| react-profiling | 0 |
| react-import-map | react,react-dom,react-dom/client,react-dom/profiling,react-is,react-reconciler,react/compiler-runtime,react/jsx-dev-runtime,react/jsx-runtime,scheduler@d5750093facf |
| route-pattern | /:user_id/:repository/blob/*name(/*path) |
| route-controller | blob |
| route-action | show |
| fetch-nonce | v2:bd88bed0-97dc-1c8b-6c98-73517115ff1f |
| current-catalog-service-hash | f3abb0cc802f3d7b95fc8762b94bdcb13bf39634c40c357301c4aa1d67a256fb |
| request-id | AF80:1463F:1496CF9:1AF1AE1:6AED762D |
| html-safe-nonce | 2edac8c90cfba2251cb60426f04eaaccca78fb4420766a9f3cf6606f3eca1963 |
| visitor-payload | eyJyZWZlcnJlciI6IiIsInJlcXVlc3RfaWQiOiJBRjgwOjE0NjNGOjE0OTZDRjk6MUFGMUFFMTo2QUVENzYyRCIsInZpc2l0b3JfaWQiOiI1ODA0NzM3ODYwOTk2MDAzMzczIiwicmVnaW9uX2VkZ2UiOiJpYWQiLCJyZWdpb25fcmVuZGVyIjoiaWFkIn0= |
| visitor-hmac | af22d6f0a0bdcbef0a2c3656d8f6b8a97f15c4bec0243474da4c36716205c002 |
| hovercard-subject-tag | repository:1406785632 |
| github-keyboard-shortcuts | repository,source-code,file-tree,copilot |
| google-site-verification | Apib7-x98H0j5cPqHWwSMm6dNU4GmODRoqxLiDzdx9I |
| octolytics-url | https://collector.github.com/github/collect |
| analytics-location | ///blob/show |
| fb:app_id | 1401488693436528 |
| apple-itunes-app | app-id=1477376905, app-argument=https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md |
| twitter:image | https://opengraph.githubassets.com/c940270926a8631ae50e012e68e8465427d747e37d7f58031ac81ab6074c89f4/loopx-agent/CS-Notes |
| twitter:card | summary_large_image |
| og:image | https://opengraph.githubassets.com/c940270926a8631ae50e012e68e8465427d747e37d7f58031ac81ab6074c89f4/loopx-agent/CS-Notes |
| og:image:alt | 我的自学笔记，终身更新. Contribute to loopx-agent/CS-Notes development by creating an account on GitHub. |
| og:image:width | 1200 |
| og:image:height | 600 |
| og:site_name | GitHub |
| og:type | object |
| hostname | github.com |
| expected-hostname | github.com |
| None | 56fb42677ae87ed3e64410323d67980aa65d0485568a7858f874e7edfc6d1c11 |
| turbo-cache-control | no-cache |
| go-import | github.com/loopx-agent/CS-Notes git https://github.com/loopx-agent/CS-Notes.git |
| octolytics-dimension-user_id | 337587101 |
| octolytics-dimension-user_login | loopx-agent |
| octolytics-dimension-repository_id | 1406785632 |
| octolytics-dimension-repository_nwo | loopx-agent/CS-Notes |
| octolytics-dimension-repository_public | true |
| octolytics-dimension-repository_is_fork | true |
| octolytics-dimension-repository_parent_id | 256262711 |
| octolytics-dimension-repository_parent_nwo | huangruiteng/CS-Notes |
| octolytics-dimension-repository_network_root_id | 256262711 |
| octolytics-dimension-repository_network_root_nwo | huangruiteng/CS-Notes |
| turbo-body-classes | logged-out env-production page-responsive |
| disable-turbo | false |
| browser-stats-url | https://api.github.com/_private/browser/stats |
| browser-errors-url | https://api.github.com/_private/browser/errors |
| release | 9c9a37e2423ed382cf39f70b5aa4b4858b40a3b0 |
| ui-target | full |
| theme-color | #1e2327 |
| color-scheme | light dark |
| Skip to content | https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#start-of-content |
| https://github.com/ |
| Sign in | https://github.com/login?return_to=https%3A%2F%2Fgithub.com%2Floopx-agent%2FCS-Notes%2Fblob%2Fmaster%2FNotes%2FReinforcement-Learning.md |
| GitHubRepository hosting, code review, and collaboration | https://github.com/platform |
| CopilotAI coding agent for tasks, projects, and automated work | https://github.com/features/copilot |
| ActionsCI/CD and automation | https://github.com/features/actions |
| Code QualityQuality analysis, fixes, and enforcement | https://github.com/features/code-quality |
| CodespacesCloud development environments | https://github.com/features/codespaces |
| Code SecurityVulnerability detection and remediation | https://github.com/security/advanced-security/code-security |
| Secret ProtectionSecret scanning and leak prevention | https://github.com/security/advanced-security/secret-protection |
| All Features | https://github.com/features |
| Documentation | https://docs.github.com/en |
| Training | https://learn.github.com/skills |
| Certifications | https://learn.github.com/credentials |
| Build with Copilot | https://github.com/features/ai |
