Python画train和val的loss

最新推荐文章于 2024-06-29 16:53:05 发布

lct不吃香菜

最新推荐文章于 2024-06-29 16:53:05 发布

阅读量327

点赞数

文章标签： python 机器学习开发语言

本文链接：https://blog.csdn.net/m0_46507285/article/details/132940365

版权

import argparse
import matplotlib.pyplot as plt
import pandas as pd


def main():
    # 在终端中使用指定路径
    # parser = argparse.ArgumentParser(description='Plot Logs')
    # parser.add_argument('-path', type=str, required=True, help="Path to '.csv' file.")
    # df = pd.read_csv(parser.parse_args().path)

    path = 'train_records.csv'
    df = pd.read_csv(path)

    epoch = df['epoch']  # 读取cvs中的epoch
    train_loss = df['train_loss\t']  # 读取cvs中的train_loss下的值
    val_loss = df['val_loss\t']  # 读取cvs中的val_loss下的值

    # 画三个图
    # psnr = df['psnr\t']
    # ssim = df['ssim\t']
    # fig, axs = plt.subplots(1, 3, figsize=(15, 3.5))
    #
    # axs[0].plot(epoch, train_loss, label='train_loss')
    # axs[0].plot(epoch, val_loss, label='val_loss')
    # axs[0].set(xlabel='Epoch', ylabel='Loss',
    #            title='Train (Val) Loss Analysis')
    # axs[0].legend()
    #
    # axs[1].plot(epoch, psnr)
    # axs[1].set(xlabel='Epoch', ylabel='PSNR (dB)',
    #            title='PSNR Analysis')
    #
    # axs[2].plot(epoch, ssim)
    # axs[2].set(xlabel='Epoch', ylabel='SSIM',
    #            title='SSIM Analysis')

    # 画一个图，自适应图的大小
    fig, axs = plt.subplots(1, 1)
    axs.plot(epoch, train_loss, label='train_loss')
    axs.plot(epoch, val_loss, label='val_loss')
    axs.set(xlabel='Epoch', ylabel='Loss',
            title='Train (Val) Loss Analysis')
    axs.legend()

    plt.tight_layout()
    plt.show()


if __name__ == '__main__':
    main()