Beit - update numpy

beit/dall_e/__init__.py

@@ -12,7 +12,7 @@ def load_model(path: str, device: torch.device = None) -> nn.Module:
         resp.raise_for_status()
 
         with io.BytesIO(resp.content) as buf:
-            return torch.load(buf, map_location=device)
+            return torch.load(buf, map_location=device, weights_only=False)
     else:
         with open(path, 'rb') as f:
-            return torch.load(f, map_location=device)
+            return torch.load(f, map_location=device, weights_only=False)

beit/masking_generator.py

@@ -77,7 +77,7 @@ def _mask(self, mask, max_mask_patches):
         return delta
 
     def __call__(self):
-        mask = np.zeros(shape=self.get_shape(), dtype=np.int)
+        mask = np.zeros(shape=self.get_shape(), dtype=int)
         mask_count = 0
         while mask_count < self.num_masking_patches:
             max_mask_patches = self.num_masking_patches - mask_count

beit/semantic_segmentation/mmcv_custom/resize_transform.py

@@ -196,7 +196,7 @@ def _resize_img(self, results):
             img, w_scale, h_scale = mmcv.imresize(
                 results['img'], results['scale'], return_scale=True)
         scale_factor = np.array([w_scale, h_scale, w_scale, h_scale],
-                                dtype=np.float32)
+                                dtype=float)
         results['img'] = img
         results['img_shape'] = img.shape
         results['pad_shape'] = img.shape  # in case that there is no padding

beit2/masking_generator.py

@@ -77,7 +77,7 @@ def _mask(self, mask, max_mask_patches):
         return delta
 
     def __call__(self):
-        mask = np.zeros(shape=self.get_shape(), dtype=np.int32)
+        mask = np.zeros(shape=self.get_shape(), dtype=int)
         mask_count = 0
         while mask_count < self.num_masking_patches:
             max_mask_patches = self.num_masking_patches - mask_count

beit2/semantic_segmentation/mmcv_custom/resize_transform.py

@@ -196,7 +196,7 @@ def _resize_img(self, results):
             img, w_scale, h_scale = mmcv.imresize(
                 results['img'], results['scale'], return_scale=True)
         scale_factor = np.array([w_scale, h_scale, w_scale, h_scale],
-                                dtype=np.float32)
+                                dtype=float)
         results['img'] = img
         results['img_shape'] = img.shape
         results['pad_shape'] = img.shape  # in case that there is no padding

beit3/randaug.py

@@ -123,7 +123,7 @@ def brightness_func(img, factor):
     '''
         same output as PIL.ImageEnhance.Contrast
     '''
-    table = (np.arange(256, dtype=np.float32) * factor).clip(0, 255).astype(np.uint8)
+    table = (np.arange(256, dtype=float) * factor).clip(0, 255).astype(np.uint8)
     out = table[img]
     return out
 
@@ -133,7 +133,7 @@ def sharpness_func(img, factor):
     The differences the this result and PIL are all on the 4 boundaries, the center
     areas are same
     '''
-    kernel = np.ones((3, 3), dtype=np.float32)
+    kernel = np.ones((3, 3), dtype=float)
     kernel[1][1] = 5
     kernel /= 13
     degenerate = cv2.filter2D(img, -1, kernel)
@@ -142,8 +142,8 @@ def sharpness_func(img, factor):
     elif factor == 1.0:
         out = img
     else:
-        out = img.astype(np.float32)
-        degenerate = degenerate.astype(np.float32)[1:-1, 1:-1, :]
+        out = img.astype(float)
+        degenerate = degenerate.astype(float)[1:-1, 1:-1, :]
         out[1:-1, 1:-1, :] = degenerate + factor * (out[1:-1, 1:-1, :] - degenerate)
         out = out.astype(np.uint8)
     return out