The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

A theoretical reconstruction of the Claude Mythos architecture, built from first principles using the available research literature.

A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.

An Asynchronous Reinforcement Learning Engine for Omni-Modal Post-Training at Scale

The agent that grows with you

Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and Gemini CLI: local deterministic AST parsing…

GLM-Image: Auto-regressive for Dense-knowledge and High-fidelity Image Generation.

[ICML 2026] GenExam: A Multidisciplinary Text-to-Image Exam

Qwen-Image-Lightning: Speed up Qwen-Image model with distillation

✨ [ICLR'26] WithAnyone is capable of generating high-quality, controllable, and ID consistent images

[ICLR 2026 Oral] DiffusionNFT: Online Diffusion Reinforcement with Forward Process

OmniInsert: Mask-Free Video Insertion of Any Reference via Diffusion Transformer Models

VeOmni: Scaling Any Modality Model Training with Model-Centric Distributed Recipe Zoo

Official code for VMix: Improving Text-to-Image Diffusion Model with Cross-Attention Mixing Control

[CVPR 2026] 🔥🔥 Official Repo of UMO: Scaling Multi-Identity Consistency for Image Customization via Matching Reward

[CVPR 2026] 🔥🔥 Official Repo of USO: Unified Style and Subject-Driven Generation via Disentangled and Reward Learning

Automatically crawl arXiv papers daily and summarize them using AI. Illustrating them using GitHub Pages.

An AI-driven daily arXiv paper crawler, analyzer, and organizer tool, focusing on AIGC

Code for MetaMorph Multimodal Understanding and Generation via Instruction Tuning

Minimalistic 4D-parallelism distributed training framework for education purpose

verl/HybridFlow: A Flexible and Efficient RL Post-Training Framework

[SIGGRAPH Asia 2025] DreamO: A Unified Framework for Image Customization

[ICCV 2025] 🔥🔥 UNO: A Universal Customization Method for Both Single and Multi-Subject Conditioning

[TMLR 2025🔥] A survey for the autoregressive models in vision.

Align Anything: Training All-modality Model with Feedback

这是一个从头训练大语言模型的项目，包括预训练、微调和直接偏好优化，模型拥有1B参数，支持中英文。

VideoGen-Eval: Agent-based System for Video Generation Evaluation

[NeurIPS 2024] Official code for PuLID: Pure and Lightning ID Customization via Contrastive Alignment