Reverse engineer anything with agents, from app behavior down to native binaries.

Optimal Qwantization Kit

FlashKDA: high-performance Kimi Delta Attention kernels

[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

TurboDiffusion: 100–200× Acceleration for Video Diffusion Models

AI-agent Skill for generating polished HTML slide decks: editorial magazine and Swiss layouts, image prompts, social covers, and a WebGL/low-power presentation runtime.

DeepGEMM: clean and efficient BLAS kernel library on GPU

FlashInfer: Kernel Library for LLM Serving

Claude Code 中文全面上手指南。基于 luongnv89/claude-howto 本土化重写，面向中国小白用户，保留命令与配置兼容性，并附学习路径与本地化校验护栏。

DeepSeek-V4 Lecture

分享AI Infra知识&代码练习：PyTorch、vLLM/SGLang、slime/vime框架入门⚡️、性能加速🚀、大模型基础🧠、AI软硬件🔧等

As the core component of the Huawei Ascend for PyTorch community, TorchNPU is a deep learning adapter plugin built by Ascend for PyTorch, enabling the PyTorch framework to directly invoke Ascend NP…

Fast and memory-efficient exact attention

TensorRT LLM provides users with an easy-to-use Python API to define Large Language Models (LLMs) and supports state-of-the-art optimizations to perform inference efficiently on NVIDIA GPUs. Tensor…

Repo for Qwen Image Finetune

Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.8, DeepSeek-V4, GLM-5.1, InternLM3, Llama4, ...) and 300+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, GLM5.3, Gemma4, Llava, P…

Train transformer language models with reinforcement learning.

🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning.

Enjoy the magic of Diffusion models!

Repo for Qwen Image Finetune

[NeurIPS 24 Spotlight] MaskLLM: Learnable Semi-structured Sparsity for Large Language Models

Officiel code for PATCH: Learnable Tile-level Hybrid Sparsity for LLMs

[ICML2025] SpargeAttention: A training-free sparse attention that accelerates any model inference.

🌟100+ 原创 LLM / RL 原理图📚，《大模型算法》作者巨献！💥（100+ LLM/RL Algorithm Maps ）

[ICML 2025] Official PyTorch implementation of "FlatQuant: Flatness Matters for LLM Quantization"

Fast and memory-efficient exact attention

LLM Finetuning with peft