Popular GitHub Repositories for Sparsity

PaddleSlim is an open-source library for deep model compression and architecture search.

A toolkit to optimize ML models for deployment for Keras and TensorFlow, including quantization and pruning.

Transformers-compatible library for applying various compression algorithms to LLMs for optimized deployment with vLLM

More readable and flexible yolov5 with more backbone(gcn, resnet, shufflenet, moblienet, efficientnet, hrnet, swin-transformer, etc) and (cbam，dcn and so on), and tensorrt

[NeurIPS'23] H2O: Heavy-Hitter Oracle for Efficient Generative Inference of Large Language Models.

An innovative library for efficient LLM inference via low-bit quantization

Reference ImageNet implementation of SelecSLS CNN architecture proposed in the SIGGRAPH 2020 paper "XNect: Real-time Multi-Person 3D Motion Capture with a Single RGB Camera". The repository also includes code for pruning the model based on implicit sparsity emerging from adaptive gradient descent methods, as detailed in the CVPR 2019 paper "On implicit filter level sparsity in Convolutional Neural Networks".

Sparse and structured neural attention mechanisms

Zero-label image classification via OpenCLIP knowledge distillation

Soft Threshold Weight Reparameterization for Learnable Sparsity

?LLaMA-MoE v2: Exploring Sparsity of LLaMA from Perspective of Mixture-of-Experts with Post-Training

E2E AutoML Model Compression Package

Accelerate Multimodal AI | 3.7x faster E2E video gen and 1.6x faster E2E image gen | Custom sparse CUDA kernels 9.3x faster than FlashAttention-3 and 2.5x faster than cuBLAS

Attention-Based Guided Structured Sparsity of Deep Neural Networks

Official PyTorch (Lightning) implementation of the NeurIPS 2020 paper "Efficient Marginalization of Discrete and Structured Latent Variables via Sparsity".

Image Denoising Codes using STROLLR learning, the Matlab implementation of the paper in ICASSP2017

WIP. Veloce is a low-code Ray-based parallelization library that makes machine learning computation novel, efficient, and heterogeneous.

model compression and optimization for deployment for Pytorch, including knowledge distillation, quantization and pruning.(知识蒸馏，量化，剪枝)

Robust Single Sample Face Recognition by Sparsity-Driven Sub-Dictionary Learning Using Deep Features

Variational Interpretable Concept Embeddings