From 6259f2a5f2a891ed753d9b68ad040241300cdab9 Mon Sep 17 00:00:00 2001
From: Patrick von Platen <patrick.v.platen@gmail.com>
Date: Mon, 6 Jun 2022 17:43:36 +0200
Subject: [PATCH] Update README.md

---
 README.md | 56 ++++++++++++++++++++++++++++++++++++++++---------------
 1 file changed, 41 insertions(+), 15 deletions(-)

diff --git a/README.md b/README.md
index e7620704..c0355307 100644
--- a/README.md
+++ b/README.md
@@ -27,25 +27,51 @@ One should be able to save both models and samplers as well as load them from th
 Example:
 
 ```python
-from diffusers import UNetModel, GaussianDDPMScheduler
 import torch
+from diffusers import UNetModel, GaussianDDPMScheduler
+import PIL
+import numpy as np
 
-# 1. Load model
-unet = UNetModel.from_pretrained("fusing/ddpm_dummy")
+generator = torch.Generator()
+generator = generator.manual_seed(6694729458485568)
 
-# 2. Do one denoising step with model
-batch_size, num_channels, height, width = 1, 3, 32, 32
-dummy_noise = torch.ones((batch_size, num_channels, height, width))
-time_step = torch.tensor([10])
-image = unet(dummy_noise, time_step)
+# 1. Load models
+scheduler = GaussianDDPMScheduler.from_config("fusing/ddpm-lsun-church")
+model = UNetModel.from_pretrained("fusing/ddpm-lsun-church").to(torch_device)
 
-# 3. Load sampler
-sampler = GaussianDDPMScheduler.from_config("fusing/ddpm_dummy")
+# 2. Sample gaussian noise
+image = scheduler.sample_noise((1, model.in_channels, model.resolution, model.resolution), device=torch_device, generator=generator)
 
-# 4. Sample image from sampler passing the model
-image = sampler.sample(model, batch_size=1)
+# 3. Denoise                                                                                                                                           
+for t in reversed(range(len(scheduler))):
+    # i) define coefficients for time step t
+    clip_image_coeff = 1 / torch.sqrt(scheduler.get_alpha_prod(t))
+    clip_noise_coeff = torch.sqrt(1 / scheduler.get_alpha_prod(t) - 1)
+    image_coeff = (1 - scheduler.get_alpha_prod(t - 1)) * torch.sqrt(scheduler.get_alpha(t)) / (1 - scheduler.get_alpha_prod(t))
+    clip_coeff = torch.sqrt(scheduler.get_alpha_prod(t - 1)) * scheduler.get_beta(t) / (1 - scheduler.get_alpha_prod(t))
 
-print(image)
+    # ii) predict noise residual
+    with torch.no_grad():
+        noise_residual = model(image, t)
+
+    # iii) compute predicted image from residual
+    # See 2nd formula at https://github.com/hojonathanho/diffusion/issues/5#issue-896554416 for comparison
+    pred_mean = clip_image_coeff * image - clip_noise_coeff * noise_residual
+    pred_mean = torch.clamp(pred_mean, -1, 1)
+    prev_image = clip_coeff * pred_mean + image_coeff * image
+
+    # iv) sample variance
+    prev_variance = scheduler.sample_variance(t, prev_image.shape, device=torch_device, generator=generator)
+
+    # v) sample  x_{t-1} ~ N(prev_image, prev_variance)
+    sampled_prev_image = prev_image + prev_variance
+    image = sampled_prev_image
+
+image_processed = image.cpu().permute(0, 2, 3, 1)
+image_processed = (image_processed + 1.0) * 127.5
+image_processed = image_processed.numpy().astype(np.uint8)
+image_pil = PIL.Image.fromarray(image_processed[0])
+image_pil.save("test.png")
 ```
 
 ## 2. `diffusers` as a collection of most import Diffusion models (GLIDE, Dalle, ...)
@@ -117,8 +143,8 @@ with tempfile.TemporaryDirectory() as tmpdirname:
 │       ├── models
 │       │   └── unet.py
 │       ├── processors
-│       └── samplers
-│           ├── gaussian.py
+│       └── schedulers
+│           ├── gaussian_ddpm.py
 ├── tests
 │   └── test_modeling_utils.py
 ```