AutoDL uses tensorboard

Table of contents

1. Training to form a log file

2. Switch the logs directory

3. Access TensorBoard in AutoPanel

---

1. Training to form a log file

example:

from torch.utils.tensorboard import SummaryWriter
import numpy as np

writer = SummaryWriter()
for x in range(1, 101) :
    writer.add_scalar('y = 2x', x, 2 * x) 
writer.close()
#单条曲线(scalar)
#add_scalar(tag, scalar_value, global_step=None, walltime=None)

#参数：

#tag ( string ) – 数据标识符
#scalar_value ( float或string/blobname ) – 要保存的值
#global_step ( int ) – 要记录的全局步长值
#walltime ( float ) – 记录训练的时间，默认 walltime (time.time()) 秒

2. Switch the logs directory

1. First end the Tensorboard process started by default and execute the command:

ps -ef | grep tensorboard | awk '{print $2}' | xargs kill -9

2. Execute the following command in the terminal to start TensorBoard

Officially, the event file of tensorboard needs to be saved to the /root/tf-logs/ path. If you don’t want to switch the save path, you only need to change the execution command.

/root/tf-logs/path

tensorboard --port 6007 --logdir tf-logs

other paths

tensorboard --port 6007 --logdir path/to/your/tf-logs/direction
#path/to/your/tf-logs/direction为你的logs文件前的路径，不需要带logs的文件名

3. Access TensorBoard in AutoPanel