Functions — github.com/LouisRouss/Diffusion-Based-Model-for-Colorization

hub / github.com/LouisRouss/Diffusion-Based-Model-for-Colorization / functions

Functions47 in github.com/LouisRouss/Diffusion-Based-Model-for-Colorization

⨍ Functions 47 ◇ Types & classes 16

↓ 8 callersFunctionextract

(a, t, x_shape)

src/diffusion.py:10

↓ 7 callersMethod__init__

(self, beta)

src/network.py:96

↓ 4 callersFunctionnormalization

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

src/network.py:13

↓ 3 callersFunctionzero_module

Zero out the parameters of a module and return it.

src/network.py:21

↓ 2 callersFunctioncheckpoint

Evaluate a function without caching intermediate activations, allowing for reduced memory at the expense of extra compute in the backward pas

src/network.py:29

↓ 2 callersFunctioncount_flops_attn

A counter for the `thop` package to count the operations in an attention operation. Meant to be used like: macs, params = thop.pr

src/network.py:343

↓ 2 callersMethodsave_model

(self,name,EMA=False)

src/trainer.py:69

↓ 1 callersMethodbackward

(ctx, *output_grads)

src/network.py:76

↓ 1 callersMethodforward

(self, x)

src/network.py:152

↓ 1 callersFunctionload_config

(path)

src/config.py:26

↓ 1 callersFunctionmain

()

main.py:5

↓ 1 callersMethodnoise_prediction

Use the NN to predict the noise added between y_{t-1} and y_t

src/diffusion.py:48

↓ 1 callersMethodnoisy_image

Compute y_noisy according to (6) p15 of [2]

src/diffusion.py:42

↓ 1 callersFunctiontimestep_embedding

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

src/network.py:45

↓ 1 callersFunctiontrain

(config)

src/train.py:5

↓ 1 callersMethodtrain

(self)

src/trainer.py:75

↓ 1 callersMethodupdate_average

(self, old, new)

src/network.py:105

↓ 1 callersMethodupdate_model_average

(self, ma_model, current_model)

(self, channels, use_conv, out_channels=None)

src/network.py:144

Method__init__

(self, channels, use_conv, out_channels=None)

src/network.py:166

Method__init__

( self, channels, emb_channels, dropout, out_channels=None, us

src/network.py:200

Method__init__

( self, channels, num_heads=1, num_head_channels=-1, use_checkpoint=Fa

( self, image_size, in_channels, model_channels, out_channels,

src/network.py:454

Method__init__

(self,path_color,path_gray)

src/dataloader.py:9

Method__init__

( #Remplacer par fichier config self, image_size = (224,224), channel_y = 3,

(ctx, run_function, length, *args)

src/network.py:67

Methodforward

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

Apply the block to a Tensor, conditioned on a timestep embedding. :param x: an [N x C x ...] Tensor of features. :param emb:

Apply QKV attention. :param qkv: an [N x (H * 3 * C) x T] tensor of Qs, Ks, and Vs. :return: an [N x (H * C) x T] tensor afte

src/network.py:371

Methodforward

Apply QKV attention. :param qkv: an [N x (3 * H * C) x T] tensor of Qs, Ks, and Vs. :return: an [N x (H * C) x T] tensor afte

src/network.py:403

Methodforward

Apply the model to an input batch. :param y: a [N x 3 x ...] Tensor of noisy colored images :param x: an [N x 1 x ...] Tenso