DTSIL","offset":41030,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-3.html#text"},{"title":"0.14 No-ops and sticky actions","offset":49628,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-4.html#text"},{"title":"0.15 PPO and SIL","offset":52259,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-4.html#text"},{"title":"0.16 The backward algorithm","offset":57987,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-4.html#text"},{"title":"0.17 Go-Explore vs. Agent57","offset":64455,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-5.html#text"},{"title":"0.18 ALE issues","offset":67581,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-5.html#text"},{"title":"0.19 Infrastructure","offset":69128,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/part-5.html#text"}],"text_code_points":69585,"source_markdown":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/3cd790c501eaa633193918533947c58e30d1939bc08577606f1d88c56d368285/source.md"}