This project aims to help engineers, researchers and students to easily find and learn the good thoughts and designs in AI-related fields, such as AI/ML/DL accelerators, chips, and systems, proposed in the top-tier architecture
A flow-based network is considered to be inefficient in parameter complexity because of reduced expressiveness of bijective mapping, which renders the models prohibitively expensive in terms of parameters. We present an alternativ
An implementation of the algorithm and experiments defined in "Ab-Initio Solution of the Many-Electron Schroedinger Equation with Deep Neural Networks", David Pfau, James S. Spencer, Alex G de G Matthews and W.M.C. Foulkes, Phys.
LA-MCTS is a new MCTS based derivative-free meta-solver. Compared to Bayesian optimization, and evolutionary algorithm, it learns to partition the search space, thereby 🌟 finding a better solution with fewer samples.