mirror of
https://github.com/storytold/artcraft.git
synced 2026-11-16 06:42:14 +00:00
speedup
This commit is contained in:
1 parent
851b6f51fe
commit
04470c1745
2 files changed
+10
-16
No files matched your search
+3
-1
@@ -1,5 +1,7 @@
|
||||
image_width: 512
|
||||
image_height: 512
|
||||
scheduler_steps: 15
|
||||
#scheduler_steps: 15
|
||||
# NB(bt): It looks like we can get down to 2 sampling steps, perhaps ?
|
||||
scheduler_steps: 2
|
||||
scheduler_samples: 1
|
||||
|
||||
@@ -88,8 +88,11 @@ pub fn run(args: Args<'_>) -> Result<RgbImage> {
|
||||
|
||||
let init_latent_dist : DiagonalGaussianDistribution = args.model_cache.vae_encode(&input_image)?;
|
||||
|
||||
// TODO(bt,2025-02-18): This takes a little bit to generate.
|
||||
println!("Generating latents from input image...");
|
||||
let latents = (init_latent_dist.sample()? * vae_scale)?.to_device(&args.configs.device)?;
|
||||
let latents = (init_latent_dist.sample()? * vae_scale)?;
|
||||
|
||||
//.to_device(&args.configs.device)?;
|
||||
|
||||
println!("Initial latents shape: {:?}", latents.shape());
|
||||
|
||||
@@ -128,21 +131,13 @@ pub fn run(args: Args<'_>) -> Result<RgbImage> {
|
||||
continue;
|
||||
}
|
||||
|
||||
println!("Processing step {}/{}", timestep_index + 1, timesteps.len());
|
||||
let latent_model_input = latents.clone();
|
||||
println!("Processing step {}/{} @ timestamp = {}", timestep_index + 1, timesteps.len(), timestep);
|
||||
|
||||
println!("Latent input shape: {:?}", latent_model_input.shape());
|
||||
|
||||
println!("Scaling model input...");
|
||||
let latent_model_input = scheduler.scale_model_input(latent_model_input, timestep)?;
|
||||
|
||||
println!("Scaled input shape: {:?}", latent_model_input.shape());
|
||||
|
||||
println!("Running UNet inference with timestep {}...", timestep);
|
||||
// TODO(bt,2025-08-18): Do we have to clone the tensor?
|
||||
let latent_model_input = scheduler.scale_model_input(latents.clone(), timestep)?;
|
||||
|
||||
let noise_pred = match args.model_cache.unet_inference(&latent_model_input, timestep as f64, &text_embeddings) {
|
||||
Ok(pred) => {
|
||||
println!("UNet inference successful");
|
||||
pred
|
||||
},
|
||||
Err(e) => {
|
||||
@@ -150,11 +145,8 @@ pub fn run(args: Args<'_>) -> Result<RgbImage> {
|
||||
return Err(anyhow::anyhow!("UNet inference failed: {}", e));
|
||||
},
|
||||
};
|
||||
println!("Noise prediction shape: {:?}", noise_pred.shape());
|
||||
|
||||
println!("Applying scheduler step...");
|
||||
latents = scheduler.step(&noise_pred, timestep, &latents)?;
|
||||
println!("Step {}/{} completed", timestep_index + 1, timesteps.len());
|
||||
}
|
||||
|
||||
println!("Diffusion process completed, decoding image...");
|
||||
|
||||
Reference in new issue
Block a user