Functions — github.com/Sin3DM/Sin3DM

hub / github.com/Sin3DM/Sin3DM / functions

Functions339 in github.com/Sin3DM/Sin3DM

↓ 25 callersFunction_extract_into_tensor

Extract values from a 1-D numpy array for a batch of indices. :param arr: the 1-D numpy array. :param timesteps: a tensor of indices int

src/diffusion/gaussian_diffusion.py:934

↓ 20 callersMethodlog

(self, *args, level=INFO)

src/diffusion/logger.py:376

↓ 11 callersMethodclose

(self)

src/diffusion/logger.py:391

↓ 9 callersFunctionget_current

()

src/diffusion/logger.py:325

↓ 9 callersMethodsave

(self)

src/diffusion/train_util.py:258

↓ 9 callersFunctiontqdm

(x)

evaluation/sifid.py:39

↓ 8 callersMethodlogkv_mean

(self, key, val)

src/diffusion/logger.py:350

↓ 8 callersFunctionmean_flat

Take the mean over all non-batch dimensions.

src/diffusion/nn.py:86

↓ 7 callersMethod__init__

(self, channels, out_channels, kernel_size, padding, is_rollout=True)

src/diffusion/unet_triplane.py:22

↓ 6 callersFunctiondecompose_featmaps

(composed_map, sizes)

src/utils/triplane_util.py:20

↓ 6 callersFunctionlog

Write the sequence of args, with no separators, to the console and output files (if you've configured an output file).

src/diffusion/logger.py:247

↓ 5 callersFunctioncalculate_activation_statistics

Calculation of the statistics used by the FID. Params: -- files : List of image files paths -- model : Instance of inception m

evaluation/sifid.py:179

↓ 5 callersMethoddecode_batch

decode a batch of points triplane_feat: (c, h, w, d)

src/encoding/model.py:320

↓ 5 callersFunctiondraw_scalar_field2D

(arr, vmin=None, vmax=None, cmap=None)

src/utils/common_util.py:16

↓ 5 callersFunctionencoding_log_dir

(exp_tag)

src/utils/parser_util.py:217

↓ 5 callersFunctionlinear

Create a linear module.

src/diffusion/nn.py:35

↓ 5 callersFunctionload_and_overwrite_args

(args, path, ignore_keys=[])

src/utils/parser_util.py:180

↓ 4 callersMethod__init__

(self, in_channels, out_channels, hidden_channels, num_hidden_layers, posenc=0)

src/encoding/blocks.py:46

↓ 4 callersMethod_scale_timesteps

(self, t)

src/diffusion/gaussian_diffusion.py:352

↓ 4 callersMethod_wrap_model

(self, model)

src/diffusion/respace.py:104

↓ 4 callersFunctioncompose_featmaps

(feat_xy, feat_xz, feat_yz)

src/utils/triplane_util.py:7

↓ 4 callersFunctiondiffusion_log_dir

(exp_tag)

src/utils/parser_util.py:221

↓ 4 callersFunctionload_voxgrid

(path, resolution=128, device="cpu")

evaluation/patch_utils.py:21

↓ 4 callersMethodlogkv

(self, key, val)

src/diffusion/logger.py:347

↓ 4 callersMethodp_mean_variance

Apply the model to get p(x_{t-1} | x_t), as well as a prediction of the initial x, x_0. :param model: the model, which takes

src/diffusion/gaussian_diffusion.py:233

↓ 4 callersMethodq_posterior_mean_variance

Compute the mean and variance of the diffusion posterior: q(x_{t-1} | x_t, x_0)

src/diffusion/gaussian_diffusion.py:209

↓ 3 callersMethod_predict_eps_from_xstart

(self, x_t, t, pred_xstart)

src/diffusion/gaussian_diffusion.py:346

↓ 3 callersMethod_vb_terms_bpd

Get a term for the variational lower-bound. The resulting units are bits (rather than nats, as one might expect). This allow

src/diffusion/gaussian_diffusion.py:736

↓ 3 callersFunctionargs_to_dict

(args, keys)

src/utils/parser_util.py:199

↓ 3 callersFunctiondiffusion_model_defaults

()

src/utils/parser_util.py:88

↓ 3 callersFunctionenable_cuda_devices

()

rendering/blender_utils.py:4

↓ 3 callersMethodencode

encode a patch of volume

src/encoding/model.py:312

↓ 3 callersFunctionencoding_feat_path

(exp_tag)

src/utils/parser_util.py:225

↓ 3 callersFunctionfind_resume_checkpoint

()

src/diffusion/train_util.py:317

↓ 3 callersMethodget_dir

(self)

src/diffusion/logger.py:388

↓ 3 callersFunctionload_triplane_data

(path, device="cuda:0", compose=True)

src/utils/triplane_util.py:44

↓ 3 callersFunctionnormalization

Make a standard normalization layer. :param channels: number of input channels. :return: an nn.Module for normalization.

