Functions — github.com/Graph-and-Geometric-Learning/hyperbolic-transformer

hub / github.com/Graph-and-Geometric-Learning/hyperbolic-transformer / functions

Functions650 in github.com/Graph-and-Geometric-Learning/hyperbolic-transformer

⨍ Functions 650◇ Types & classes 92

↓ 35 callersMethodreset_parameters

(self)

large/gnns.py:53

↓ 35 callersMethodreset_parameters

(self)

medium/gnns.py:57

↓ 19 callersMethod__init__

(self, in_channels, hidden_channels, out_channels, num_layers, dropout=.5)

medium/gnns.py:39

↓ 18 callersMethod__init__

(self, in_channels, hidden_channels, out_channels, num_layers, dropout=.5)

large/gnns.py:35

↓ 9 callersFunction_inner

(u, v, keepdim: bool = False, dim: int = -1)

large/manifolds/lorentz_math.py:37

↓ 9 callersFunction_inner

(u, v, keepdim: bool = False, dim: int = -1)

medium/manifolds/lorentz_math.py:39

↓ 9 callersFunction_inner

(u, v, keepdim: bool = False, dim: int = -1)

Hypformer/manifolds/lorentz_math.py:39

↓ 9 callersMethodget_idx_split

train_prop: The proportion of dataset for train split. Between 0 and 1. valid_prop: The proportion of dataset for validation split. B

large/dataset.py:58

↓ 7 callersFunction_inner0

(v, k: torch.Tensor, keepdim: bool = False, dim: int = -1)

large/manifolds/lorentz_math.py:70

↓ 7 callersMethodmid_point

Compute the midpoint of points on the Lorentz manifold. Parameters ---------- x : torch.Tensor Points on

medium/manifolds/lorentz.py:506

↓ 6 callersFunction_norm

(u, keepdim: bool = False, dim: int = -1)

large/manifolds/lorentz_math.py:280

↓ 6 callersFunction_norm

(u, keepdim: bool = False, dim: int = -1)

medium/manifolds/lorentz_math.py:286

↓ 6 callersFunction_norm

(u, keepdim: bool = False, dim: int = -1)

Hypformer/manifolds/lorentz_math.py:286

↓ 6 callersMethodmid_point

Compute the midpoint of points on the Lorentz manifold. Parameters ---------- x : torch.Tensor Points on

large/manifolds/lorentz.py:537

↓ 5 callersMethod__init__

Initializes the HypCLS class with the given parameters. Parameters: - `manifold` (Manifold): The manifold object.

large/manifolds/layer.py:169

↓ 5 callersMethod__init__

(self, manifold, in_channels, out_channels, bias=True)

medium/manifolds/hyp_layer.py:128

↓ 5 callersMethod__init__

(self, manifold, in_channels, out_channels, bias=True)

Hypformer/manifolds/hyp_layer.py:122

↓ 5 callersMethod__init__

Initializes the HypCLS class with the given parameters. Parameters: - `manifold` (Manifold): The manifold object.

Hypformer/manifolds/layer.py:169

↓ 5 callersFunction_dist0

(x, k: torch.Tensor, keepdim: bool = False, dim: int = -1)

medium/manifolds/lorentz_math.py:136

↓ 5 callersFunction_dist0

(x, k: torch.Tensor, keepdim: bool = False, dim: int = -1)

Hypformer/manifolds/lorentz_math.py:136

↓ 5 callersFunctionclamp

(x: torch.Tensor, min: float = float("-inf"), max: float = float("+inf"))

large/manifolds/utils.py:237

↓ 5 callersFunctionclamp

(x: torch.Tensor, min: float = float("-inf"), max: float = float("+inf"))

medium/manifolds/manifold_utils.py:238

↓ 5 callersFunctionclamp

(x: torch.Tensor, min: float = float("-inf"), max: float = float("+inf"))

Hypformer/manifolds/utils.py:237

↓ 5 callersMethodexpmap0

Perform the exponential map from the origin. Parameters: u (torch.Tensor): Tangent vector. project (bool): I

Hypformer/manifolds/lorentz.py:253

↓ 5 callersMethodmid_point

Compute the midpoint of points on the manifold. Parameters: x (torch.Tensor): Points on the manifold. w (tor

Hypformer/manifolds/lorentz.py:500

↓ 5 callersMethodnorm

Compute the norm of a tangent vector. Parameters: u (torch.Tensor): Tangent vector. keepdim (bool): If True,

Hypformer/manifolds/lorentz.py:163

↓ 4 callersFunction_inner0

(v, k: torch.Tensor, keepdim: bool = False, dim: int = -1)

medium/manifolds/lorentz_math.py:72

↓ 4 callersFunction_inner0

(v, k: torch.Tensor, keepdim: bool = False, dim: int = -1)

