kfac 话题在 GitHub 上没有填写简介，这里按 Star 数汇总最相关的热门仓库。
PyTorch implementation of Advantage Actor Critic (A2C), Proximal Policy Optimization (PPO), Scalable trust-region method for deep reinforcement learning using Kronecker-factored approximation (ACKTR) and Generative Adversarial Imitation Learning (GAIL).
{KFAC,EKFAC,Diagonal,Implicit} Fisher Matrices and finite width NTKs in PyTorch
Pytorch implementation of KFAC and E-KFAC (Natural Gradient).
Distributed K-FAC preconditioner for PyTorch
Bayesian Low-Rank Adaptation for Large Language Models
Newton-Muon + Preconditioned Optimizers for MoE Training at scale, with out-of-the-box support for MuP and FSDP support for Muon, built on top of Megatron-LM and TransformerEngine.
Convolutions and more as einsum for PyTorch
Laplace HypoPINN: Physics-informed neural networks for travel-time inversion, hypocenter localization and local network-weight uncertainty.
Matrix-multiplication-only KFAC; Code for ICML 2023 paper on Simplifying Momentum-based Positive-definite Submanifold Optimization with Applications to Deep Learning
该项目在 GitHub 上没有填写描述
Deliver AI-summarized local and national news, NOAA alerts, and full stories privately over Meshtastic mesh networks without internet access.
unRL (AKA "unreal") is a set of libraries providing Reinforcement Learning algorithms implemented in PyTorch or Jax.
Curvature-normalised rank-r parameter-space steering of language models, with no dictionary. 5-8x the raw behaviour gradient at matched capability cost on Gemma 4 12B.