網(wǎng)絡(luò)上查找到兩種可視化方式：
其一：Visdom包(https://zhuanlan.zhihu.com/p/32025746)
其二：tensorboardx調(diào)用tensorboard(參考 http://www.itdecent.cn/p/46eb3004beca)
本文采用第二種方式，安裝文件

pip install tensorboardX
pip install tensorflow

測(cè)試代碼

'''
參考 http://www.itdecent.cn/p/46eb3004beca
'''
import numpy as np
from tensorboardX import SummaryWriter

writer = SummaryWriter()
for epoch in range(100):
    #print(epoch,epoch*np.sin(epoch),epoch*np.cos(epoch))
    writer.add_scalar('scalar/test',np.random.rand())
    writer.add_scalars('scalar/scalars_test',{'xsinx':epoch*np.sin(epoch),'xcosx':epoch*np.cos(epoch)},epoch)
writer.close()

目錄下生成文件位于runs文件夾下，調(diào)用

tensorboard --logdir runs

提示 TensorBoard 1.13.1 at http://DESKTOP-SM5RK2N:6006 (Press CTRL+C to quit)
但是本機(jī)輸入http://DESKTOP-SM5RK2N:6006 在瀏覽器中無法打開，參考https://blog.csdn.net/zhangkaixu321/article/details/79166089，訪問地址修改為

http://localhost:6006

成功訪問

tensorboardx訪問成功