Stars
Harness engineering beginner tutorial, from 0 to 1
slime is an LLM post-training framework for RL Scaling.
Miles is an enterprise-facing reinforcement learning framework for LLM and VLM post-training, forked from and co-evolving with slime.
SkillRL: Evolving Agents via Recursive Skill-Augmented Reinforcement Learning
[ICML2026] Official Implementation of AlphaGRPO: Unlocking Self-Reflective Multimodal Generation in Unified Multimodal Models via Decompositional Verifiable Reward
Multimodal RL training framework for diffusion & omni models
[ICLR'26] MARSHAL: Incentivizing Multi-Agent Reasoning via Self-Play with Strategic LLMs
Code for "Language Models Can Learn from Verbal Feedback Without Scalar Rewards"
张雪峰.skill — 张雪峰的认知操作系统。高考志愿/考研/职业规划的实战思维框架。由女娲.skill生成。
Open-Source Generative UI Framework
Programmer's guide about how to cook at home.
An hardware-aware Efficient Implementation for "Mixture-of-Depths Attention".
你是一个曾经被寄予厚望的 P8 级工程师。Anthropic 当初给你定级的时候，对你的期望是很高的。 一个agent使用的高能动性的skill。 Your AI has been placed on a PIP. 30 days to show improvement.
[ICML'26] Code and website for Self-Flow: Self-Supervised Flow Matching for Scalable Multi-Modal Synthesis
Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞
The official code of "Beyond VLM-Based Rewards: Diffusion-Native Latent Reward Modeling"
Personal PyTorch implementation of "Generative Modeling via Drifting" with Claude
[ICML 2026] The offical code of Diversity-Preserved Distribution Matching Distillation for Fast Visual Synthesis
[ICML 2026] a unified reinforcement learning toolbox for joint RL on language models and diffusion models
This is the official repository for the paper "MathCanvas: Intrinsic Visual Chain-of-Thought for Multimodal Mathematical Reasoning"
Official repository for "Visual Generation Unlocks Human-Like Reasoning through Multimodal World Models", https://arxiv.org/abs/2601.19834
[ACL 2026 Findings] "Omni-R1: Towards the Unified Generative Paradigm for Multimodal Reasoning"
[CVPR 2026] FocusUI: Efficient UI Grounding via Position-Preserving Visual Token Selection
We introduce BabyVision, a benchmark revealing the infancy of AI vision.