rose-1h-nmr
A pretrained model and adaptation protocol for 1H NMR spectrum embeddings.
SITE CAPTUREBased on information published on the service's website. Selection and review policy
WHAT IT DOES
What does it do?
ROSE is a pretrained 1H NMR foundation model that provides inference and a paper based adaptation protocol. It supports zero shot use of the pretrained weights and adaptation by tuning the encoder. The code and pretrained weights are available under Apache-2.0.
KEY FEATURES
Key features
- Run inference with a pretrained 1H NMR model
- Use weights in a zero shot setting
- Adapt the encoder and configure parameter groups
- Follow protocols for peak and search tasks