GitHub - Danau5tin/ai-trains-ai: RL-training an AI agent to RL-train AI agents.
RL-training an AI agent to RL-train AI agents. Contribute to Danau5tin/ai-trains-ai development by creating an account on GitHub.
Technical Wiki
Summary
A self-hosted, Wikipedia-style wiki documenting reinforcement learning post-training with LoRA on consumer-grade hardware (4x DGX Spark, 4x RTX PRO 6000). Runs locally with only Python 3, supports editing, talk pages, history, and diffs. Public read-only copy deploys via GitHub Pages.
Highlights
- Covers RL post-training concepts, algorithms, hardware specs, models, frameworks, and run-time estimates for home setups
- Zero-dependency local server (Python 3 only) with full wiki features: editing, talk pages, history, diffs, and a questions form
- Public read-only copy auto-deploys to GitHub Pages on every push; history backed by git, questions form opens prefilled GitHub issues
- Targets specific home hardware configurations: 4x DGX Spark and 4x RTX PRO 6000
- Seed-based page management: original articles in seed.wiki, hand-edited pages are never overwritten
auto-generated
Context
Audience
ML practitioners and hobbyists who want to run reinforcement learning post-training on consumer or prosumer hardware at home
DomainReinforcement Learning
Formatself-hosted wiki with public static site
Accessopen source (GitHub), free to run locally or view online
Source Overview
Wikipedia-style wiki documenting RL post-training with LoRA on home hardware (4x DGX Spark, 4x RTX PRO 6000), covering concepts, algorithms, hardware, models, frameworks, run-time estimates, and embedded videos and papers.
Runs locally with only Python 3, supporting page editing, talk pages, history, diffs, and a questions form. Public copy at 0xsero.github.io/home-rl-wiki rebuilds on every push via GitHub Actions; history and the questions form are backed by git and prefilled GitHub issues.
Topics
Related
LoRA (Low-Rank Adaptation)RLHF / RL post-trainingDGX SparkRTX PRO 6000PPO / DPO algorithmsGitHub Pages
