NVIDIA AI Updates: June 16, 2026

1. NVIDIA Adds LoRA Fine-Tuning to BioNeMo Recipes for Biological Foundation Models

NVIDIA. NVIDIA published guidance for parameter-efficient fine-tuning of large biological foundation models using Low-Rank Adaptation in its BioNeMo Recipes framework, training roughly 1 percent of model parameters while preserving competitive accuracy. The post demonstrates the approach with two case studies, protein secondary structure prediction using ESM2-3B and DNA splice-site classification using Evo2-1B. It integrates the NVIDIA Transformer Engine and scale-out strategies so billion-parameter models can be fine-tuned on a single workstation GPU. Source