reinfrocement-learning 开源项目推荐（66 个）

WindyLab /LLM-RL-Papers

追踪 arXiv 上 LLM 与 RL 在控制领域的最新交叉研究。如有优秀论文，欢迎提交 PR。

BarisYazici /deep-rl-grasping

训练用于机器人抓取的深度强化学习模型。可从原始深度、RGBD 和自编码器等不同感知层中选择。在不同场景和对象数据集上测试学习到的模型

yyzpiero /EVO-PopulationBasedTraining

使用消息传递接口（MPI）进行强化学习的基于种群的训练（PBT）

silencial /DeepRL

Berkeley CS285 2019 作业解答

anas-zafar /agi-cognitive-foundations

综述论文《Thinking Beyond Tokens: From Brain-Inspired Intelligence to Cognitive Foundations for Artificial General Intelligence and its Societal Impact》的官方 GitHub 页面

DistRL-lab /distrl-open

DistRL：面向设备端控制智能体的异步分布式强化学习框架

HKUST-Aerial-Robotics /FC-Family

A Series of Research on Advancing Aerial Robots Beyond Mobility

Vindulamj /eco-drive

[ECC 2022] Codebase for the paper titled "Learning Eco-Driving Strategies at Signalized Intersections".

chagmgang /pysc2_rl

该项目在 GitHub 上没有填写描述

PiperLiu /five-in-a-row-AI

♟️ 部署AI五连胜游戏。包括前端、后端和深度强化学习代码。基于 vue3 与 flask 部署的强化学习五子棋 AlphaGo 实践。

kyegomez /HindsightReplay

My implementation of Hindsight replay in PyTorch: "Hindsight Experience Replay"

F-a-b-r-i-z-i-o /maze-game

Q-learning project where an agent learns by himself to find the exit inside a maze. The project is implemented as a level-based game.

godelcomplete9 /attention-grammar

Symbolic primitives for reasoning parsing. If language is too expensive for your SOTA tiny model, use meta tags.

naifmeh /smartbotjs

JS smart crawler using reinforcement learning

zyxue /rljs

RLjs currently serves as an interactive playground for learning reinforcement learning.

SyedAanif /AI-ML-Python

This repo contains learnings about Artificial Intelligence and Machine Learning using Python Jupyter Notebooks downloaded as HTML

okefemi12 /ecommerce-recommender-system

Production-grade Hybrid Recommender System combining Sequential Transformers (SASRec) and Offline RL (DQN) with Sim-to-Real training. Features sub-50ms inference via TFLite/FastAPI, Dockerized deployment, and CI/CD pipelines

Anshumaan-Chauhan02 /DQNAS

Neural Architecture Search for Convolutional Neural Networks using Reinforcement Learning

coco11563 /Traceable_Automatic_Feature_Transformation_via_Cascading_Actor-Critic_Agents

Code for <Traceable Automatic Feature Transformation via Cascading Actor-Critic Agents>

unitn-sml /recourse-fare

(Explainable) Algorithmic Recourse with Reinforcement Learning and MCTS (FARE and E-FARE)