Function conv3d_patch_embed 

pub fn conv3d_patch_embed(
    patch: &Vjepa2PatchEmbedWeights,
    video_ncthw: &[f32],
    frames: usize,
    height: usize,
    width: usize,
) -> Result<Vec<f32>, Error>

3-D conv patch embedding: input [C, T, H, W] → tokens [seq, embed_dim].