model_weights/model.decoder.layers.self_attention.linear_qkv.weight/.zarray

{
    "chunks": [
        1,
        2688,
        18432
    ],
    "compressor": null,
    "dtype": "bfloat16",
    "fill_value": null,
    "filters": null,
    "order": "C",
    "shape": [
        96,
        21504,
        18432
    ],
    "zarr_format": 2
}