#pytorch Repositories
1,936 repositories total
A PyTorch impl of EfficientDet faithful to the original Google impl w/ ported weights
[CVPR 2025] MatAnyone: Stable Video Matting with Consistent Memory Propagation
Repository for training models for music source separation.
A coding-free framework built on PyTorch for reproducible deep learning studies. PyTorch Ecosystem. ๐ 28 knowledge distillation methods presented at TPAMI, CVPR, ICLR, ECCV, NeurIPS, ICCV, AAAI, etc. have been implemented so far. ๐ Trained models, training logs, and configurations are available to ensure reproducibility and benchmarking.
Unofficial Parallel WaveGAN (+ MelGAN & Multi-band MelGAN & HiFi-GAN & StyleMelGAN) with Pytorch
Autoresearch for GPU kernels. Give it any PyTorch model, go to sleep, wake up to optimized Triton kernels.
Official code for "FeatUp: A Model-Agnostic Frameworkfor Features at Any Resolution" ICLR 2024
[TPAMI 2025] ViewCrafter: Taming Video Diffusion Models for High-fidelity Novel View Synthesis
Fun-ASR speech recognition models, with native Hugging Face Transformers support for Fun-ASR-Nano and separate FunASR, vLLM and llama.cpp deployment paths.
FB (Facebook) + GEMM (General Matrix-Matrix Multiplication) - https://code.fb.com/ml-applications/fbgemm/
[CVPR 2025 Highlight] DepthCrafter: Generating Consistent Long Depth Sequences for Open-world Videos
[SIGGRAPH Asia 2023 (Technical Communications)] EasyVolcap: Accelerating Neural Volumetric Video Research
Official code of "HybrIK: A Hybrid Analytical-Neural Inverse Kinematics Solution for 3D Human Pose and Shape Estimation", CVPR 2021
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.
Pretrained EfficientNet, EfficientNet-Lite, MixNet, MobileNetV3 / V2, MNASNet A1 and B1, FBNet, Single-Path NAS
solo-learn: a library of self-supervised methods for visual representation learning powered by Pytorch Lightning
an implementation of 3D Ken Burns Effect from a Single Image using PyTorch
[ICML 2023] SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models
[CVPR 2025] Fast3R: Towards 3D Reconstruction of 1000+ Images in One Forward Pass
SoulX-FlashTalk is the first 14B model to achieve sub-second start-up latency (0.87s) while maintaining a real-time throughput of 32 FPS on an 8xH800 node.
A powerful and flexible machine learning platform for drug discovery
Bayesian Convolutional Neural Network with Variational Inference based on Bayes by Backprop in PyTorch.
A PyTorch library entirely dedicated to neural differential equations, implicit models and related numerical methods
The most comprehensive plug-and-play modules for the entire network in 2025, shared for free! CVPR2025, AAAI2025, ICLR2025, TNNLS2025, arXiv2025...... Covers all fields of artificial intelligence (machine learning, deep learning, etc.), suitable for image classification, object detection, instance segmentation, semantic segmentation, panoptic segmentation, pose estimation, medical image segmentation, video object segmentation, image matting, image editing, single object tracking, multiple object tracking, person re-identification, RGBT, image denoising, rain removal, fog removal, shadow removal, blur removal, super-resolution, reflection removal, moirรฉ pattern removal, image restoration, image inpainting, hyperspectral image restoration, image fusion, image colorization, high dynamic range imaging, video and image compression, 3D point clouds, 3D object detection, 3D semantic segmentation, 3D pose estimation, and various other computer vision and image processing tasks, as well as natural language processing, large language models, multimodal, and other AI-related tasks. Continuously updating......
a fast and user-friendly runtime for transformer inference (Bert, Albert, GPT2, Decoders, etc) on CPU and GPU.
Hibiki is a model for streaming speech translation (also known as simultaneous translation). Unlike offline translationโwhere one waits for the end of the source utterance to start translating--- Hibiki adapts its flow to accumulate just enough context to produce a correct translation in real-time, chunk by chunk.