Issues

• bf16 params: fp32 optimizer state is cast to bf16 on load_state_dict and step() fails

  NVIDIA-NeMo/Emerging-Optimizers#316 · Filter by author v0ropaev opened on Oct 20, 2026
  6

• [Optimizer Proposal] NEAT: conflict-aware optimization via directional gradient correction

  enhancement
  NVIDIA-NeMo/Emerging-Optimizers#295 · Filter by author ItCodinTime opened on Sep 5, 2026

• Docs: note where library defaults diverge from the arXiv 2607.20548 recipe (momentum, weight decay, NS iterations, RMS scale)

  NVIDIA-NeMo/Emerging-Optimizers#259 · Filter by author adpena opened on Jul 24, 2026
  1

• Locoprop-S and Newton-Muon support via TE, Megatron-LM changes.

  enhancement
  NVIDIA-NeMo/Emerging-Optimizers#215 · Filter by author plugyawn opened on May 30, 2026
  4

• Adding symmetry-compatible / equivariant optimizers for general matrix-valued parameters

  enhancement
  NVIDIA-NeMo/Emerging-Optimizers#188 · Filter by author timlautk opened on May 20, 2026
  5