# D-JEPA: pusht-adapted-predictor-full

- Architecture: `pretrained_predictor`.
- Tensor count: 309.
- Includes pretrained predictor weights: true.
- Original checkpoint SHA256: `ccbd6ca30c550b60a186c5d0a1fa5f1a43b7a08b704b19037586ff3de52b0887`.
- Exported tensor-file SHA256: `26cb528b6ba5b3b6e7d464eeb9a75f599ac28641617666e0638c9e51de2b9083`.

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.

Publication hold: upstream weight redistribution terms must be confirmed before upload.