Hypformer/manifolds/lorentz_math.py:72

↓ 4 callersFunctioneval_acc

true: (n, 1) pred: (n, c)

large/eval.py:140

↓ 4 callersMethodnorm

Compute the norm of a tangent vector. Parameters: u (torch.Tensor): Tangent vector. keepdim (bool): If True,

large/manifolds/lorentz.py:183

↓ 4 callersMethodnorm

Compute the norm of a tangent vector. Parameters: u (torch.Tensor): Tangent vector. keepdim (bool): If True,

medium/manifolds/lorentz.py:146

↓ 4 callersMethodprint_statistics

(self, run=None, mode='max_acc')

large/logger.py:39

↓ 3 callersFunction_dist

(x, y, k: torch.Tensor, keepdim: bool = False, dim: int = -1)

medium/manifolds/lorentz_math.py:107

↓ 3 callersFunction_dist

(x, y, k: torch.Tensor, keepdim: bool = False, dim: int = -1)

Hypformer/manifolds/lorentz_math.py:107

↓ 3 callersFunction_dist0

(x, k: torch.Tensor, keepdim: bool = False, dim: int = -1)

large/manifolds/lorentz_math.py:134

↓ 3 callersFunction_logmap

(x, y, k, dim: int = -1)

medium/manifolds/lorentz_math.py:398

↓ 3 callersFunction_logmap

(x, y, k, dim: int = -1)

Hypformer/manifolds/lorentz_math.py:398

↓ 3 callersFunction_logmap0

(y, k, dim: int = -1)

medium/manifolds/lorentz_math.py:425

↓ 3 callersFunction_logmap0

(y, k, dim: int = -1)

Hypformer/manifolds/lorentz_math.py:425

↓ 3 callersFunction_logmap0back

(x, k, dim: int = -1)

medium/manifolds/lorentz_math.py:454

↓ 3 callersFunction_logmap0back

(x, k, dim: int = -1)

Hypformer/manifolds/lorentz_math.py:454

↓ 3 callersFunctionarcosh

(x: torch.Tensor)

large/manifolds/lorentz_math.py:7

↓ 3 callersFunctionarcosh

Computes the inverse hyperbolic cosine safely.

medium/manifolds/lorentz_math.py:8

↓ 3 callersFunctionarcosh

Computes the inverse hyperbolic cosine safely.

Hypformer/manifolds/lorentz_math.py:8

↓ 3 callersMethodexpmap0

Perform the exponential map from the origin. Parameters: u (torch.Tensor): Tangent vector. project (bool): I

large/manifolds/lorentz.py:273

↓ 3 callersMethodexpmap0

Perform the exponential map from the origin. Parameters: u (torch.Tensor): Tangent vector. project (bool): I

medium/manifolds/lorentz.py:236

↓ 3 callersMethodinner

Compute the inner product of two tangent vectors at a point. Parameters: x (torch.Tensor): Point on the manifold.

large/manifolds/lorentz.py:331

↓ 3 callersMethodinner

Compute the inner product of two tangent vectors at a point. Parameters: x (torch.Tensor): Point on the manifold.

Hypformer/manifolds/lorentz.py:311

↓ 3 callersMethodlogmap0

Perform the logarithmic map from the origin. Parameters: y (torch.Tensor): Point on the manifold. dim (int):

large/manifolds/lorentz.py:305

↓ 3 callersMethodlogmap0

Perform the logarithmic map from the origin. Parameters: y (torch.Tensor): Point on the manifold. dim (int):

medium/manifolds/lorentz.py:268

↓ 3 callersFunctionrand_train_test_idx

randomly splits label into train/valid/test splits

large/data_utils.py:13

↓ 3 callersMethodsave

(self, params, results, filename)

large/logger.py:144

↓ 3 callersFunctionsqrt

(x: torch.Tensor)

medium/manifolds/manifold_utils.py:218

↓ 2 callersMethod__init__

(self, manifold_in, manifold_hidden, manifold_out, in_channels, hidden_channels, num_layers=2, num_heads=1,

large/hypformer.py:180

↓ 2 callersMethod__init__

(self, manifold_in, manifold_hidden, manifold_out, in_channels, hidden_channels, args=None)

medium/hypformer.py:176

↓ 2 callersMethod__init__

(self, manifold_in, manifold_hidden, manifold_out, in_channels, hidden_channels, num_layers=2, num_heads=1,

Hypformer/hypformer.py:154

↓ 2 callersFunction_dist

(x, y, k: torch.Tensor, keepdim: bool = False, dim: int = -1)

large/manifolds/lorentz_math.py:105

↓ 2 callersMethodadd_result

(self, run, result)

large/logger.py:18

↓ 2 callersMethodbackward

(ctx: Any, grad_output: torch.Tensor)

large/manifolds/utils.py:256

↓ 2 callersMethodcinner

(self, x, y)

large/manifolds/layer.py:191

↓ 2 callersMethodcinner

(self, x, y)

