import numpy as np
from tensorboardX import SummaryWriter
writer=SummaryWriter(log_dir="scala",comment="base_scala")
for epoch in range(100):
writer.add_scala('scala/test',np.random.rand(),epoch)
writer.add_scalars("scala/scalars_test",{"xsinx":epoch *np.sin(epoch),"xcosx":epoch*np.cos(epoch)},epoch)
writer.close()
1. from tensorboardXimport SummaryWriter
2. Then define a SummaryWriter () instance.
Parameter SummaryWriter () is: def __init __ (self, log_dir = "scala", comment = "base_scala", ** kwargs):
Log_dir file which is generated by the release of the directory,
comment for the file name.
The default directory folder directory to generate runs .
3. writer.add_scalar('scalar/test', np.random.rand(), epoch)
Data type scalar type
The first argument can be simply understood as the name of the saved map
The second parameter is the Y-axis data to be understood that
The third parameter to be understood that the X-axis data
When the data is more than one Y-axis, may be used writer.add_scalars ()
4. If the default directory, tensorboard --logdir runs
In the present embodiment, tensorboard --logdir scalar
5. Finally, call writer.close ()