Functions — github.com/LaurentMazare/diffusers-rs

hub / github.com/LaurentMazare/diffusers-rs / functions

Functions171 in github.com/LaurentMazare/diffusers-rs

⨍ Functions 171◇ Types & classes 91

↓ 16 callersMethodget

(&self, name: &str)

src/utils.rs:25

↓ 12 callersMethodforward

(&self, xs: &Tensor, temb: Option<&Tensor>)

src/models/resnet.rs:77

↓ 10 callersMethodencode

Returns the distribution in the latent space.

src/models/vae.rs:276

↓ 6 callersMethoddecode

Takes as input some sampled values.

src/models/vae.rs:282

↓ 6 callersFunctioninterp

One-dimensional linear interpolation for monotonically increasing sample points, mimicking np.interp(). Based on https://github.com/pytorch/pytorch/i

src/schedulers/mod.rs:61

↓ 6 callersMethodstate_in_first_order

(&self)

src/schedulers/k_dpm_2_discrete.rs:209

↓ 5 callersMethodforward

(&self, xs: &Tensor, timestep: f64, encoder_hidden_states: &Tensor)

src/models/unet_2d.rs:243

↓ 5 callersMethodstate_in_first_order

(&self)

src/schedulers/k_dpm_2_ancestral_discrete.rs:254

↓ 5 callersMethodstate_in_first_order

(&self)

src/schedulers/heun_discrete.rs:145

↓ 4 callersFunctionbetas_for_alpha_bar

Create a beta schedule that discretizes the given alpha_t_bar function, which defines the cumulative product of `(1-beta)` over time from `t = [0,1]`.

src/schedulers/mod.rs:44

↓ 4 callersMethodbuild_clip_transformer

( &self, clip_weights: &str, device: tch::Device, )

src/pipelines/stable_diffusion.rs:179

↓ 4 callersMethodbuild_scheduler

(&self, n_steps: usize)

src/pipelines/stable_diffusion.rs:175

↓ 4 callersMethodbuild_unet

( &self, unet_weights: &str, device: Device, in_channels: i64, )

src/pipelines/stable_diffusion.rs:162

↓ 4 callersMethodbuild_vae

( &self, vae_weights: &str, device: Device, )

src/pipelines/stable_diffusion.rs:150

↓ 4 callersMethodscale_model_input

Ensures interchangeability with schedulers that need to scale the denoising model input depending on the current timestep.

src/schedulers/ddim.rs:94

↓ 4 callersMethodstep

Performs a backward step during inference.

src/schedulers/ddim.rs:99

↓ 4 callersMethodtimesteps

(&self)

src/schedulers/ddim.rs:88

↓ 3 callersMethodindex_for_timestep

(&self, timestep: f64)

src/schedulers/k_dpm_2_ancestral_discrete.rs:223

↓ 3 callersMethodindex_for_timestep

(&self, timestep: f64)

src/schedulers/k_dpm_2_discrete.rs:180

↓ 3 callersMethodindex_for_timestep

(&self, timestep: f64)

src/schedulers/heun_discrete.rs:119

↓ 3 callersMethodinit_noise_sigma

(&self)

src/schedulers/ddim.rs:149

↓ 3 callersMethodreshape_heads_to_batch_dim

(&self, xs: &Tensor)

src/models/attention.rs:82

↓ 3 callersMethodshape

(&self, xs: &Tensor, seq_len: i64, bsz: i64)

src/transformers/clip.rs:556

↓ 3 callersMethodtranspose_for_scores

(&self, xs: Tensor)

src/models/attention.rs:352

↓ 2 callersMethodattention

(&self, query: &Tensor, key: &Tensor, value: &Tensor)

src/models/attention.rs:128

↓ 2 callersMethodforward_with_additional_residuals

( &self, xs: &Tensor, timestep: f64, encoder_hidden_states: &Tensor, d

src/models/unet_2d.rs:247

↓ 2 callersMethodget_variance

(&self, timestep: usize)

src/schedulers/ddpm.rs:93

↓ 2 callersFunctionoutput_filename

( basename: &str, sample_idx: i64, num_samples: i64, timestep_idx: Option<usize>, )

examples/stable-diffusion/main.rs:175

↓ 2 callersFunctionoutput_filename

( basename: &str, sample_idx: i64, num_samples: i64, timestep_idx: Option<usize>, )

