Python Tensorboard中存在错误:TypeError:不支持的操作数类型为-:';str';和';int';
我正在尝试使用TensorBoard在模型训练期间创建洞察。通过它,我想检查我的模型是如何训练的,哪个epoch提供了最好的结果,等等。但是,我遇到了以下错误:Python Tensorboard中存在错误:TypeError:不支持的操作数类型为-:';str';和';int';,python,tensorflow,tensorboard,Python,Tensorflow,Tensorboard,我正在尝试使用TensorBoard在模型训练期间创建洞察。通过它,我想检查我的模型是如何训练的,哪个epoch提供了最好的结果,等等。但是,我遇到了以下错误: TypeError Traceback (most recent call last) <ipython-input-33-e7c9b8bd2922> in <module> 18 model.fit([x_train['input1'],
TypeError Traceback (most recent call last)
<ipython-input-33-e7c9b8bd2922> in <module>
18 model.fit([x_train['input1'], x_train['input2']], y_train, batch_size = 32, epochs = 10, verbose = 0,
---> 19 callbacks = [tf_board])
20
21
对于迟来的回答,很抱歉,你可能已经解决了你的问题,但我在这里发布给未来和我有同样问题的人 您只需将Tensorflow、TensorBoard(和TensorBoard插件配置文件)更新为版本
=2.2.0
,如中所述。Tensorboard回调在以前的版本中可能需要不同的参数。尝试使用此(5,10)
,而不是此'5,10'
from tensorflow.keras.callbacks import TensorBoard
import datetime # optional
# Important!
%load_ext tensorboard
%reload_ext tensorboard
# Pass a directory to save logs
log_dir = '/home/pedro/MyDrive/Transfer_Learning/Models/logs'
# Let's create our callback
# TensorBoard(directory, histogram_freq, update_freq, 5)
# -> str(range(profile_batch)) or str(profile_batch)
tf_board = TensorBoard(log_dir, histogram_freq=8, update_freq='batch', profile_batch='5, 10')
# And fit our callback inside our training data
model.fit([x_train['input1'], x_train['input2']], y_train, batch_size = 32, epochs = 10, verbose = 0,
callbacks = [tf_board])
# Use this magic function to visualize
%tensorboard --logdir '/home/pedro/MyDrive/Transfer_Learning/Models/logs'