TensorBoard ValueError: Expected scalar shape, saw shape: (1,)

Question

When I use the callback function TensorBoard as the following:

skip_training = False

tensorboad_cb = TensorBoard('logs')

def train_model(model, callbacks_list):
    '''
    Input: 
        Model and callback list,
    Return: 
        Model with best-checkpoint weights.
    '''
    ## TYPE YOUR CODE for task 10 here:
    #history = model.fit(X_tr, y_tr, batch_size=4096, epochs=20, verbose=1, validation_data=(X_va, y_va), callbacks=[callbacks_list])
    if not skip_training:
      history = model.fit(X_tr, y_tr, batch_size=4096, epochs=1, verbose=1, callbacks=[tensorboad_cb])
      model.save(checkpoint_name)
    else:
      model = load_model(checkpoint_name)
    return model

model = train_model(model, callbacks_list)

I get this error:

287/287 [==============================] - 180s 626ms/step - loss: 0.1103 - f1_score: 0.6010 - acc: 0.9565 --------------------------------------------------------------------------- ValueError Traceback (most recent call last) in () 17 return model 18 ---> 19 model = train_model(model, callbacks_list) 20 21 #history = model.fit(train_data, epochs=15, validation_data=valid_data, validation_steps=50, callbacks=[checkpoint_callback])

9 frames /usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/check_ops.py in assert_scalar(tensor, name, message) 2282 if context.executing_eagerly(): 2283 raise ValueError('%sExpected scalar shape, saw shape: %s.' -> 2284 % (message or '', shape,)) 2285 else: 2286 raise ValueError('%sExpected scalar shape for %s, saw shape: %s.'

ValueError: Expected scalar shape, saw shape: (1,).

Answer 1

I have same issue on tensorflow 2.6.2, but on tensorflow 2.4.1 it works well.

finally, I fixed it by comment the code:

comment tensorboard code