jhgk

app.py

@@ -32,11 +32,13 @@ def model_inference(
     images, text,
 ):
     
-    print(type(images))
-    print(images[0])
-    images = Image.open(images[0][0])
+    # print(type(images))
+    # print(images[0])
+    # images = Image.open(images[0][0])
+    # print(images)
+    # print(type(images))
+    images = [{"type": "image", "image": Image.open(image[0])} for image in images]
     print(images)
-    print(type(images))
     # model = Qwen2VLForConditionalGeneration.from_pretrained(
     # "Qwen/Qwen2-VL-7B-Instruct", torch_dtype="auto", device_map="auto"
     # )
@@ -55,13 +57,7 @@ def model_inference(
     messages = [
         {
             "role": "user",
-            "content": [
-                {
-                    "type": "image",
-                    "image": images,
-                },
-                {"type": "text", "text": text},
-            ],
+            "content": images.append({"type": "text", "text": text}),
         }
     ]