examples/controlnet/main.rs:91

↓ 2 callersMethodreshape_batch_dim_to_heads

(&self, xs: &Tensor)

src/models/attention.rs:89

↓ 2 callersFunctionrun

(args: Args)

examples/stable-diffusion-img2img/main.rs:131

↓ 2 callersFunctionrun

(args: Args)

examples/stable-diffusion/main.rs:202

↓ 2 callersFunctionrun

(args: Args)

examples/controlnet/main.rs:147

↓ 2 callersMethodsample

(&self)

src/models/vae.rs:228

↓ 1 callersMethodadd_noise

(&self, original: &Tensor, noise: Tensor, timestep: usize)

src/schedulers/ddim.rs:142

↓ 1 callersMethodbpe

(&self, token: &str)

src/transformers/clip.rs:410

↓ 1 callersMethodclip_weights

(&self)

examples/stable-diffusion-img2img/main.rs:91

↓ 1 callersMethodclip_weights

(&self)

examples/stable-diffusion/main.rs:144

↓ 1 callersMethodclip_weights

(&self)

examples/stable-diffusion-inpaint/main.rs:100

↓ 1 callersMethodconvert_model_output

Convert the model output to the corresponding type that the algorithm (DPM-Solver / DPM-Solver++) needs. DPM-Solver is designed to discretize an inte

src/schedulers/dpmsolver_multistep.rs:151

↓ 1 callersMethoddpm_solver_first_order_update

One step for the first-order DPM-Solver (equivalent to DDIM). See https://arxiv.org/abs/2206.00927 for the detailed derivation.

src/schedulers/dpmsolver_multistep.rs:218

↓ 1 callersMethodencode_pad

(&self, s: &str, pad_size_to: Option<usize>)

src/transformers/clip.rs:455

↓ 1 callersFunctionensure_data_dir

(safetensors)

scripts/get_weights.py:11

↓ 1 callersFunctionfile_open

(path: P)

src/utils.rs:6

↓ 1 callersMethodget_lms_coefficient

Compute a linear multistep coefficient

src/schedulers/lms_discrete.rs:108

↓ 1 callersMethodget_prev_sample

( &self, sample: Tensor, timestep: usize, prev_timestep: isize, model_

src/schedulers/pndm.rs:164

↓ 1 callersFunctionget_safetensors

(safetensors, weight_bits)

scripts/get_weights.py:29

↓ 1 callersFunctionget_urls

(sd_version, weight_bits)

scripts/get_weights.py:53

↓ 1 callersFunctionget_vocab

(vocab_url)

scripts/get_weights.py:44

↓ 1 callersFunctionimage_preprocess

(path: T)

examples/stable-diffusion-img2img/main.rs:122

↓ 1 callersMethodimage_preprocess

(&self, path: T)

examples/controlnet/main.rs:124

↓ 1 callersFunctionintegrate

(f: F, a: f64, b: f64, target_absolute_error: f64)

src/schedulers/integrate.rs:27

↓ 1 callersFunctionintegrate_core

Integrate f(x) from [-1.0, 1.0]

src/schedulers/integrate.rs:51

↓ 1 callersMethodmultistep_dpm_solver_second_order_update

One step for the second-order multistep DPM-Solver.

src/schedulers/dpmsolver_multistep.rs:240

↓ 1 callersMethodmultistep_dpm_solver_third_order_update

One step for the third-order multistep DPM-Solver

src/schedulers/dpmsolver_multistep.rs:293

↓ 1 callersFunctionprepare_mask_and_masked_image

( path_input: T, path_mask: T, )

examples/stable-diffusion-inpaint/main.rs:131

↓ 1 callersFunctionrun

(args: Args)

examples/stable-diffusion-inpaint/main.rs:145

↓ 1 callersMethodscale

(self, c: f64)

src/schedulers/integrate.rs:18

↓ 1 callersMethodsliced_attention

( &self, query: &Tensor, key: &Tensor, value: &Tensor, sequence_length

src/models/attention.rs:96

↓ 1 callersMethodstep_plms

Step function propagating the sample with the linear multi-step method. This has one forward pass with multiple times to approximate the solution.

