Neural Network Artifacts as a New Data Modality

The workshop treats trained models and their artefacts as data. Researchers examine weights, gradients, representations and optimization histories across model populations, seeking methods to compare, search, predict and modify model behaviour and understand relationships among models. Session: 08:30–18:30 (Europe/Paris). A paid NeurIPS virtual pass includes workshop livestreams; on-site access follows the host-location registration rules.