areas: training

• Cramming: Training a Language Model on a Single GPU in One Day — 2022, to-read
• How to Train State-Of-The-Art Models Using TorchVision’s Latest Primitives — 2021, deep-read
• Very Deep Convolutional Networks for Large-Scale Image Recognition — 2015, skimmed
• Deep Residual Learning for Image Recognition — 2015, deep-read
• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift — 2015, deep-read