src/schedulers/pndm.rs:122

↓ 1 callersMethodunet_weights

(&self)

examples/stable-diffusion-img2img/main.rs:111

↓ 1 callersMethodunet_weights

(&self)

examples/stable-diffusion/main.rs:164

↓ 1 callersMethodunet_weights

(&self)

examples/stable-diffusion-inpaint/main.rs:120

↓ 1 callersMethodvae_weights

(&self)

examples/stable-diffusion-img2img/main.rs:101

↓ 1 callersMethodvae_weights

(&self)

examples/stable-diffusion/main.rs:154

↓ 1 callersMethodvae_weights

(&self)

examples/stable-diffusion-inpaint/main.rs:110

Methodadd_noise

(&self, original_samples: &Tensor, noise: Tensor, timestep: f64)

src/schedulers/k_dpm_2_ancestral_discrete.rs:330

Methodadd_noise

(&self, original_samples: &Tensor, noise: Tensor, timestep: f64)

src/schedulers/k_dpm_2_discrete.rs:281

Methodadd_noise

(&self, original_samples: &Tensor, noise: Tensor, timestep: f64)

src/schedulers/lms_discrete.rs:173

Methodadd_noise

(&self, original_samples: &Tensor, noise: Tensor, timestep: f64)

src/schedulers/euler_ancestral_discrete.rs:139

Methodadd_noise

(&self, original_samples: &Tensor, noise: Tensor, timestep: usize)

src/schedulers/ddpm.rs:182

Methodadd_noise

(&self, original_samples: &Tensor, noise: Tensor, timestep: f64)

src/schedulers/euler_discrete.rs:147

Methodadd_noise

(&self, original: &Tensor, noise: Tensor, timestep: usize)

src/schedulers/pndm.rs:217

Methodadd_noise

(&self, original_samples: &Tensor, noise: Tensor, timestep: usize)

src/schedulers/dpmsolver_multistep.rs:400

Methodadd_noise

(&self, original_samples: &Tensor, noise: Tensor, timestep: f64)

src/schedulers/heun_discrete.rs:214

Methodbuild_causal_attention_mask

https://github.com/huggingface/transformers/blob/674f750a57431222fa2832503a108df3badf1564/src/transformers/models/clip/modeling_clip.py#L678

src/transformers/clip.rs:684

Methodcreate

Creates a new CLIP tokenizer, this takes as input the path for the bpe vocabulary file.

src/transformers/clip.rs:351

Methoddecode

The inverse of the tokenization process, takes as input a list of tokens and returns a string that produces this tokenization.

src/transformers/clip.rs:492

Methoddefault

()

src/schedulers/k_dpm_2_ancestral_discrete.rs:19

Methoddefault

()

src/schedulers/ddim.rs:35

Methoddefault

()

src/schedulers/k_dpm_2_discrete.rs:19

Methoddefault

()

src/schedulers/lms_discrete.rs:23

Methoddefault

()

src/schedulers/euler_ancestral_discrete.rs:19

Methoddefault

()

src/schedulers/ddpm.rs:14

Methoddefault

()

src/schedulers/euler_discrete.rs:19

Methoddefault

()

src/schedulers/pndm.rs:26

Methoddefault

()

src/schedulers/dpmsolver_multistep.rs:62

Methoddefault

()

src/schedulers/heun_discrete.rs:19

Methoddefault

()

src/models/resnet.rs:30

Methoddefault

()

src/models/unet_2d_blocks.rs:90

Methoddefault

()

src/models/unet_2d.rs:33

Methoddefault

https://huggingface.co/lllyasviel/sd-controlnet-canny/blob/main/config.json

src/models/controlnet.rs:68

Methoddefault

()

src/models/attention.rs:214

The main tokenization entry point, takes as input a string and returns the list of tokens.

src/transformers/clip.rs:486

Methodforward

(&self, xs: &Tensor)

src/transformers/clip.rs:18

Methodforward

(&self, xs: &Tensor)

src/models/embeddings.rs:20

Methodforward

(&self, xs: &Tensor)

src/models/unet_2d_blocks.rs:35

Methodforward

(&self, xs: &Tensor)

src/models/controlnet.rs:43

Methodforward

(&self, xs: &Tensor)

src/models/attention.rs:16

next →1–100 of 171, ranked by callers