Skip to content

@THUDM THUKEG

Change the repository type filter

All

    Repositories list

    130 repositories

    • slime

      Public
      slime is an LLM post-training framework for RL Scaling.
      Python
      •
      Apache License 2.0
      •1.2k•8.1k•213•213•Updated Aug 16, 2026Aug 16, 2026

    • SCALE-CUA

      Public
      Open-source framework for computer use agents: VeriGen verifiable task synthesis, online RL training (AgentRL), and OSWorld/ScienceBoard evaluation.
      reinforcement-learningllm-agentscomputer-use-agents
      reinforcement-learningllm-agentscomputer-use-agents
      Python
      •2•50•3•0•Updated Aug 3, 2026Aug 3, 2026

    • KARL

      Public
      Python
      •1•3•0•0•Updated Jul 7, 2026Jul 7, 2026

    • CodeRM-NT

      Public
      [Findings of ACL 2026] CodeRM-NT: Reward Model for Code RL without Unit Tests
      Python
      •0•0•0•0•Updated Jul 1, 2026Jul 1, 2026

    • DeepDive

      Public
      DeepDive: Advancing Deep Search Agents with Knowledge Graphs and Multi-Turn RL
      Python
      •38•341•0•0•Updated Jun 17, 2026Jun 17, 2026

    • INFTY

      Public
      INFTY Engine: An Optimization Toolkit to Support Continual AI
      Python
      •
      MIT License
      •11•574•6•0•Updated Jun 8, 2026Jun 8, 2026

    • ReST-RL

      Public
      Reinforcing LLM Reasoning through Self-Training and Value-Guided Decoding
      Python
      •
      MIT License
      •0•18•0•0•Updated May 6, 2026May 6, 2026

    • CaRR

      Public
      This repository contains the code and data for the paper "Chaining the Evidence: Robust Reinforcement Learning for Deep Search Agents with Citation-Aware Rubric…
      Python
      •
      MIT License
      •9•73•0•0•Updated Apr 8, 2026Apr 8, 2026

    • IndexCache

      Public
      IndexCache: Accelerating Sparse Attention via Cross-Layer Index Reuse
      glmsparse-attentionllm
      glmsparse-attentionllmlong-contextdeepseek
      Other
      •11•132•7•1•Updated Mar 14, 2026Mar 14, 2026

    • AgentBench

      Public
      A Comprehensive Benchmark to Evaluate LLMs as Agents (ICLR'24)
      gpt-4llmchatgpt
      gpt-4llmchatgptllm-agent
      Python
      •
      Apache License 2.0
      •274•3.7k•63•12•Updated Feb 8, 2026Feb 8, 2026

    • DataSciBench

      Public
      DataSciBench: An LLM Agent Benchmark for Data Science (Findings of ACL 2026)
      Python
      •8•66•1•0•Updated Jan 21, 2026Jan 21, 2026

    • AgentRL

      Public
      Scaling Agentic Reinforcement Learning with a Multi-Turn, Multi-Task Framework
      Python
      •
      MIT License
      •27•342•9•0•Updated Jan 17, 2026Jan 17, 2026

    • MobileRL

      Public
      Python
      •
      MIT License
      •9•97•3•0•Updated Dec 23, 2025Dec 23, 2025

    • ComputerRL

      Public
      Python
      •
      Apache License 2.0
      •8•43•4•0•Updated Nov 7, 2025Nov 7, 2025

    • PETra

      Public
      Python
      •0•2•0•0•Updated Nov 5, 2025Nov 5, 2025

    • AlignBench

      Public
      大模型多维度中文对齐评测基准 (ACL 2024)
      large-language-modelsllmchatgpt
      large-language-modelsllmchatgptchatglm
      Python
      •30•431•15•0•Updated Oct 25, 2025Oct 25, 2025

    • LLM4CardGame

      Public
      Python
      •2•17•2•0•Updated Oct 15, 2025Oct 15, 2025

    • TDRM

      Public
      Python
      •
      Apache License 2.0
      •1•9•0•0•Updated Sep 25, 2025Sep 25, 2025

    • Android-Lab

      Public
      Python
      •
      MIT License
      •26•323•2•0•Updated Aug 18, 2025Aug 18, 2025

    • SWE-Dev

      Public
      [ACL25' Findings] SWE-Dev is an SWE agent with a scalable test case construction pipeline.
      Python
      •
      MIT License
      •0•65•1•0•Updated Jul 21, 2025Jul 21, 2025

    • z-ai-sdk-typescript

      Public
      Typescript SDK for Z.ai - Not yet released.
      TypeScript
      •
      MIT License
      •2•9•1•0•Updated Jul 17, 2025Jul 17, 2025

    • BiPro

      Public
      code and data for Paper: BIPro: Zero-shot Chinese Poem Generation via Block Inverse Prompting Constrained Generation Framework(ACL 2025 main)
      Python
      •0•6•0•0•Updated Jun 28, 2025Jun 28, 2025

    • LongWriter

      Public
      [ICLR 2025] LongWriter: Unleashing 10,000+ Word Generation from Long Context LLMs
      fine-tuningllmlong-context
      fine-tuningllmlong-contextlong-text
      Python
      •
      Apache License 2.0
      •182•1.9k•30•2•Updated Jun 24, 2025Jun 24, 2025

    • TreeRL

      Public
      TreeRL: LLM Reinforcement Learning with On-Policy Tree Search in ACL'25
      Python
      •
      Apache License 2.0
      •12•104•4•0•Updated Jun 16, 2025Jun 16, 2025

    • WebRL

      Public
      Building Open LLM Web Agents with Self-Evolving Online Curriculum RL
      Python
      •38•539•0•0•Updated Jun 6, 2025Jun 6, 2025

    • AndroidGen

      Public
      Python
      •
      Apache License 2.0
      •1•14•1•0•Updated May 29, 2025May 29, 2025

    • AlignMMBench

      Public
      code, data and model for Paper: AlignMMBench: Evaluating Chinese Multimodal Alignment in Large Vision-Language Models (ACL'25 main)
      Python
      •2•6•1•0•Updated May 20, 2025May 20, 2025

    • CogKit

      Public
      Finetuning and inference tools for the CogView4 and CogVideoX model series.
      finetunevideo-generationtext2
      finetunevideo-generationtext2cogvideox
      Python
      •
      Apache License 2.0
      •19•130•18•1•Updated May 14, 2025May 14, 2025

    • VisualAgentBench

      Public
      Towards Large Multimodal Models as Visual Foundation Agents
      gptllm-agentmultimodal-large-language-models
      gptllm-agentmultimodal-large-language-models
      Python
      •
      Apache License 2.0
      •13•274•16•1•Updated Apr 24, 2025Apr 24, 2025

    • MoELoRA_Riemannian

      Public
      Source code of paper: A Stronger Mixture of Low-Rank Experts for Fine-Tuning Foundation Models. (ICML 2025)
      Python
      •2•39•0•0•Updated Apr 2, 2025Apr 2, 2025
    ProTip! When viewing an organization's repositories, you can use the props. filter to filter by custom property.