bring in prediction of v objective, combining the findings from progr…

…essive distillation paper and imagen-video to the eventual extension of dalle2 to make-a-video

README.md

@@ -1298,4 +1298,14 @@ For detailed information on training the diffusion prior, please refer to the [d
 }
 ```
 
+```bibtex
+@article{Salimans2022ProgressiveDF,
+    title   = {Progressive Distillation for Fast Sampling of Diffusion Models},
+    author  = {Tim Salimans and Jonathan Ho},
+    journal = {ArXiv},
+    year    = {2022},
+    volume  = {abs/2202.00512}
+}
+```
+
 *Creating noise from data is easy; creating data from noise is generative modeling.* - <a href="https://arxiv.org/abs/2011.13456">Yang Song's paper</a>

dalle2_pytorch/dalle2_pytorch.py

@@ -619,14 +619,20 @@ def q_posterior(self, x_start, x_t, t):
         posterior_log_variance_clipped = extract(self.posterior_log_variance_clipped, t, x_t.shape)
         return posterior_mean, posterior_variance, posterior_log_variance_clipped
 
-    def q_sample(self, x_start, t, noise=None):
+    def q_sample(self, x_start, t, noise = None):
         noise = default(noise, lambda: torch.randn_like(x_start))
 
         return (
             extract(self.sqrt_alphas_cumprod, t, x_start.shape) * x_start +
             extract(self.sqrt_one_minus_alphas_cumprod, t, x_start.shape) * noise
         )
 
+    def calculate_v(self, x_start, t, noise = None):
+        return (
+            extract(self.sqrt_alphas_cumprod, t, x_start.shape) * noise -
+            extract(self.sqrt_one_minus_alphas_cumprod, t, x_start.shape) * x_start
+        )
+
     def q_sample_from_to(self, x_from, from_t, to_t, noise = None):
         shape = x_from.shape
         noise = default(noise, lambda: torch.randn_like(x_from))
@@ -638,6 +644,12 @@ def q_sample_from_to(self, x_from, from_t, to_t, noise = None):
 
         return x_from * (alpha_next / alpha) + noise * (sigma_next * alpha - sigma * alpha_next) / alpha
 
+    def predict_start_from_v(self, x_t, t, v):
+        return (
+            extract(self.sqrt_alphas_cumprod, t, x_t.shape) * x_t -
+            extract(self.sqrt_one_minus_alphas_cumprod, t, x_t.shape) * v
+        )
+
     def predict_start_from_noise(self, x_t, t, noise):
         return (
             extract(self.sqrt_recip_alphas_cumprod, t, x_t.shape) * x_t -
@@ -1146,6 +1158,7 @@ def __init__(
         image_cond_drop_prob = None,
         loss_type = "l2",
         predict_x_start = True,
+        predict_v = False,
         beta_schedule = "cosine",
         condition_on_text_encodings = True,  # the paper suggests this is needed, but you can turn it off for your CLIP preprocessed text embed -> image embed training
         sampling_clamp_l2norm = False,       # whether to l2norm clamp the image embed at each denoising iteration (analogous to -1 to 1 clipping for usual DDPMs)
@@ -1197,6 +1210,7 @@ def __init__(
         # in paper, they do not predict the noise, but predict x0 directly for image embedding, claiming empirically better results. I'll just offer both.
