# D-JEPA: pusht-predictive-adaptation

- Architecture: `predictor_patch`.
- Tensor count: 22.
- Includes pretrained predictor weights: false.
- Original checkpoint SHA256: `ccbd6ca30c550b60a186c5d0a1fa5f1a43b7a08b704b19037586ff3de52b0887`.
- Exported tensor-file SHA256: `b2bcb7880fb9477a7bb9aaa3c3380b66e4cd5e1767a4093f990c72766ff8e49a`.

Load `model.pt` with `torch.load(..., map_location="cpu", weights_only=True)`. Use the matching D-JEPA code profile and JSON configuration. Tensor values are unchanged; metadata and optimizer state have been separated.

Project license is selected by the authors before public upload.