Estimate pairwise alignment from a list of correspondences Args: corr_P (FloatTensor): Correspondnces P corr_Q (_type_): _descrip

Compute reasonable near and far planes (lower and upper bounds on depth). This assumes that all of an example's views are of roughly the same thin

r""" Context manager for validation/inference with averaged parameters. Equivalent to: ema.store() ema.copy_

Create a beta schedule that discretizes the given alpha_t_bar function, which defines the cumulative product of (1-beta) over time from t = [

Custom colormap: red (0) -> yellow (0.5) -> green (1).

Custom colormap to visualize pruning

Merge nested dictionaries of tensors.

Convert primitive modules to float16.

Convert primitive modules to float32, undoing convert_module_to_f16().

Convert the torso of the model to float16.

Convert the torso of the model to float32.

src_feat = src_feat / (src_feat.norm(dim=1, p=2, keepdim=True) + eps) trg_feat = trg_feat / (trg_feat.norm(dim=1, p=2, keepdim=True) + eps)

DINOv2 ViT-B/14 model (optionally) pretrained on the LVD-142M dataset.

DINOv2 ViT-L/14 model (optionally) pretrained on the LVD-142M dataset.

DINOv2 ViT-S/14 model (optionally) pretrained on the LVD-142M dataset.

Args: x: tensor of shape [..., dim] diag_cov: An optional tensor of shape `(..., dim)` representing t

Multi-Head linear attention proposed in "Transformers are RNNs" Args: queries: [N, L, H, D] keys: [N, S, H, D]

Multi-head scaled dot-product attention, a.k.a full attention. Args: queries: [N, L, H, D] keys: [N, S, H, D]

Args: x (torch.Tensor): [N, L, C] source (torch.Tensor): [N, S, C] x_mask (torch.Tensor): [N, L] (optiona

Args: feat0 (torch.Tensor): [N, L, C] feat1 (torch.Tensor): [N, S, C] mask0 (torch.Tensor): [N, L] (optio

IMPORTANT: this model is in (v b), NOT (b v), due to some historical issues. keep this in mind when performing any operation related to the vi

Apply the module to `x` given `emb` timestep embeddings.