juiceb0xc0de·Jul 7CartoLLM: Mid-Training Atlas Across Three Pretraining Checkpoints — Informed Weight GainsA small open-source team handed me a model while it was still training and asked me to read the inside of it.
Ed Plowman·Feb 16The Golden Ticket Theory: Why Distillation Is a Shortcut to a Dead End…Finding a golden ticket was just the first obstacle, while that means chololate today, it won't win you Wonka's knowledge & factory...
InTechKoala InsightsbyNithin Bharadwaj·Jan 6Mastering PyTorch Hooks: Debug, Monitor, and Control Your Model Like a Pro
Kartekeya Sharma·Jun 18, 2025Pruning ResNet at Initialization: My Journey into Sparse Training on CIFAR-10Pruning ResNet at Initialization: My Journey into Sparse Training on CIFAR-10
Mehdi Yazdani·May 18, 2025Why Regularization Isn’t Enough: A Better Way to Train Neural Networks with Two ObjectivesWhen training neural networks, we often juggle two competing objectives. for example, maximizing predictive performance while also meeting…A response icon1
