The solution to the display disorder of several events files in the tensorboard

In the use of tensorboard When visualizing network parameters , When multiple events After the document , stay tensorboard The display will be very chaotic , Very difficult to observe . As shown in the figure below , This picture is a document of only two experiments , It's already confused .

My previous solution is to delete the previous files every time I run , Ensure that only the currently generated events file , But this is cumbersome , Especially when more parameters need to be adjusted , Each deletion is not only tiring , And it's easy to make mistakes , Later, inspired by this article , Since they are saved in different paths, they can be displayed separately , Then every time you generate events Save the files in different paths , So I thought of using timestamp as path name , While solving the display disorder, the training time is also recorded , Kill two birds with one stone .

from datetime import datetimeTIMESTAMP = "{0:%Y-%m-%dT%H-%M-%S/}".format(datetime.now())...train_log_dir = 'logs/train/' + TIMESTAMPtest_log_dir = 'logs/test/'   + TIMESTAMPmegred = tf.summary.merge_all()with tf.Session() as sess:    writer_train = tf.summary.FileWriter(train_log_dir,sess.graph)    writer_test = tf.summary.FileWriter(test_log_dir)        ...other code...    writer_train.add_summary(summary_str_train,step)    writer_test.add_summary(summary_str_test,step)

The effect is as follows .