src/diffusion/nn.py:93

↓ 3 callersMethodquery_tex

(self, points)

data/mesh_sampler.py:108

↓ 3 callersMethodquery_tex

(self, points)

data/mesh_sampler_pbr.py:115

↓ 3 callersFunctionzero_module

Zero out the parameters of a module and return it.

src/diffusion/nn.py:68

↓ 2 callersMethod__init__

(self, geo_feat_channels, tex_feat_channels, feat_channel_up, mlp_hidden_channels, mlp_hidden_layers, use_tex=

src/encoding/networks.py:22

↓ 2 callersMethod__init__

(self)

evaluation/lpips.py:53

↓ 2 callersMethod_compute_norms

(self, grad_scale=1.0)

src/diffusion/fp16_util.py:217

↓ 2 callersMethod_predict_xstart_from_eps

(self, x_t, t, eps)

src/diffusion/gaussian_diffusion.py:329

↓ 2 callersMethod_resize_aabb

(self, featmap_size)

src/encoding/model.py:351

↓ 2 callersMethod_truncate

(self, s)

src/diffusion/logger.py:80

↓ 2 callersFunctionadd_base_options

(parser)

src/utils/parser_util.py:6

↓ 2 callersFunctionadd_dict_to_argparser

(parser, default_dict)

src/utils/parser_util.py:189

↓ 2 callersFunctionapprox_standard_normal_cdf

A fast approximation of the cumulative distribution function of the standard normal.

src/diffusion/losses.py:42

↓ 2 callersMethodbackward

(ctx, *output_grads)

src/diffusion/nn.py:153

↓ 2 callersFunctioncalculate_activation_statistics

Calculation of the statistics used by the FID. Returns: -- mu : The mean over samples of the activations of the inception model. -- sig

evaluation/ssfid.py:65

↓ 2 callersFunctioncalculate_frechet_distance

Numpy implementation of the Frechet Distance. The Frechet distance between two multivariate Gaussians X_1 ~ N(mu_1, C_1) and X_2 ~ N(mu_2, C_2

evaluation/sifid.py:122

↓ 2 callersFunctioncalculate_multiview_sifid_given_paths

Calculates multi view SIFID

evaluation/sifid.py:245

↓ 2 callersFunctioncompose_triplane_channelwise

(feat_maps)

src/encoding/blocks.py:164

↓ 2 callersFunctionconfigure

If comm is provided, average all numerical stats across that comm

src/diffusion/logger.py:442

↓ 2 callersFunctioncreate_model_and_diffusion_from_args

Create model and diffusion from args.

src/diffusion/script_util.py:7

↓ 2 callersFunctiondecode

(args, paths)

src/sample.py:51

↓ 2 callersMethoddecode

(self, x, feat_maps, aabb=None)

src/encoding/networks.py:89

↓ 2 callersMethoddecode_grid

decode feature volume at grid points

src/encoding/model.py:336

↓ 2 callersMethoddecode_texmesh

(self, save_dir, triplane_feat, reso, n_faces=10000, n_surf_pc=-1, texture_reso=2048, only_largest_cc=True,

src/encoding/model.py:363

↓ 2 callersFunctiondecompose_triplane_channelwise

(composed_map, sizes)

src/encoding/blocks.py:180

↓ 2 callersFunctiondiffusion_defaults

()

src/utils/parser_util.py:75

↓ 2 callersMethoddumpkvs

(self)

src/diffusion/logger.py:355

↓ 2 callersMethodevaluate

(self)

src/encoding/model.py:291

↓ 2 callersFunctionextract_valid_patches_unfold

extract near-surface patches of a 3D shape using torch.unfold Args: voxels (torch.Tensor): a 3D shape volume of size (H, W, D) pa

