Focused on image, video, and audio deep learning models and training/deployment systems. I maintain several popular PyTorch (and recently JAX) model repositories.
The model collections are focused on providing pretrained weights for the latest model architectures. I adapt some weights from other organizations and deep learning frameworks but also conduct research and train many of them myself. The models I've been working recently require thousands of GPU-hours to train, tune, and verify.
Training, testing, and maintaining these codebases and model weights requires considerable GPU (or TPU) resources and significant electricity/cooling bills. I don't need funding to put bread on the table, but I do need additional resources to offer and train more models with better techniques. I'm currently working on semi/self-supervised training and transfer learning benchmarking.
If you are an organization who benefits from the models, code, or techniques I publish, please consider contributing so that I can build more.
15 sponsors are funding rwightman’s work.
Featured work
-
rwightman/pytorch-image-models
PyTorch image models, scripts, pretrained weights -- ResNet, ResNeXT, EfficientNet, EfficientNetV2, NFNet, Vision Transformer, MixNet, MobileNet-V3/V2, RegNet, DPN, CSPNet, and more
Python 13,334 -
rwightman/gen-efficientnet-pytorch
Pretrained EfficientNet, EfficientNet-Lite, MixNet, MobileNetV3 / V2, MNASNet A1 and B1, FBNet, Single-Path NAS
Python 1,391 -
rwightman/efficientdet-pytorch
A PyTorch impl of EfficientDet faithful to the original Google impl w/ ported weights
Python 1,163 -
rwightman/efficientnet-jax
EfficientNet, MobileNetV3, MobileNetV2, MixNet, etc in JAX w/ Flax Linen and Objax
Python 80