Tensorflow2.1报错 ProfilerNotRunningError: Cannot stop profiling. No profiler is running.

最近学习tensorflow与案例代码一致运行却不通过

Tensorflow2.1在创建TensorBoard callback时，日志输出文件采用./callbacks这种写法时间会报出tensorflow.python.eager.profiler.ProfilerNotRunningError: Cannot stop profiling. No profiler is running.

 

logdir = './callbacks'  # 注意，这里有个bug，不能使用./callbacks这种形式表达，可能与tensorflow内部实现有关
if not os.path.exists(logdir):
    os.mkdir(logdir)
output_model_file = os.path.join(logdir, "mnist_model.h5")  # 定义输出的model文件，并放入logdir文件夹
callbacks = [
    keras.callbacks.TensorBoard(logdir), #tensorboard出错
    keras.callbacks.ModelCheckpoint(output_model_file, save_best_only=True),
    keras.callbacks.EarlyStopping(patience=5, min_delta=1e-3)
]

当采用logdir = ”callbacks“这种不含有./这种windows寻找当前根目录写法时就可以顺利运行，可能是默认在当前根目录下吧

运行的非常顺利