 
         self.predict_x_start = predict_x_start
+        self.predict_v = predict_v # takes precedence over predict_x_start
 
         # @crowsonkb 's suggestion - https://github.com/lucidrains/DALLE2-pytorch/issues/60#issue-1226116132
 
@@ -1226,7 +1240,9 @@ def p_mean_variance(self, x, t, text_cond, self_cond = None, clip_denoised = Fal
 
         pred = self.net.forward_with_cond_scale(x, t, cond_scale = cond_scale, self_cond = self_cond, **text_cond)
 
-        if self.predict_x_start:
+        if self.predict_v:
+            x_start = self.noise_scheduler.predict_start_from_v(x, t = t, v = pred)
+        elif self.predict_x_start:
             x_start = pred
         else:
             x_start = self.noise_scheduler.predict_start_from_noise(x, t = t, noise = pred)
@@ -1299,7 +1315,9 @@ def p_sample_loop_ddim(self, shape, text_cond, *, timesteps, eta = 1., cond_scal
 
             # derive x0
 
-            if self.predict_x_start:
+            if self.predict_v:
+                x_start = self.noise_scheduler.predict_start_from_v(image_embed, t = time_cond, v = pred)
+            elif self.predict_x_start:
                 x_start = pred
             else:
                 x_start = self.noise_scheduler.predict_start_from_noise(image_embed, t = time_cond, noise = pred_noise)
@@ -1314,7 +1332,7 @@ def p_sample_loop_ddim(self, shape, text_cond, *, timesteps, eta = 1., cond_scal
 
             # predict noise
 
-            if self.predict_x_start:
+            if self.predict_x_start or self.predict_v:
                 pred_noise = self.noise_scheduler.predict_noise_from_start(image_embed, t = time_cond, x0 = x_start)
             else:
                 pred_noise = pred
@@ -1372,7 +1390,12 @@ def p_losses(self, image_embed, times, text_cond, noise = None):
         if self.predict_x_start and self.training_clamp_l2norm:
             pred = self.l2norm_clamp_embed(pred)
 
-        target = noise if not self.predict_x_start else image_embed
+        if self.predict_v:
+            target = self.noise_scheduler.calculate_v(image_embed, times, noise)
+        elif self.predict_x_start:
+            target = image_embed
+        else:
+            target = noise
 
         loss = self.noise_scheduler.loss_fn(pred, target)
         return loss
@@ -2448,6 +2471,7 @@ def __init__(
         loss_type = 'l2',
         beta_schedule = None,
         predict_x_start = False,
+        predict_v = False,
         predict_x_start_for_latent_diffusion = False,
         image_sizes = None,                         # for cascading ddpm, image size at each stage
         random_crop_sizes = None,                   # whether to random crop the image at that stage in the cascade (super resoluting convolutions at the end may be able to generalize on smaller crops)
@@ -2620,6 +2644,10 @@ def __init__(
 
         self.predict_x_start = cast_tuple(predict_x_start, len(unets)) if not predict_x_start_for_latent_diffusion else tuple(map(lambda t: isinstance(t, VQGanVAE), self.vaes))
 
+        # predict v
+
+        self.predict_v = cast_tuple(predict_v, len(unets))
+
         # input image range
 
         self.input_image_range = (-1. if not auto_normalize_img else 0., 1.)
@@ -2731,14 +2759,16 @@ def dynamic_threshold(self, x):
         x = x.clamp(-s, s) / s
         return x
 
-    def p_mean_variance(self, unet, x, t, image_embed, noise_scheduler, text_encodings = None, lowres_cond_img = None, self_cond = None, clip_denoised = True, predict_x_start = False, learned_variance = False, cond_scale = 1., model_output = None, lowres_noise_level = None):
+    def p_mean_variance(self, unet, x, t, image_embed, noise_scheduler, text_encodings = None, lowres_cond_img = None, self_cond = None, clip_denoised = True, predict_x_start = False, predict_v = False, learned_variance = False, cond_scale = 1., model_output = None, lowres_noise_level = None):
         assert not (cond_scale != 1. and not self.can_classifier_guidance), 'the decoder was not trained with conditional dropout, and thus one cannot use classifier free guidance (cond_scale anything other than 1)'
 
         model_output = default(model_output, lambda: unet.forward_with_cond_scale(x, t, image_embed = image_embed, text_encodings = text_encodings, cond_scale = cond_scale, lowres_cond_img = lowres_cond_img, self_cond = self_cond, lowres_noise_level = lowres_noise_level))
 
