Upload jsconv_cpm.py
Browse files- jsconv_cpm.py +31 -0
jsconv_cpm.py
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import json
|
| 2 |
+
|
| 3 |
+
# Путь к файлу JSON
|
| 4 |
+
json_file = 'd:/Dropbox/YandexDisk/Dataset/output.json'
|
| 5 |
+
|
| 6 |
+
# Чтение JSON из файла
|
| 7 |
+
with open(json_file, 'r') as f:
|
| 8 |
+
data = json.load(f)
|
| 9 |
+
|
| 10 |
+
# Создание нового списка словарей
|
| 11 |
+
new_data = []
|
| 12 |
+
|
| 13 |
+
for item in data:
|
| 14 |
+
conversations = item['conversations']
|
| 15 |
+
for conv in conversations:
|
| 16 |
+
if conv['from'] == 'user':
|
| 17 |
+
query = conv['value']
|
| 18 |
+
elif conv['from'] == 'assistant':
|
| 19 |
+
response = conv['value']
|
| 20 |
+
|
| 21 |
+
new_item = {
|
| 22 |
+
"query": query,
|
| 23 |
+
"response": response,
|
| 24 |
+
"images": item['image']
|
| 25 |
+
}
|
| 26 |
+
new_data.append(new_item)
|
| 27 |
+
|
| 28 |
+
# Запись нового списка словарей в файл
|
| 29 |
+
new_json_file = 'd:/Dropbox/YandexDisk/Dataset/new2_vl_data.json'
|
| 30 |
+
with open(new_json_file, 'w') as f:
|
| 31 |
+
json.dump(new_data, f, indent=4)
|