| Customize Copilot | https://docs.github.com/en/copilot/how-tos/copilot-on-github/customize-copilot/customize-copilot-overview |
| Integrations | https://github.com/marketplace |
| Trending repos | https://github.com/trending |
| GitHub Sponsors | https://github.com/open-source/sponsors |
| Developer articles | https://github.com/resources/articles |
| ChangelogYour source for the latest features, improvements, and fixes across GitHub products, all in one place. | https://github.blog/changelog/ |
| Enterprises | https://github.com/enterprise |
| Team | https://github.com/team |
| Startups | https://github.com/enterprise/startups |
| Nonprofits | https://github.com/solutions/industry/nonprofits |
| Healthcare | https://github.com/solutions/industry/healthcare |
| Financial services | https://github.com/solutions/industry/financial-services |
| Manufacturing | https://github.com/solutions/industry/manufacturing |
| Government | https://github.com/solutions/industry/government |
| See all industries | https://github.com/solutions/industry |
| Why GitHub | https://github.com/why-github |
| Customer Stories | https://github.com/customer-stories |
| Business insights | https://github.com/solutions/executive-insights |
| Trust Center | https://github.com/trust-center |
| GitHub Advanced SecurityEnterprise-grade code security | https://github.com/security/advanced-security |
| Copilot for businessEnterprise-grade AI features | https://github.com/features/copilot/copilot-business |
| Premium SupportEnterprise-grade 24/7 support | https://github.com/enterprise/premium-support |
| Documentation | https://docs.github.com/en |
| Resource center | https://github.com/resources |
| Blog | https://github.blog/ |
| Events & webinars | https://github.com/resources/events |
| Community Forum | https://github.com/orgs/community/discussions |
| Customer Stories | https://github.com/customer-stories |
| Integrations | https://github.com/marketplace |
| Customer support | https://help.github.com/ |
| Training | https://learn.github.com/skills |
| GitHub Status | https://www.githubstatus.com/ |
| About | https://github.com/about |
| Partners | https://github.com/partners |
| Changelog | https://github.blog/changelog/ |
| GitHub UniverseThe event for developers, by developers — talks, workshops, and what’s next for GitHub. | https://githubuniverse.com/ |
| Pricing | https://github.com/pricing |
| Sign in | https://github.com/login?return_to=https%3A%2F%2Fgithub.com%2Floopx-agent%2FCS-Notes%2Fblob%2Fmaster%2FNotes%2FReinforcement-Learning.md |
| Sign up | https://github.com/signup?ref_cta=Sign+up&ref_loc=header+logged+out&ref_page=%2F%3Cuser-name%3E%2F%3Crepo-name%3E%2Fblob%2Fshow&source=header-repo&source_repo=loopx-agent%2FCS-Notes |
| Reload | https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md |
| Reload | https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md |
| Reload | https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md |
| loopx-agent | https://github.com/loopx-agent |
| CS-Notes | https://github.com/loopx-agent/CS-Notes |
| Code | https://github.com/loopx-agent/CS-Notes |
| Pull requests | https://github.com/loopx-agent/CS-Notes/pulls |
| Actions | https://github.com/loopx-agent/CS-Notes/actions |
| Projects | https://github.com/loopx-agent/CS-Notes/projects |
| Security and quality | https://github.com/loopx-agent/CS-Notes/security |
| Insights | https://github.com/loopx-agent/CS-Notes/pulse |
| https://github.com/loopx-agent/CS-Notes |
| CS-Notes | https://github.com/loopx-agent/CS-Notes/tree/master |
| Notes | https://github.com/loopx-agent/CS-Notes/tree/master/Notes |
| History | https://github.com/loopx-agent/CS-Notes/commits/master/Notes/Reinforcement-Learning.md |
| https://github.com/loopx-agent/CS-Notes/commits/master/Notes/Reinforcement-Learning.md |
| CS-Notes | https://github.com/loopx-agent/CS-Notes/tree/master |
| Notes | https://github.com/loopx-agent/CS-Notes/tree/master/Notes |
| Raw | https://github.com/loopx-agent/CS-Notes/raw/refs/heads/master/Notes/Reinforcement-Learning.md |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#reinforcement-learning |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#基本问题 |
| 深度强化学习（一）强化学习概述 - iker peng的文章 - 知乎 | https://zhuanlan.zhihu.com/p/22542101 |