         pred, var_interp_frac_unnormalized = self.parse_unet_output(learned_variance, model_output)
 
-        if predict_x_start:
+        if predict_v:
+            x_start = noise_scheduler.predict_start_from_v(x, t = t, v = pred)
+        elif predict_x_start:
             x_start = pred
         else:
             x_start = noise_scheduler.predict_start_from_noise(x, t = t, noise = pred)
@@ -2765,9 +2795,9 @@ def p_mean_variance(self, unet, x, t, image_embed, noise_scheduler, text_encodin
         return model_mean, posterior_variance, posterior_log_variance, x_start
 
     @torch.no_grad()
-    def p_sample(self, unet, x, t, image_embed, noise_scheduler, text_encodings = None, cond_scale = 1., lowres_cond_img = None, self_cond = None, predict_x_start = False, learned_variance = False, clip_denoised = True, lowres_noise_level = None):
+    def p_sample(self, unet, x, t, image_embed, noise_scheduler, text_encodings = None, cond_scale = 1., lowres_cond_img = None, self_cond = None, predict_x_start = False, predict_v = False, learned_variance = False, clip_denoised = True, lowres_noise_level = None):
         b, *_, device = *x.shape, x.device
-        model_mean, _, model_log_variance, x_start = self.p_mean_variance(unet, x = x, t = t, image_embed = image_embed, text_encodings = text_encodings, cond_scale = cond_scale, lowres_cond_img = lowres_cond_img, self_cond = self_cond, clip_denoised = clip_denoised, predict_x_start = predict_x_start, noise_scheduler = noise_scheduler, learned_variance = learned_variance, lowres_noise_level = lowres_noise_level)
+        model_mean, _, model_log_variance, x_start = self.p_mean_variance(unet, x = x, t = t, image_embed = image_embed, text_encodings = text_encodings, cond_scale = cond_scale, lowres_cond_img = lowres_cond_img, self_cond = self_cond, clip_denoised = clip_denoised, predict_x_start = predict_x_start, predict_v = predict_v, noise_scheduler = noise_scheduler, learned_variance = learned_variance, lowres_noise_level = lowres_noise_level)
         noise = torch.randn_like(x)
         # no noise when t == 0
         nonzero_mask = (1 - (t == 0).float()).reshape(b, *((1,) * (len(x.shape) - 1)))
@@ -2782,6 +2812,7 @@ def p_sample_loop_ddpm(
         image_embed,
         noise_scheduler,
         predict_x_start = False,
+        predict_v = False,
         learned_variance = False,
         clip_denoised = True,
         lowres_cond_img = None,
@@ -2840,6 +2871,7 @@ def p_sample_loop_ddpm(
                     lowres_cond_img = lowres_cond_img,
                     lowres_noise_level = lowres_noise_level,
                     predict_x_start = predict_x_start,
+                    predict_v = predict_v,
                     noise_scheduler = noise_scheduler,
                     learned_variance = learned_variance,
                     clip_denoised = clip_denoised
@@ -2865,6 +2897,7 @@ def p_sample_loop_ddim(
         timesteps,
         eta = 1.,
         predict_x_start = False,
+        predict_v = False,
         learned_variance = False,
         clip_denoised = True,
         lowres_cond_img = None,
@@ -2926,7 +2959,9 @@ def p_sample_loop_ddim(
 
                 # predict x0
 
-                if predict_x_start:
+                if predict_v:
+                    x_start = noise_scheduler.predict_start_from_v(img, t = time_cond, v = pred)
+                elif predict_x_start:
                     x_start = pred
                 else:
                     x_start = noise_scheduler.predict_start_from_noise(img, t = time_cond, noise = pred)
@@ -2938,8 +2973,8 @@ def p_sample_loop_ddim(
 
                 # predict noise
 
-                if predict_x_start:
-                    pred_noise = noise_scheduler.predict_noise_from_start(img, t = time_cond, x0 = pred)
+                if predict_x_start or predict_v:
+                    pred_noise = noise_scheduler.predict_noise_from_start(img, t = time_cond, x0 = x_start)
                 else:
                     pred_noise = pred
 
@@ -2975,7 +3010,7 @@ def p_sample_loop(self, *args, noise_scheduler, timesteps = None, **kwargs):
 
