mozhu621

superpowers. An agentic skills framework & software development methodology that works.

ConStory-Bench. ConStory-Bench (ACL 26')

claude-code. Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.

CHAIN. Python

VisGym. Official Repository of VisGym: Diverse, Customizable, Scalable Environments for Multimodal Agents

PEAR. [ICLR 2026]Open repository for paper "PEAR: Phase Entropy Aware Reward for Efficient Reasoning"

CL-bench. CL-bench: A Benchmark for Context Learning

gemini-cli. An open-source AI agent that brings the power of Gemini directly into your terminal.

Kimi-Linear.

InfoQA. Python

ml-agents. The Unity Machine Learning Agents Toolkit (ML-Agents) is an open-source project that enables games and simulations to serve as environments for training intelligent agents using deep reinforcement learning and imitation learning.

jamify. JAM: A Tiny Flow-based Song Generator with Fine-grained Controllability and Aesthetic Alignment

GLM-4.5. GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models

lca-baselines. Baselines for all tasks from Long Code Arena benchmarks 🏟️

slime. slime is an LLM post-training framework for RL Scaling.

VisualReasonBench. https://docs.google.com/presentation/d/13wSxfrFKtroNnvdFSS0EKtA56oYau_soGPTUynzVstI

RLVR-Decomposed. [NeurIPS 2025] Implementation for the paper "The Surprising Effectiveness of Negative Reinforcement in LLM Reasoning"

creative-writing-bench. HTML

SuperWriter. Python

Megatron-LM. Ongoing research training transformer models at scale

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

ReasoningPathCompression. [NeurIPS 2025] Official implementation of "Reasoning Path Compression: Compressing Generation Trajectories for Efficient LLM Reasoning"

ParScale. Parallel Scaling Law for Language Model — Beyond Parameter and Inference Time Scaling

Qwen3. Qwen3 is the large language model series developed by Qwen team, Alibaba Cloud.

ring-flash-attention. Ring attention implementation with flash attention

Skywork-OR1. Unleashing the Power of Reinforcement Learning for Math and Code Reasoners

WritingBench. WritingBench: A Comprehensive Benchmark for Generative Writing

kg-gen. [NeurIPS '25] Knowledge Graph Generation from Any Text

TangoFlux. [ICLR 2026] TangoFlux: Super Fast and Faithful Text to Audio Generation with Flow Matching

LongWriter-V. [ACM MM25] LongWriter-V: Enabling Ultra-Long and High-Fidelity Generation in Vision-Language Models

Satori. [ICML 2025] Satori: Reinforcement Learning with Chain-of-Action-Thought Enhances LLM Reasoning via Autoregressive Search

NeMo-Aligner. Scalable toolkit for efficient model alignment

FastVideo. A unified inference and post-training framework for accelerated video generation.

PaperGallery. A curated gallery and toolkit designed to provide inspiration for scientific illustrations, project sites, and visual storytelling in research.

Awesome-LLM-Long-Context-Modeling. 📰 Must-read papers and blogs on LLM based Long Context Modeling 🔥

Kimi-k1.5.

suri. Suri: Multi-constraint instruction following for long-form text generation [EMNLP’24]

LongBench. LongBench v2 and LongBench (ACL 25'&24')

CONFACT. Python

O1-Journey. O1 Replication Journey

LongGenBench. Jupyter Notebook

g1. g1: Using Llama-3.1 70b on Groq to create o1-like reasoning chains

LongWriter. [ICLR 2025] LongWriter: Unleashing 10,000+ Word Generation from Long Context LLMs

ai-for-grant-writing. A curated list of resources for using LLMs to develop more competitive grant applications.

CogView. Text-to-Image generation. The repo for NeurIPS 2021 paper "CogView: Mastering Text-to-Image Generation via Transformers".

DPO-ST. [ACL 2024] Self-Training with Direct Preference Optimization Improves Chain-of-Thought Reasoning

EasyContext. Memory optimization and training recipes to extrapolate language models' context length to 1 million tokens, with minimal hardware.

HALOs. A library with extensible implementations of DPO, KTO, PPO, ORPO, and other human-aware loss functions (HALOs).

flexflow-train. Automatically Discovering Fast Parallelization Strategies for Distributed Deep Neural Network Training

SentiStream. Python

self-rag. This includes the original implementation of SELF-RAG: Learning to Retrieve, Generate and Critique through self-reflection by Akari Asai, Zeqiu Wu, Yizhong Wang, Avirup Sil, and Hannaneh Hajishirzi.

SentiStream.

in-context-ralm. Python

aclpubcheck. Tools for checking ACL paper submissions

ai-town. A MIT-licensed, deployable starter kit for building and customizing your own version of AI town - a virtual town where AI characters live, chat and socialize.

transformers. 🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.

MorphStream. This project aims at building a scalable transactional stream processing engine on modern hardware. It allows ACID transactions to be run directly on streaming data. It shares similar project vision with Flink

Awesome-Online-Continual-Learning.

ProPETL. One Network, Many Masks: Towards More Parameter-Efficient Transfer Learning

GPTCache. Semantic cache for LLMs. Fully integrated with LangChain and llama_index.

ChatPaper. Use ChatGPT to summarize the arXiv papers. 全流程加速科研，利用chatgpt进行论文全文总结+专业翻译+润色+审稿+审稿回复

tuning_playbook. A playbook for systematically maximizing the performance of deep learning models.

bert4keras. keras implement of transformers for humans

accelerate. 🚀 A simple way to launch, train, and use PyTorch models on almost any device and distributed configuration, automatic mixed precision (including fp8), and easy-to-configure FSDP and DeepSpeed support

BERT-whitening. 简单的向量白化改善句向量质量

chinese_sentiment. 用tensorflow进行中文自然语言处理的情感分析