Open-source models
Alongside the Rocky Mountain Snowpack dataset, we publish open-source models trained on it. They're research artifacts for the snow-science and machine-learning communities — explore them, break them, build on them.
snowGAN
Our generative family over snowpack imagery, published on Hugging Face under the RMDig organization in two variants, one per imaging modality:
- snowGAN-magnified-profile — image-to-image model over magnified snowpack profile imagery (stratigraphy from pit walls).
- snowGAN-core — image-to-image model over extracted snow core cross-sections.
Learning to generate realistic snowpack imagery forces the models to internalize the structure of real snowpack — layering, grain texture, density transitions — which is what makes the generative work a stepping stone for the analysis models that follow.
snowGradient — experimental
Current model experiments live in the open snowGradient repository — the testing ground for candidate architectures in our avalanche-risk research line (AvAI — Avalanche AI). This work is research in progress: it is not part of the AvAI app today, and when it first ships it will be clearly labeled as a research preview.
License & contact
See each model card for license and usage details; the underlying dataset is CC-BY-4.0. Questions or collaboration: support@rmdig.ai.