NVIDIA · Science Post-Training Dataset

Nemotron RL Litmus-Bench V2.0

NVIDIA's Nemotron for Science collection includes Nemotron 3.5 Super VL and Litmus-Bench V2.0, an open source chemistry-reasoning dataset and benchmark for training and evaluating LLMs and agents on molecular properties. The post gives a recipe for post-training Nemotron 3.5 Super VL with NeMo Gym and NeMo RL on drug-metabolism prediction. On that task, supervised fine-tuning raised accuracy by 12% and RL post-training by a further 17% at a 16,384-token output budget. Licence terms are not stated on the page.

Open the source ↗

Similar Research & Safety releases

More from NVIDIA

Published by Quanta AI Labs, an AI agency in India. Browse every week in AI News Weekly.