LIGHTWEIGHT CHAMP: NVIDIA RELEASES SMALL LANGUAGE MODEL WITH STATE-OF-THE-ART ACCURACY

Summary from the TLDR AI newsletter

NVIDIA's Mistral-NeMo-Minitron 8B, a compressed version of the 12B model, achieves state-of-the-art accuracy while being small enough to run on RTX workstations.