Update app.py

app.py

@@ -377,19 +377,22 @@ LORA_NAME = "ohami_filter_autorun"
 
 # ========== HELPERS ==========
 def create_dataset(images, *captions):
-    destination_folder = f"datasets_{uuid.uuid4()}"
-    os.makedirs(destination_folder, exist_ok=True)
-
-    jsonl_file_path = os.path.join(destination_folder, "metadata.jsonl")
-    with open(jsonl_file_path, "a") as jsonl_file:
-        for index, image in enumerate(images):
-            new_image_path = shutil.copy(str(image), destination_folder)
-            caption = captions[index]
-            file_name = os.path.basename(new_image_path)
-            data = {"file_name": file_name, "prompt": caption}
-            jsonl_file.write(json.dumps(data) + "\n")
-
-    return destination_folder
+    if len(images) != len(captions):
+        raise ValueError("Number of images and captions must be the same.")
+
+    destination_folder = Path(f"/tmp/datasets_{uuid.uuid4()}")
+    destination_folder.mkdir(parents=True, exist_ok=True)
+
+    jsonl_file_path = destination_folder / "metadata.jsonl"
+
+    with jsonl_file_path.open("a", encoding="utf-8") as jsonl_file:
+        for image_path, caption in zip(images, captions):
+            new_image_path = shutil.copy(str(image_path), destination_folder)
+            file_name = Path(new_image_path).name
+            entry = {"file_name": file_name, "prompt": caption}
+            jsonl_file.write(json.dumps(entry, ensure_ascii=False) + "\n")
+
+    return str(destination_folder)
 
 def recursive_update(d, u):
     for k, v in u.items():