Upload modeling_ddim.py

modeling_ddim.py

@@ -34,61 +34,49 @@ class DDIM(DiffusionPipeline):
         inference_step_times = range(0, num_trained_timesteps, num_trained_timesteps // num_inference_steps)
 
         self.unet.to(torch_device)
-
-        # Sample gaussian noise to begin loop
         image = self.noise_scheduler.sample_noise(
             (batch_size, self.unet.in_channels, self.unet.resolution, self.unet.resolution),
             device=torch_device,
             generator=generator,
         )
 
-        # See formulas (9), (10) and (7) of DDIM paper https://arxiv.org/pdf/2010.02502.pdf
-        # Ideally, read DDIM paper in-detail understanding
-
-        # Notation (<variable name> -> <name in paper>
-        # - pred_noise_t -> e_theta(x_t, t)
-        # - pred_original_image -> f_theta(x_t, t) or x_0
-        # - std_dev_t -> sigma_t
-
         for t in tqdm.tqdm(reversed(range(num_inference_steps)), total=num_inference_steps):
-            # 1. predict noise residual
-            with torch.no_grad():
-                pred_noise_t = self.unet(image, inference_step_times[t])
-
-            # 2. get actual t and t-1
+            # get actual t and t-1
             train_step = inference_step_times[t]
             prev_train_step = inference_step_times[t - 1] if t > 0 else -1
 
-            # 3. compute alphas, betas
+            # compute alphas
             alpha_prod_t = self.noise_scheduler.get_alpha_prod(train_step)
             alpha_prod_t_prev = self.noise_scheduler.get_alpha_prod(prev_train_step)
+            alpha_prod_t_rsqrt = 1 / alpha_prod_t.sqrt()
+            alpha_prod_t_prev_rsqrt = 1 / alpha_prod_t_prev.sqrt()
             beta_prod_t_sqrt = (1 - alpha_prod_t).sqrt()
             beta_prod_t_prev_sqrt = (1 - alpha_prod_t_prev).sqrt()
 
-            # 4. Compute predicted previous image from predicted noise
-            # First: compute predicted original image from predicted noise also called
-            # "predicted x_0" of formula (12) from https://arxiv.org/pdf/2010.02502.pdf
-            pred_original_image = (image - beta_prod_t_sqrt * pred_noise_t) / alpha_prod_t.sqrt()
-            # Second: Clip "predicted x_0"
-            pred_original_image = torch.clamp(pred_original_image, -1, 1)
-            # Third: Compute variance: "sigma_t" -> see 
-#            std_dev_t = (1 - alpha_prod_t / alpha_prod_t_prev).sqrt() * beta_prod_t_prev_sqrt / beta_prod_t_sqrt
-            std_dev_t = (1 - alpha_prod_t / alpha_prod_t_prev).sqrt()
-            std_dev_t = std_dev_t * eta
-            # Fourth: Compute "direction pointing to x_t" of formula (12) from https://arxiv.org/pdf/2010.02502.pdf
-            pred_image_direction = (1 - alpha_prod_t_prev - std_dev_t**2).sqrt() * pred_noise_t
-
-            # Fourth: Compute outer formula (DDIM formula)
-            pred_prev_image = alpha_prod_t_prev.sqrt() * pred_original_image + pred_image_direction 
+            # compute relevant coefficients
+            coeff_1 = (
+                (alpha_prod_t_prev - alpha_prod_t).sqrt()
+                * alpha_prod_t_prev_rsqrt
+                * beta_prod_t_prev_sqrt
+                / beta_prod_t_sqrt
+                * eta
+            )
+            coeff_2 = ((1 - alpha_prod_t_prev) - coeff_1**2).sqrt()
+
+            # model forward
+            with torch.no_grad():
+                noise_residual = self.unet(image, train_step)
+
+            # predict mean of prev image
+            pred_mean = alpha_prod_t_rsqrt * (image - beta_prod_t_sqrt * noise_residual)
+            pred_mean = torch.clamp(pred_mean, -1, 1)
+            pred_mean = (1 / alpha_prod_t_prev_rsqrt) * pred_mean + coeff_2 * noise_residual
 
             # if eta > 0.0 add noise. Note eta = 1.0 essentially corresponds to DDPM
             if eta > 0.0:
                 noise = self.noise_scheduler.sample_noise(image.shape, device=image.device, generator=generator)
-                prev_image = pred_prev_image + std_dev_t * noise
+                image = pred_mean + coeff_1 * noise
             else:
-                prev_image = pred_prev_image
-
-            # Set current image to prev_image: x_t -> x_t-1
-            image = prev_image
+                image = pred_mean
 
         return image