| 深度强化学习系列（二）强化学习基础 - iker peng的文章 - 知乎 | https://zhuanlan.zhihu.com/p/23436744 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#rl-与监督学习--推荐系统的区别 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#什么时候-rl-是必要的 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#为什么大模型通常不是从零-rl |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#核心难点与理论抓手 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#mdp--pomdp |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#state--observation--belief--reward |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#credit-assignment |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#counterfactual-branch |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#exploration-vs-exploitation |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#on-policy--off-policy |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#behavior-cloning--sft |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#policy-support--kl-shift |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#rollout--trajectory--episode |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#counterfactual-evaluation--off-policy-evaluation |
| Counterfactual Reasoning and Learning Systems | https://jmlr.csail.mit.edu/beta/papers/v14/bottou13a.html |
| arXiv | https://arxiv.org/abs/1209.2355 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#haty |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#markov-factor-replacement |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#y |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#y-1 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#haty_is |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#overlapclipping-与置信区间 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#w_iclip |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#doubly-robust用-predictor-降方差 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#hatv_drpi |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#从-evaluation-到-learning |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#theta |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#reward-hacking |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#continual-learning--interference |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#专题从-rl-不易遗忘到推荐系统缺陷 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#on-policy-为什么可能缓解遗忘 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/rl-sft-forgetting-curve.jpg |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#low-curvature--off-principal-更新 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/rlvr-off-principal-optimization.jpg |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#high-entropy--forking-tokens |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/high-entropy-forking-tokens-rl.jpg |
| 青稞社区转载全文 | https://qingkeai.online/archives/D9j6JtwN |
| 知乎原文 | https://zhuanlan.zhihu.com/p/1976336943878005134 |
| Reinforcement Fine-Tuning Naturally Mitigates Forgetting in Continual Post-Training | https://arxiv.org/abs/2507.04218 |
| SFT Memorizes, RL Generalizes | https://arxiv.org/abs/2501.17161 |
| The Path Not Taken | https://arxiv.org/abs/2510.19222 |
| Beyond the 80/20 Rule | https://arxiv.org/abs/2510.23840 |
| Retaining by Doing | https://arxiv.org/abs/2509.19065 |
| RL's Razor | https://arxiv.org/abs/2509.04259 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#bias-variance-bootstrapping |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#td-error-accumulation-td-误差累积 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#算法谱系 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#ppo |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20250504001256092.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#lmathrmcliptheta |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#r_ttheta |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#ppo-epoch为什么-ppo-是-near-on-policy |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#deepseek-r1 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20250504010037799.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20250504010054840.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#grpo-group-relative-policy-optimization--deepseekmath |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20250504001047002.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20250502002245050.