- [R] ResNet strikes back: An improved training procedure in timm. There has been significant progress on best practices for training neural nets since ResNet's introduction in 2015. With such advances, a vanilla ResNet-50 reaches 80.4% top-1 accuracy on ImageNet without extra data or distillation. https://arxiv.org/abs/2110.00476 45 comments machinelearning
Linking pages
- How to Train State-Of-The-Art Models Using TorchVision’s Latest Primitives | PyTorch https://pytorch.org/blog/how-to-train-state-of-the-art-models-using-torchvision-latest-primitives/ 3 comments
- GitHub - 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 https://github.com/rwightman/pytorch-image-models 2 comments
- Making Machine Learning More Accessible for Application Developers | Frank’s Ramblings https://frankzliu.com/blog/making-machine-learning-more-accessible-for-application-developers 0 comments
- Learning on tree architectures outperforms a convolutional feedforward network | Scientific Reports https://www.nature.com/articles/s41598-023-27986-6 0 comments
Would you like to stay up to date with Computer science? Checkout Computer science
Weekly.
Related searches:
Search whole site: site:arxiv.org
Search title: [2110.00476] ResNet strikes back: An improved training procedure in timm
See how to search.