         return self.p_sample_loop_ddim(*args, noise_scheduler = noise_scheduler, timesteps = timesteps, **kwargs)
 
-    def p_losses(self, unet, x_start, times, *, image_embed, noise_scheduler, lowres_cond_img = None, text_encodings = None, predict_x_start = False, noise = None, learned_variance = False, clip_denoised = False, is_latent_diffusion = False, lowres_noise_level = None):
+    def p_losses(self, unet, x_start, times, *, image_embed, noise_scheduler, lowres_cond_img = None, text_encodings = None, predict_x_start = False, predict_v = False, noise = None, learned_variance = False, clip_denoised = False, is_latent_diffusion = False, lowres_noise_level = None):
         noise = default(noise, lambda: torch.randn_like(x_start))
 
         # normalize to [-1, 1]
@@ -3020,7 +3055,12 @@ def p_losses(self, unet, x_start, times, *, image_embed, noise_scheduler, lowres
 
         pred, _ = self.parse_unet_output(learned_variance, unet_output)
 
-        target = noise if not predict_x_start else x_start
+        if predict_v:
+            target = noise_scheduler.calculate_v(x_start, times, noise)
+        elif predict_x_start:
+            target = x_start
+        else:
+            target = noise
 
         loss = noise_scheduler.loss_fn(pred, target, reduction = 'none')
         loss = reduce(loss, 'b ... -> b (...)', 'mean')
@@ -3106,7 +3146,7 @@ def sample(
         num_unets = self.num_unets
         cond_scale = cast_tuple(cond_scale, num_unets)
 
-        for unet_number, unet, vae, channel, image_size, predict_x_start, learned_variance, noise_scheduler, lowres_cond, sample_timesteps, unet_cond_scale in tqdm(zip(range(1, num_unets + 1), self.unets, self.vaes, self.sample_channels, self.image_sizes, self.predict_x_start, self.learned_variance, self.noise_schedulers, self.lowres_conds, self.sample_timesteps, cond_scale)):
+        for unet_number, unet, vae, channel, image_size, predict_x_start, predict_v, learned_variance, noise_scheduler, lowres_cond, sample_timesteps, unet_cond_scale in tqdm(zip(range(1, num_unets + 1), self.unets, self.vaes, self.sample_channels, self.image_sizes, self.predict_x_start, self.predict_v, self.learned_variance, self.noise_schedulers, self.lowres_conds, self.sample_timesteps, cond_scale)):
             if unet_number < start_at_unet_number:
                 continue  # It's the easiest way to do it
 
@@ -3142,6 +3182,7 @@ def sample(
                     text_encodings = text_encodings,
                     cond_scale = unet_cond_scale,
                     predict_x_start = predict_x_start,
+                    predict_v = predict_v,
                     learned_variance = learned_variance,
                     clip_denoised = not is_latent_diffusion,
                     lowres_cond_img = lowres_cond_img,
@@ -3181,6 +3222,7 @@ def forward(
         lowres_conditioner  = self.lowres_conds[unet_index]
         target_image_size   = self.image_sizes[unet_index]
         predict_x_start     = self.predict_x_start[unet_index]
+        predict_v           = self.predict_v[unet_index]
         random_crop_size    = self.random_crop_sizes[unet_index]
         learned_variance    = self.learned_variance[unet_index]
         b, c, h, w, device, = *image.shape, image.device
@@ -3219,7 +3261,7 @@ def forward(
             image = vae.encode(image)
             lowres_cond_img = maybe(vae.encode)(lowres_cond_img)
 
-        losses = self.p_losses(unet, image, times, image_embed = image_embed, text_encodings = text_encodings, lowres_cond_img = lowres_cond_img, predict_x_start = predict_x_start, learned_variance = learned_variance, is_latent_diffusion = is_latent_diffusion, noise_scheduler = noise_scheduler, lowres_noise_level = lowres_noise_level)
+        losses = self.p_losses(unet, image, times, image_embed = image_embed, text_encodings = text_encodings, lowres_cond_img = lowres_cond_img, predict_x_start = predict_x_start, predict_v = predict_v, learned_variance = learned_variance, is_latent_diffusion = is_latent_diffusion, noise_scheduler = noise_scheduler, lowres_noise_level = lowres_noise_level)
 
         if not return_lowres_cond_image:
             return losses

dalle2_pytorch/version.py

@@ -1 +1 @@
-__version__ = '1.10.9'
+__version__ = '1.11.1'