utils/select_200k.py

import json
import random

with open(
    "/mnt/petrelfs/zhuchenglin/LLaVA/playground/data/llava_v1_5_mix665k.json"
) as f:
    data = json.load(f)

selected_data = random.sample(data, 200000)

# 将选择的元素写入新的JSON文件
with open(
    "/mnt/petrelfs/zhuchenglin/LLaVA/playground/data/llava_v1_5_mix200k.json",
    "w",
) as f:
    json.dump(selected_data, f)