Miroier

Miroier's Stars

• thu-ml/SageAttention

  Quantized Attention that achieves speedups of 2.1-3.1x and 2.7-5.1x compared to FlashAttention2 and xformers, respectively, without lossing end-to-end metrics across various models.

  Language:Cuda67233

• tlc-pack/libflash_attn

  Standalone Flash Attention v2 kernel without libtorch dependency

  Language:C++9913

• google/bloaty

  Bloaty: a size profiler for binaries

  Language:C++4.8k348

• srush/Triton-Puzzles

  Puzzles for learning Triton

  Language:Jupyter Notebook1.2k92

• ZonePG/cs-notes

  my cs notes

  Language:Jupyter Notebook272

• tgale96/grouped_gemm

  PyTorch bindings for CUTLASS grouped GEMM.

  Language:Cuda5740

• bytedance/ABQ-LLM

  An acceleration library that supports arbitrary bit-width combinatorial quantization operations

  Language:C++22925

• FlagOpen/FlagGems

  FlagGems is an operator library for large language models implemented in Triton Language.

  Language:Python35852

• bytedance/flux

  A fast communication-overlapping library for tensor parallelism on GPUs.

  Language:C++23719

• kendryte/nncase

  Open deep learning compiler stack for Kendryte AI accelerators ✨

  Language:C#756185

• feifeibear/LLMSpeculativeSampling

  Fast inference from large lauguage models via speculative decoding

  Language:Python60063

• HazyResearch/ThunderKittens

  Tile primitives for speedy kernels

  Language:Cuda1.7k78

• KnowingNothing/compiler-and-arch

  A list of tutorials, paper, talks, and open-source projects for emerging compiler and architecture

  40135

• TiledTensor/TiledCUDA

  TiledCUDA is a highly efficient kernel template library designed to elevate CUDA C’s level of abstraction for processing tiles.

  Language:C++16710

• Sergei-Korneev/obsidian-local-images-plus

  This repo is a reincarnation of obsidian-local-images plugin which main aim was downloading images in md notes to local storage.

  Language:TypeScript27021

• FoundationVision/VAR

  [NeurIPS 2024 Oral][GPT beats diffusion🔥] [scaling laws in visual generation📈] Official impl. of "Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction". An *ultra-simple, user-friendly yet state-of-the-art* codebase for autoregressive image generation!

  Language:Python6.1k407

• xdit-project/xDiT

  xDiT: A Scalable Inference Engine for Diffusion Transformers (DiTs) with Massive Parallelism

  Language:Python91679

• FlagOpen/FlagScale

  FlagScale is a large model toolkit based on open-sourced projects.

  Language:Python19449

• online-judge-tools/verification-helper

  a testing framework for snippet libraries used in competitive programming

  Language:Python23356

• wting/autojump

  A cd command that learns - easily navigate directories from the command line

  Language:Python16.3k705

• alangrainger/share-note

  Instantly share an Obsidian note with the full theme exactly like you see in your vault. Data is shared encrypted by default, and only you and the person you send it to have the key.

  Language:TypeScript33717

• pengsida/learning_research

  本人的科研经验

  6.1k358

• Jack47/hack-SysML

  The road to hack SysML and become an system expert

  Language:Emacs Lisp44754

• weishengying/cutlass_flash_atten_fp8

  使用 cutlass 仓库在 ada 架构上实现 fp8 的 flash attention

  Language:Cuda533

• CaiJimmy/hugo-theme-stack

  Card-style Hugo theme designed for bloggers

  Language:HTML5.1k1.7k

• ELS-RD/kernl

  Kernl lets you run PyTorch transformer models several times faster on GPU with a single line of code, and is designed to be easily hackable.

  Language:Jupyter Notebook1.5k94

• ROCm/rocWMMA

  rocWMMA

  Language:C++9426

• AnswerDotAI/gpu.cpp

  A lightweight library for portable low-level GPU computation using WebGPU.

  Language:C++3.8k177

• te42kyfo/gpu-benches

  collection of benchmarks to measure basic GPU capabilities

  Language:Jupyter Notebook26441

• sjfeng1999/gpu-arch-microbenchmark

  Dissecting NVIDIA GPU Architecture

  Language:Cuda8225