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#j_mathrmgrpotheta |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#mathbbeleftfrac1gsum_i1gfrac1o_isum_t1o_iminleftr_itthetahataimathrmcliprittheta-1-epsilon-1epsilonhata_iright |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#r_ittheta |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#hata_i |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20250504001745451.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20250504002517628.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#why-rl-work |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#llm-post-trainingrlhf--rft--reasoning-rl |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#rlhf--基于人类反馈的强化学习 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#intro |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20251002031823547.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20251002031905100.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#历史发展 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20251002032014395.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20251002032114981.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20251002032323673.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20251002032405417.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#rlhf-步骤 |
| InstructGPT | https://openai.com/blog/instruction-following/ |
| https://huggingface.co/blog/rlhf | https://huggingface.co/blog/rlhf |
| Proximal Policy Optimization | https://openai.com/blog/openai-baselines-ppo/ |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/ChatGPT_Diagram.svg |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#instructgpt--介绍rlhf的数据工程 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20251002032216911.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#ultrafeedback |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20251002032525480.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#推理模型与-rlopenai-o1--cot |
| 如何理解OpenAI o1 | https://mp.weixin.qq.com/s/QdVSq8q7wLWtPakdZdqidA |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#cot |
| OpenAI研究员、思维树作者姚顺雨专访：人生是一场无限流游戏丨独家 | https://mp.weixin.qq.com/s/MdPI-X1HvRxFuX_Z0Ju_ug |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#rl-工程与系统 |
| http://xhslink.com/o/2MvDgQlctwI | http://xhslink.com/o/2MvDgQlctwI |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#挑战训推一体 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20251007143316174.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#partial-rollout |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#intro-1 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20251007143441936.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#mooncake--rl |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning/image-20251005222409248.png |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#verl |
| https://arxiv.org/abs/2409.19256 | https://arxiv.org/abs/2409.19256 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#intro-2 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#应用形态 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#search-r1 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#zerosearch |
| https://alibaba-nlp.github.io/ZeroSearch/ | https://alibaba-nlp.github.io/ZeroSearch/ |
| https://github.com/Alibaba-nlp/ZeroSearch | https://github.com/Alibaba-nlp/ZeroSearch |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#agent-memory--context-routing |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#其它应用领域 |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#游戏 |
| AI挑战黑神话！死亡1000次，我训练的AI终于击败了首个BOSS【图灵计划10】 | https://www.bilibili.com/video/BV1qE421c7mU |
| 【DQN只狼实战教程】手把手带你实现用强化学习DQN打只狼里的boss（第一期） | https://www.bilibili.com/video/BV1by4y1n7pe |
| https://github.com/loopx-agent/CS-Notes/blob/master/Notes/Reinforcement-Learning.md#机器人 |
| 一群不懂ActorCritic强化学习和PPO的人强行读Humanoid-Gym代码四小时_哔哩哔哩_bilibili | https://b23.tv/25KisIK |
|
| https://github.com |
| Terms | https://docs.github.com/site-policy/github-terms/github-terms-of-service |
| Privacy | https://docs.github.com/site-policy/privacy-policies/github-privacy-statement |
| Security | https://github.com/security |
| Status | https://www.githubstatus.com/ |
| Community | https://github.community/ |
| Docs | https://docs.github.com/ |
| Contact | https://support.github.com?tags=dotcom-footer |
