Model reference · open weights

neurovfm-encoder

neurovfm-encoder is an open-weight embedding model from mlinslab. neurovfm-encoder (BF16) weighs 287 MB; the smallest configuration that runs it is RTX 3060 12 GB.

For self-hosting, compare this build's requirements below with the GPU you plan to rent.

• Neurovfm-encoder is a 3D Vision Transformer visual foundation model developed by mlinslab for image-feature-extraction in clinical neuroimaging.
• It was pretrained on 5.24 million uncurated MRI and CT volumes using the Volumetric Joint-Embedding Predictive Architecture.
• The model weights are distributed under the CC-BY-NC-SA 4.0 license.

Summary of the mlinslab/neurovfm-encoder model card, 2026-10-01

Hugging Face ↗