evaluation/patch_utils.py:46

↓ 2 callersFunctionget_args_by_group

(parser, args, group_name)

src/utils/parser_util.py:172

↓ 2 callersFunctionget_blob_logdir

()

src/diffusion/train_util.py:311

↓ 2 callersFunctionget_param_groups_and_shapes

(named_model_params)

src/diffusion/fp16_util.py:82

↓ 2 callersFunctionload_sdfgrid2vox

(path, binarize=True, resolution=128, device="cpu")

evaluation/patch_utils.py:8

↓ 2 callersFunctionmake_master_params

Copy model parameters into a (differently-shaped) list of full-precision parameters.

src/diffusion/fp16_util.py:35

↓ 2 callersFunctionnormal_kl

Compute the KL divergence between two gaussians. Shapes are automatically broadcasted, so batches can be compared to scalars, among othe

src/diffusion/losses.py:12

↓ 2 callersFunctionnormalize

(x, eps=1e-10)

evaluation/lpips.py:19

↓ 2 callersFunctionnormalize_aabb

(v, reso, enlarge_scale=1.03, mult=8)

data/utils.py:19

↓ 2 callersMethodq_sample

Diffuse the data for a given number of diffusion steps. In other words, sample from q(x_t | x_0). :param x_start: the initi

src/diffusion/gaussian_diffusion.py:189

↓ 2 callersMethodquery_sdf

(self, points)

data/mesh_sampler.py:104

↓ 2 callersMethodquery_sdf

(self, points)

data/mesh_sampler_pbr.py:111

↓ 2 callersMethodreset_aabb

(self, aabb)

src/encoding/networks.py:54

↓ 2 callersFunctionsample_diffusion

(args)

src/sample.py:6

↓ 2 callersMethodsample_feature_plane2D

Sample feature map at given coordinates

src/encoding/networks.py:79

↓ 2 callersMethodsample_feature_plane2D

Sample feature map at given coordinates

src/encoding/networks.py:182

↓ 2 callersMethodsample_feature_plane2D

Sample feature map at given coordinates

src/encoding/networks.py:289

↓ 2 callersFunctionsample_grid_points_aabb

(aabb, resolution)

data/utils.py:4

↓ 2 callersFunctionsave_triplane_data

(path, feat_xy, feat_xz, feat_yz)

src/utils/triplane_util.py:38

↓ 2 callersFunctionseed_all

(seed)

src/utils/common_util.py:7

↓ 2 callersFunctiontimestep_embedding

Create sinusoidal timestep embeddings. :param timesteps: a 1-D Tensor of N indices, one per batch element. These may b

src/diffusion/nn.py:103

↓ 2 callersMethodtrain

(self, data_path)

src/encoding/model.py:239

↓ 2 callersFunctiontrain_ae

(args)

src/train.py:8

↓ 2 callersFunctionunflatten_master_params

(param_group, master_param)

src/diffusion/fp16_util.py:78

↓ 2 callersMethodzero_grad

(self)

src/diffusion/fp16_util.py:173

↓ 2 callersFunctionzero_master_grads

(master_params)

src/diffusion/fp16_util.py:128

↓ 2 callersFunctionzero_module

Zero out the parameters of a module and return it.

src/encoding/blocks.py:99

↓ 1 callersMethod__init__

(self, model, timestep_map, rescale_timesteps, original_num_steps)

src/diffusion/respace.py:117

↓ 1 callersMethod_anneal_lr

(self)

src/diffusion/train_util.py:241

↓ 1 callersFunction_configure_default_logger

()

src/diffusion/logger.py:474

↓ 1 callersMethod_do_log

(self, args)

src/diffusion/logger.py:397

↓ 1 callersMethod_forward_batch

forward a batch of data

src/encoding/model.py:186

↓ 1 callersMethod_load

(self, path)

data/mesh_sampler.py:16

↓ 1 callersMethod_load

(self, path)

data/mesh_sampler_pbr.py:18

↓ 1 callersMethod_load_and_sync_parameters

(self)

src/diffusion/train_util.py:101

↓ 1 callersMethod_load_data

(self, path, sdf_renorm=False)

src/encoding/model.py:51

next →1–100 of 339, ranked by callers