Merge branch 'main' into releasev1.0.3

projects/prompt_to_prompt/models/seq_aligner.py

@@ -34,16 +34,16 @@ def mis_match_char(self, x, y):
 
 def get_matrix(size_x, size_y, gap):
     matrix = []
-    for i in range(len(size_x) + 1):
+    for i in range(size_x + 1):
         sub_matrix = []
-        for j in range(len(size_y) + 1):
+        for j in range(size_y + 1):
             sub_matrix.append(0)
         matrix.append(sub_matrix)
-    for j in range(1, len(size_y) + 1):
+    for j in range(1, size_y + 1):
         matrix[0][j] = j * gap
-    for i in range(1, len(size_x) + 1):
+    for i in range(1, size_x + 1):
         matrix[i][0] = i * gap
-    return matrix
+    return np.array(matrix)
 
 
 # def get_matrix(size_x, size_y, gap):

projects/prompt_to_prompt/visualize.ipynb

@@ -98,10 +98,11 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "#### 1. VAE reconstruction as the a standard result\n"
+    "#### 1. VAE reconstruction as the standard result\n"
    ]
   },
   {
@@ -123,11 +124,12 @@
    "source": [
     "from models import ptp_utils\n",
     "import torch\n",
+    "import mmcv\n",
     "\n",
-    "image_path = 'projects/prompt_to_prompt/assets/gnochi_mirror.jpeg'\n",
+    "image = mmcv.imread('https://github.com/open-mmlab/mmagic/assets/12782558/c851dcf8-bcfa-424f-9eeb-d01a48ea7127', channel_order='rgb')\n",
     "prompt = \"a cat sitting next to a mirror\"\n",
     "# using vae to reconstruct the image\n",
-    "image_tensor = ptp_utils.load_512(image_path).to('cuda')\n",
+    "image_tensor = ptp_utils.load_512(image).to('cuda')\n",
     "vae = StableDiffuser.vae\n",
     "with torch.no_grad():\n",
     "    vae_rec = vae(image_tensor, return_dict=False)[0]\n",