up

Author: patrickvonplaten

models/vision/ddpm/example.py

@@ -1,20 +1,26 @@
 #!/usr/bin/env python3
 import tempfile
 import sys
-
+import os
+import pathlib
 from modeling_ddpm import DDPM
+import PIL.Image
+import numpy as np
 
-model_id = sys.argv[1]
+model_ids = ["ddpm-lsun-cat", "ddpm-lsun-cat-ema", "ddpm-lsun-church-ema", "ddpm-lsun-church", "ddpm-lsun-bedroom", "ddpm-lsun-bedroom-ema", "ddpm-cifar10-ema", "ddpm-lsun-cifar10", "ddpm-lsun-celeba-hq", "ddpm-lsun-celeba-hq-ema"]
 
-ddpm = DDPM.from_pretrained(model_id)
-image = ddpm()
+for model_id in model_ids:
 
-import PIL.Image
-import numpy as np
-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")
+    path = os.path.join("/home/patrick/images/hf", model_id)
+    pathlib.Path(path).mkdir(parents=True, exist_ok=True)
+
+    ddpm = DDPM.from_pretrained("fusing/" + model_id)
+    image = ddpm(batch_size=4)
+
+    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)
 
-import ipdb; ipdb.set_trace()
+    for i in range(image_processed.shape[0]):
+        image_pil = PIL.Image.fromarray(image_processed[i])
+        image_pil.save(os.path.join(path, f"image_{i}.png"))

models/vision/ddpm/modeling_ddpm.py

@@ -33,7 +33,7 @@ def __call__(self, batch_size=1, generator=None, torch_device=None):
 
         self.unet.to(torch_device)
         # 1. Sample gaussian noise
-        image = self.noise_scheduler.sample_noise((1, self.unet.in_channels, self.unet.resolution, self.unet.resolution), device=torch_device, generator=generator)
+        image = self.noise_scheduler.sample_noise((batch_size, self.unet.in_channels, self.unet.resolution, self.unet.resolution), device=torch_device, generator=generator)
         for t in tqdm.tqdm(reversed(range(len(self.noise_scheduler))), total=len(self.noise_scheduler)):
             # i) define coefficients for time step t
             clip_image_coeff = 1 / torch.sqrt(self.noise_scheduler.get_alpha_prod(t))

src/diffusers/schedulers/gaussian_ddpm.py

@@ -108,7 +108,7 @@ def get_alpha_prod(self, time_step):
 
     def sample_variance(self, time_step, shape, device, generator=None):
         variance = self.log_variance[time_step]
-        nonzero_mask = torch.tensor([1 - (time_step == 0)], device=device).float()[None, :].repeat(shape[0], 1)
+        nonzero_mask = torch.tensor([1 - (time_step == 0)], device=device).float()[None, :]
 
         noise = self.sample_noise(shape, device=device, generator=generator)
 

tests/test_modeling_utils.py

@@ -76,7 +76,7 @@ def floats_tensor(shape, scale=1.0, rng=None, name=None):
 class ModelTesterMixin(unittest.TestCase):
     @property
     def dummy_input(self):
-        batch_size = 1
+        batch_size = 4
         num_channels = 3
         sizes = (32, 32)