reinforcement-learning

unslothai/unsloth

Unsloth ships a desktop app to run and fine-tune LLMs locally — Unsloth now offers a native desktop app for Windows, macOS and Linux that lets users run and train LLMs and diffusion models locally, supporting GGUF, MLX, Qwen3.8, DeepSeek-V4, MiniMax-H3, Gemma 4 and FLUX. It targets self-hosted, privacy-conscious users who want fine-tuning and inference without cloud dependency.

rohitg00/ai-engineering-from-scratch

A 542-lesson curriculum for building AI systems from the ground up — A free, MIT-licensed course with 542 lessons across 20 phases covering model internals, retrieval pipelines, and agent runtimes in Python, TypeScript, Rust, and Julia. Learners write code, run evals, and inspect failures, with labs and project challenges for self-paced or agent-assisted study.

labmlai/annotated_deep_learning_paper_implementations

A literate-programming library of 60+ annotated deep learning papers — This repo pairs PyTorch code with side-by-side explanatory notes for transformers, GANs, RL algorithms, optimizers and more, rendered on a companion website. It's actively updated weekly and now points to a separate 'Annotated Implementation Skill' for using coding agents to implement new papers in the same style. Useful for learners and engineers wanting to read algorithms as documented code rather than just papers.

bojieli/ai-agent-book

An open-source book on AI Agent design with 109 runnable labs — A free, open 10-chapter book on AI Agent engineering, framed as LLM + context + tools, by Li Bojie. It ships 109 hands-on experiments, PDF/EPUB downloads, an online reader, and translations into 15 languages, aimed at engineers building real agent systems.

ray-project/ray

Ray scales Python and ML workloads from laptop to cluster — Ray pairs a distributed runtime (tasks, actors, objects) with AI libraries for data, training, tuning, RL and serving. It runs anywhere from a laptop to Kubernetes clusters, and now includes LLM serving/inference tooling plus a distributed debugger and dashboard, making it relevant for teams scaling both classic ML and large language model workloads.

sgl-project/sglang

Unified inference engine for LLMs, VLMs, and diffusion models — SGLang now bundles SGLang Diffusion, a built-in image/video generation engine, alongside its existing LLM and vision-language serving stack. It targets agentic workloads, large-scale serving, and RL rollouts, with broad hardware and model support including Blackwell GPUs, MoE architectures, and models like DeepSeek, Qwen, and GLM.

microsoft/agent-lightning

A lightweight RL framework to train AI agents using real harnesses — Agent Lightning v1.0 is a ~3,500-line framework that trains agents via reinforcement learning without modifying existing agent harnesses, tools, or environments, using a proxy layer and native Kubernetes job support. It targets MLOps and RL researchers, with examples showing SWE-bench Verified gains from fine-tuning Qwen models on limited training samples.

owainlewis/awesome-artificial-intelligence

Curated list refocused on building production generative AI systems — This long-running awesome list has shifted focus to developers building, evaluating, and deploying generative AI and agentic systems, not just classic ML theory. Entries face a strict quality bar and the list is reviewed weekly by automation, making it a practical resource for engineers working with LLMs, retrieval, and coding agents.

snap-stanford/memorilla

Compressing Entire Document Collections into 16 Memory Tokens — Memorilla is a Stanford research system that condenses an entire document collection into a fixed set of query-dependent memory tokens for a frozen LLM. With just 16 tokens, it beats Mem0 and MemoRAG on average and even outperforms feeding every document on FactKG (84.33% vs 80.43%) using 108x fewer tokens, while also boosting an agent's TextWorld success rate from 18.6% to 24.2%.

joeynyc/awesome-microduck

Tracking the Pre-Launch Ecosystem of Pollen's Open-Source Duck Robot — A curated list for Microduck, the Apache-2.0 biped duck robot from Pollen Robotics and Hugging Face whose walking, kicking and recovery behaviors are MuJoCo-trained ONNX policies. Since hardware ships only around December 2026, this list focuses on what's usable now: simulators, training tools, MCP/agent integrations and community coverage, for roboticists and RL researchers tracking the project ahead of release.