Published July 2, 2026
| Version v1
Model
Open
Autoencoder and normalising flow checkpoints.
Authors/Creators
-
1.
Commissariat à l'Énergie Atomique et aux Énergies Alternatives
- 2. NSF-Simons AI Institute for the Sky
- 3. Departement of Astronomy and Astrophysics, University of Chicago
Description
The files in this repository correspond to the weights of the pretrained autoencoder and normalising flow used for the work "SHINE: Shear inference environment - Generative forward modelling for radio weak gravitational lensing". The asociate repository for this work is https://github.com/CentofantiEze/Radio_WL_Generative_Forward_Model.
The models are defined in the probabilistic-shear repository: https://github.com/b-remy/probabilistic-shear.
Contents:
- Autoencoder (AE): chekcpoints for the full AE and its distilled versions (half and quarter size).
- Normalising Flow (NF): checkpoint for the realNVP NF.
Files
model_checkpoints.zip
Additional details
Software
- Repository URL
- https://github.com/CentofantiEze/Radio_WL_Generative_Forward_Model
- Programming language
- Python