Removed clip dependency

app.py

@@ -5,7 +5,6 @@ import torchvision.transforms as transforms
 from dalle.models import StoryDalle
 import argparse
 from PIL import Image
-import numpy as np
 from torchvision.utils import save_image
 import tensorflow_hub as hub
 import gdown

dalle/utils/utils.py

@@ -10,12 +10,8 @@ import urllib
 import hashlib
 import tarfile
 import torch
-import clip
 import numpy as np
-from PIL import Image
-from torch.nn import functional as F
 from tqdm import tqdm
-import torchvision.utils as vutils
 import matplotlib.pyplot as plt
 
 
@@ -26,25 +22,6 @@ def set_seed(seed: int):
     torch.cuda.manual_seed_all(seed)
 
 
-@torch.no_grad()
-def clip_score(prompt: str,
-               images: np.ndarray,
-               model_clip: torch.nn.Module,
-               preprocess_clip,
-               device: str) -> np.ndarray:
-    images = [preprocess_clip(Image.fromarray((image*255).astype(np.uint8))) for image in images]
-    images = torch.stack(images, dim=0).to(device=device)
-    texts = clip.tokenize(prompt).to(device=device)
-    texts = torch.repeat_interleave(texts, images.shape[0], dim=0)
-
-    image_features = model_clip.encode_image(images)
-    text_features = model_clip.encode_text(texts)
-
-    scores = F.cosine_similarity(image_features, text_features).squeeze()
-    rank = torch.argsort(scores, descending=True).cpu().numpy()
-    return rank
-
-
 def download(url: str, root: str) -> str:
     os.makedirs(root, exist_ok=True)
     filename = os.path.basename(url)
@@ -128,4 +105,4 @@ def save_image(ground_truth, images, out_dir, batch_idx):
     #             fid.write(texts[i][idx] + '\n')
     #         fid.write('\n\n')
     #     fid.close()
-    return
+    return