VQA_model/PickScore-main/fix_numpy_issue.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

"""
修复numpy和transformers兼容性问题
"""

import subprocess
import sys

def run_command(command):
    """运行命令并打印输出"""
    print(f"执行: {command}")
    result = subprocess.run(command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
    if result.stdout:
        print(result.stdout.decode('utf-8'))
    if result.stderr:
        print(f"错误: {result.stderr.decode('utf-8')}")
    return result.returncode == 0

def main():
    """修复环境问题"""
    print("开始修复环境问题...")
    
    # 第1步：卸载当前的numpy和transformers
    print("步骤1: 卸载当前的numpy和transformers")
    run_command("pip uninstall -y numpy transformers")
    
    # 第2步：安装兼容的numpy版本
    print("步骤2: 安装兼容的numpy版本")
    run_command("pip install numpy==1.23.5")
    
    # 第3步：安装兼容的transformers版本
    print("步骤3: 安装兼容的transformers版本")
    run_command("pip install transformers==4.26.1")
    
    # 第4步：安装pydantic v1版本
    print("步骤4: 安装pydantic v1版本")
    run_command("pip uninstall -y pydantic")
    run_command("pip install pydantic==1.10.8")
    
    # 第5步：安装CLIP模型
    print("步骤5: 安装CLIP模型")
    run_command("pip install ftfy regex")
    run_command("pip install git+https://github.com/openai/CLIP.git")
    
    print("\n环境修复完成!")
    print("请使用以下命令重新运行你的脚本:")
    print("python batch_process.py --txt_path /home/wangjuntong/VQA_model/dataset/video_editing_score/video_quality_with_prompts.txt --output_path /home/wangjuntong/VQA_model/PickScore-main/output")

if __name__ == "__main__":
    main()