Hypformer/manifolds/layer.py:191

↓ 2 callersFunctioncompute_degrees

(edge_index, num_nodes)

large/data_utils.py:304

↓ 2 callersFunctionevaluate_large

(model, dataset, split_idx, eval_func, criterion, args, degrees, threshold, device="cpu", result=None)

large/eval.py:45

↓ 2 callersFunctioneven_quantile_labels

partitions vals into nclasses by a quantile based split, where the first class is less than the 1/nclasses quantile, second class is less tha

large/data_utils.py:79

↓ 2 callersMethodexpmap

Perform the exponential map to move from a point in the tangent space to the manifold. Parameters: x (torch.Tensor): Poi

large/manifolds/lorentz.py:251

↓ 2 callersMethodexpmap

Perform the exponential map to move from a point in the tangent space to the manifold. Parameters: x (torch.Tensor): Poi

medium/manifolds/lorentz.py:214

↓ 2 callersMethodexpmap

Perform the exponential map to move from a point in the tangent space to the manifold. Parameters: x (torch.Tensor): Poi

Hypformer/manifolds/lorentz.py:231

↓ 2 callersMethodfp

(x, p=2)

large/hypformer.py:81

↓ 2 callersMethodfp

(x, p=2)

medium/hypformer.py:79

↓ 2 callersMethodfp

(x, p=2)

Hypformer/hypformer.py:42

↓ 2 callersMethodfull_attention

(self, qs, ks, vs, output_attn=False)

large/hypformer.py:61

↓ 2 callersMethodfull_attention

(self, qs, ks, vs, output_attn=False)

medium/hypformer.py:59

↓ 2 callersMethodfull_attention

(self, qs, ks, vs, output_attn=False)

Hypformer/hypformer.py:47

↓ 2 callersMethodget_attentions

(self, x_input)

medium/hypformer.py:248

↓ 2 callersMethodinner

Compute the inner product of two tangent vectors at a point. Parameters: x (torch.Tensor): Point on the manifold.

medium/manifolds/lorentz.py:294

↓ 2 callersMethodlinear_focus_attention

(self, hyp_qs, hyp_ks, hyp_vs, output_attn=False)

large/hypformer.py:86

↓ 2 callersMethodlinear_focus_attention

(self, hyp_qs, hyp_ks, hyp_vs, output_attn=False)

medium/hypformer.py:84

↓ 2 callersMethodlinear_focus_attention

(self, hyp_qs, hyp_ks, hyp_vs, output_attn=False)

Hypformer/hypformer.py:66

↓ 2 callersFunctionload_dataset

Loader for NCDataset Returns NCDataset

large/dataset.py:89

↓ 2 callersFunctionload_fixed_splits

(data_dir, dataset, name, protocol)

large/data_utils.py:39

↓ 2 callersFunctionload_papers100M

(data_dir)

large/dataset.py:301

↓ 2 callersMethodlogmap0

Perform the logarithmic map from the origin. Parameters: y (torch.Tensor): Point on the manifold. dim (int):

Hypformer/manifolds/lorentz.py:285

↓ 2 callersFunctionnormalize_feat

Row-normalize np or sparse matrix.

medium/data_utils.py:63

↓ 2 callersFunctionpapers100M_sub

(data_dir)

large/dataset.py:628

↓ 2 callersFunctionparse_method

(args, c, d, device)

large/parse.py:4

↓ 2 callersFunctionparser_add_main_args

(parser)

large/parse.py:15

↓ 2 callersMethodprint_statistics

(self, run=None, mode='max_acc')

medium/logger.py:31

↓ 2 callersMethodrandom_normal

Create a random point on the manifold, induced by a normal distribution on the tangent space of zero. Parameters: size:

Hypformer/manifolds/lorentz.py:455

↓ 2 callersMethodreset_parameters

Reset layer parameters.

large/manifolds/layer.py:149

↓ 2 callersMethodreset_parameters

(self)

medium/manifolds/hyp_layer.py:109

↓ 2 callersMethodreset_parameters

(self)

Hypformer/manifolds/hyp_layer.py:105

↓ 2 callersMethodreset_parameters

Reset layer parameters.

Hypformer/manifolds/layer.py:149

↓ 2 callersMethodsave

(self, params, results, filename)

medium/logger.py:87

↓ 2 callersFunctionsplit_data

(labels, val_prop, test_prop, seed=1234)

medium/data_utils.py:168

↓ 2 callersFunctionsqrt

(x: torch.Tensor)

large/manifolds/utils.py:217

↓ 2 callersMethodstep

Performs a single optimization step.

large/manifolds/layer.py:255

↓ 2 callersMethodzero_grad

Sets the gradients of all optimized tensors to zero.

large/manifolds/layer.py:260

↓ 1 callersFunction_egrad2rgrad

(x, grad, k, dim: int = -1)

large/manifolds/lorentz_math.py:485

↓ 1 callersFunction_egrad2rgrad

(x, grad, k, dim: int = -1)

medium/manifolds/lorentz_math.py:491

↓ 1 callersFunction_egrad2rgrad

(x, grad, k, dim: int = -1)

Hypformer/manifolds/lorentz_math.py:491

next →1–100 of 650, ranked by callers