{
  "scope": "Genuine original V-JEPA ViT-L frozen features; each photograph repeated for 16 frames; appearance only, no motion or action recognition.",
  "provenance": {
    "architecture": "Original V-JEPA ViT-L/16, 24 layers, width 1024, 16 heads, 2-frame tubelets",
    "uniformPower": true,
    "sourceRepository": "https://github.com/facebookresearch/jepa",
    "sourceLicense": "https://github.com/facebookresearch/jepa/blob/51c59d518fc63c08464af6de585f78ac0c7ed4d5/LICENSE",
    "revision": "51c59d518fc63c08464af6de585f78ac0c7ed4d5",
    "checkpointUrl": "https://dl.fbaipublicfiles.com/jepa/vitl16/vitl16.pth.tar",
    "checkpointSha256": "0cd3bfce97d04891a31eac70d573ef94f423d858a6b0dc0b3ddf5f8d00b33fe4",
    "checkpointKey": "target_encoder",
    "removedPrefix": "module.backbone.",
    "strictLoadedTensors": 293,
    "ignoredTensors": [],
    "sourceHashes": {
      "src/models/vision_transformer.py": "3d2dd861d535434c4202c1b488436433fc545d558899a48c392c240b4afe8abf",
      "src/models/utils/modules.py": "b4e2f4259a2255173d1059865691e7e45974d121781601035c44c589a1bf62e7",
      "src/models/utils/patch_embed.py": "8167c7a9a9a15dfff81a0e3b8cdfcbfab6881cc0da396cc130c4729ccba75c11",
      "src/models/utils/pos_embs.py": "3fff7d09a8e99eaaca0c3e436ecc5794dc9e62245fe3e5489e8cd37a23f771a8",
      "src/masks/utils.py": "2297fb30aabdbc635a7b9e9793806cebec62367f032d4b9bd57fbb4e1681a381",
      "src/utils/tensors.py": "774a0f81eea52d2fcc4373818d8aab04e790c14f029e37e415ec0e5f6df2a8cb"
    },
    "programSha256": "6cc077a9422e1728eb123f755239451f2d59a8344001d7a0ac7492e8d792e8c9",
    "parameterCount": 305490944,
    "python": "3.13.13",
    "torch": "2.12.0+cpu",
    "numpy": "2.1.2",
    "pillow": "12.2.0",
    "device": "cpu",
    "dtype": "float32",
    "command": "python examples/vjepa_export.py --checkpoint .context/paper-series/weights/vjepa-vitl16.pth.tar --official-dir .context/paper-series/research/jepa1 --photos-dir public/examples/vit-demo --output public/examples/vjepa-demo",
    "adapter": "Hash-pinned native author encoder without numerical edits; local import wiring only; original absolute 3D sin/cos positions retained."
  },
  "preprocessing": {
    "resize": "PIL bilinear, shortest edge 224; floor long edge",
    "crop": "224x224 center crop; floor offsets",
    "frames": 16,
    "temporalInput": "16 identical copies of a still photograph, no measured motion",
    "mean": [
      0.485,
      0.456,
      0.406
    ],
    "std": [
      0.229,
      0.224,
      0.225
    ],
    "pooling": "Mean of all 1568 final-layer patch features, without extra feature normalization"
  },
  "photos": [
    {
      "id": "cat",
      "title": "Chelsea the cat",
      "source": {
        "url": "https://raw.githubusercontent.com/scikit-image/scikit-image/v0.25.2/skimage/data/chelsea.png",
        "sha256": "596aa1e7cb875eb79f437e310381d26b338a81c2da23439704a73c4651e8c4bb",
        "author": "Stefan van der Walt",
        "license": "CC0",
        "licenseUrl": "https://github.com/scikit-image/scikit-image/blob/v0.25.2/skimage/data/_fetchers.py"
      },
      "inputImage": "input-cat.png",
      "inputSha256": "e2110d52adf04eec7374c86ffde8a547f767ea726c5e7a8116936670343b0cfc",
      "inputShape": [
        1,
        3,
        16,
        224,
        224
      ],
      "resizedSize": [
        336,
        224
      ],
      "cropBox": [
        56,
        0,
        280,
        224
      ],
      "tokenShape": [
        1,
        1568,
        1024
      ],
      "meanTokenL2": 43.44794845581055,
      "pooledL2": 23.38578224182129,
      "pooledFirst8": [
        1.6696064472198486,
        0.6896965503692627,
        -1.1278133392333984,
        -0.6494178175926208,
        0.07895063608884811,
        1.5066343545913696,
        0.22465631365776062,
        1.344786524772644
      ]
    },
    {
      "id": "coffee",
      "title": "Cup of coffee",
      "source": {
        "url": "https://raw.githubusercontent.com/scikit-image/scikit-image/v0.25.2/skimage/data/coffee.png",
        "sha256": "cc02f8ca188b167c775a7101b5d767d1e71792cf762c33d6fa15a4599b5a8de7",
        "author": "Rachel Michetti",
        "license": "CC0",
        "licenseUrl": "https://github.com/scikit-image/scikit-image/blob/v0.25.2/skimage/data/_fetchers.py"
      },
      "inputImage": "input-coffee.png",
      "inputSha256": "a6eaec698c2cb60269e5d645f7fbe37ade709a435638a651031d487eff0fc28c",
      "inputShape": [
        1,
        3,
        16,
        224,
        224
      ],
      "resizedSize": [
        336,
        224
      ],
      "cropBox": [
        56,
        0,
        280,
        224
      ],
      "tokenShape": [
        1,
        1568,
        1024
      ],
      "meanTokenL2": 43.13788604736328,
      "pooledL2": 21.512243270874023,
      "pooledFirst8": [
        1.0002093315124512,
        0.13911016285419464,
        0.0074328030459582806,
        0.29598888754844666,
        -0.20811359584331512,
        -0.025577113032341003,
        -0.05052928626537323,
        0.7041651606559753
      ]
    }
  ],
  "cosineSimilarity": [
    [
      1.0,
      0.745404839515686
    ],
    [
      0.745404839515686,
      0.9999998807907104
    ]
  ],
  "features": {
    "file": "pooled-features.npy",
    "shape": [
      2,
      1024
    ],
    "sha256": "8edfed7b9a5d57ef9c7581dcdf08cfa6a8d5d319e1be26f7e8c61a6c6210c12b"
  }
}
