🚀 Text2Grad: Converting natural language feedback into gradient signals for precise model optimization. Revolutionizing RLHF with span-level rewards and targeted improvements across code generation…

A curated list of papers and resources on Reward Hacking, Emergent Misalignment, and Proxy Exploitation in Large Models

AI agents running research on single-GPU nanochat training automatically

SkillOpt is a text-space optimizer that trains reusable natural-language skills for frozen LLM agents through trajectory-driven edits, validation-gated updates, and deployable best_skill.md artifacts.

SkillsBench evaluates how well skills work and how effective agents are at using them.

A curated collection of papers and resources on On-Policy Distillation for Large Language Models.

The agent that grows with you

Train the smallest LM you can that fits in 16MB. Best model wins!

tLLM is an test-time training extension of vLLM

Optimize prompts, code, and more with AI-powered Reflective Optimization

[Survey] A Comprehensive Survey of Self-Evolving AI Agents: A New Paradigm Bridging Foundation Models and Lifelong Agentic Systems

你想蒸馏的下一个员工，何必是同事。蒸馏任何人的思维方式——心智模型、决策启发式、表达DNA。Distill how anyone thinks.

PPT Template of ShanghaiTech University. Include Powerpoint, Markdown Marp, LaTeX Beamer

提供多款 Shadowrocket 规则，拥有强劲的广告过滤功能。每日 8 时重新构建规则。

分流规则、重写写规则及脚本。

verl-agent is an extension of veRL, designed for training LLM/VLM agents via RL. verl-agent is also the official code for paper "Group-in-Group Policy Optimization for LLM Agent Training"

Official code for paper "TimeMaster: Training Time-Series Multimodal LLMs to Reason via Reinforcement Learning"

Search-R1: An Efficient, Scalable RL Training Framework for Reasoning & Search Engine Calling interleaved LLM based on veRL

SkillRL: Evolving Agents via Recursive Skill-Augmented Reinforcement Learning

🎃 A fast, out-of-the-box terminal built for AI coding.

原汁原昧 Claude Code 可运行,可构建, 可调试版; 生产级工程化, 企业级可靠性; 安全无毒, 内存泄露修复

Claude Code Snapshot for Research. All original source code is the property of Anthropic.

OpenClaw-RL: Train any agent simply by talking

A curated list of awesome open source libraries to deploy, monitor, version and scale your machine learning

Curated list of AutoResearch use cases with optimization traces and open source implementations

TimeCopilot: the GenAI Forecasting Agent. Built on LLMs and Time Series Foundation Models, it lets you forecast, cross-validate, and detect anomalies using multiple foundation models through a sing…