huggingface/candle
1
0
Fork 0
mirror of https://github.com/huggingface/candle.git synced 2025-06-17 02:58:50 +00:00
Code Issues Packages Projects Releases Wiki Activity

Line-up the wuerstchen model with the python implementation. (#901)

Browse Source 
* Line-up the wuerstchen model with the python implementation.

* Missing cos.

* Fix the picture denormalization.
This commit is contained in:
Laurent Mazare
2023-09-19 21:59:44 +01:00
committed by GitHub
parent 7ad82b87e4
commit 67a486d18d
4 changed files with 10 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
candle-examples/examples/wuerstchen/main.rs
View File

@ -373,7 +373,6 @@ fn run(args: Args) -> Result<()> {
); );
let image = vqgan.decode(&(&latents * 0.3764)?)?; let image = vqgan.decode(&(&latents * 0.3764)?)?;
// TODO: Add the clamping between 0 and 1. // TODO: Add the clamping between 0 and 1.
let image = ((image / 2.)? + 0.5)?.to_device(&Device::Cpu)?;
let image = (image * 255.)?.to_dtype(DType::U8)?.i(0)?; let image = (image * 255.)?.to_dtype(DType::U8)?.i(0)?;
let image_filename = output_filename(&final_image, idx + 1, num_samples, None); let image_filename = output_filename(&final_image, idx + 1, num_samples, None);
candle_examples::save_image(&image, image_filename)? candle_examples::save_image(&image, image_filename)?

6
candle-transformers/src/models/stable_diffusion/clip.rs
View File

@ -12,6 +12,7 @@ use candle_nn::Module;
pub enum Activation { pub enum Activation {
QuickGelu, QuickGelu,
Gelu, Gelu,
GeluErf,
} }
impl Module for Activation { impl Module for Activation {
@ -19,6 +20,7 @@ impl Module for Activation {
match self { match self {
Activation::QuickGelu => xs * nn::ops::sigmoid(&(xs * 1.702f64)?)?, Activation::QuickGelu => xs * nn::ops::sigmoid(&(xs * 1.702f64)?)?,
Activation::Gelu => xs.gelu(), Activation::Gelu => xs.gelu(),
Activation::GeluErf => xs.gelu_erf(),
} }
} }
} }
@ -111,7 +113,7 @@ impl Config {
num_hidden_layers: 24, num_hidden_layers: 24,
num_attention_heads: 16, num_attention_heads: 16,
projection_dim: 1024, projection_dim: 1024,
activation: Activation::Gelu, activation: Activation::GeluErf,
} }
} }
@ -126,7 +128,7 @@ impl Config {
num_hidden_layers: 32, num_hidden_layers: 32,
num_attention_heads: 20, num_attention_heads: 20,
projection_dim: 512, projection_dim: 512,
activation: Activation::Gelu, activation: Activation::GeluErf,
} }
} }
} }

4
candle-transformers/src/models/wuerstchen/common.rs
View File

@ -100,7 +100,7 @@ impl GlobalResponseNorm {
impl Module for GlobalResponseNorm { impl Module for GlobalResponseNorm {
fn forward(&self, xs: &Tensor) -> Result<Tensor> { fn forward(&self, xs: &Tensor) -> Result<Tensor> {
let agg_norm = xs.sqr()?.sum_keepdim((1, 2))?; let agg_norm = xs.sqr()?.sum_keepdim((1, 2))?.sqrt()?;
let stand_div_norm = let stand_div_norm =
agg_norm.broadcast_div(&(agg_norm.mean_keepdim(D::Minus1)? + 1e-6)?)?; agg_norm.broadcast_div(&(agg_norm.mean_keepdim(D::Minus1)? + 1e-6)?)?;
xs.broadcast_mul(&stand_div_norm)? xs.broadcast_mul(&stand_div_norm)?
@ -152,7 +152,7 @@ impl ResBlock {
.permute((0, 2, 3, 1))?; .permute((0, 2, 3, 1))?;
let xs = xs let xs = xs
.apply(&self.channelwise_lin1)? .apply(&self.channelwise_lin1)?
.gelu()? .gelu_erf()?
.apply(&self.channelwise_grn)? .apply(&self.channelwise_grn)?
.apply(&self.channelwise_lin2)? .apply(&self.channelwise_lin2)?
.permute((0, 3, 1, 2))?; .permute((0, 3, 1, 2))?;

6
candle-transformers/src/models/wuerstchen/ddpm.rs
View File

@ -52,8 +52,10 @@ impl DDPMWScheduler {
} else { } else {
t t
}; };
let alpha_cumprod = let alpha_cumprod = ((t + s) / (1. + s) * std::f64::consts::PI * 0.5)
((t + s) / (1. + s) * std::f64::consts::PI * 0.5).powi(2) / self.init_alpha_cumprod; .cos()
.powi(2)
/ self.init_alpha_cumprod;
alpha_cumprod.clamp(0.0001, 0.9999) alpha_cumprod.clamp(0.0001, 0.9999)
} }