4 pieces · Artificial Intelligence, Machine Learning & Computing Research · Est. 2007
Recursive autodiff degrades badly for high-order derivatives on noisy data. Mollifier layers replace it with convolution against analytic smoothing kernels.
By Institute for Joint Cognition & AI · 6 min · 7 August 2026
A technical comparison of grid search, random search, Bayesian optimization, and population-based training for tuning machine learning hyperparameters.
By Institute for Joint Cognition & AI · 12 min · 14 February 2024
How LoRA, QLoRA, and adapter modules cut fine-tuning cost by orders of magnitude, with the memory/quality trade-offs engineers actually hit in practice.
By Institute for Joint Cognition & AI · 12 min · 10 July 2023
Transformer architecture explained for ML practitioners: attention, self-attention, Q/K/V, multi-head attention, positional encoding, and why transformers replaced RNNs.
By Institute for Joint Cognition & AI · 10 min · 15 May 2023