Trying to run step 2 of V7.0 Results in following error

[IdsQme]

ValueError Traceback (most recent call last)
Cell In[4], line 12
1 from act_train_rhythm import *;
3 train_params_p2 = {
4 "divisor" : 4,
5 "train_epochs" : 16,
(...)
10 "data_split_count" : 80
11 };
---> 12 model_p2 = step2_build_model()

File ~\Downloads\osumapper-master\v7.0\act_train_rhythm.py:234, in step2_build_model()
233 def step2_build_model():
--> 234 model_v7 = build_model();
235 return model_v7

File ~\Downloads\osumapper-master\v7.0\act_train_rhythm.py:192, in build_model()
190 plt.figure()
191 plt.xlabel('Epoch')
--> 192 plt.ylabel('Mean Abs Error [Limitless]')
193 plt.plot(history.epoch, np.array(history.history['mean_absolute_error']),
194 label='Train MAE')
195 plt.plot(history.epoch, np.array(history.history['val_mean_absolute_error']),
196 label = 'Val MAE')

File ~\anaconda3\Lib\site-packages\keras\src\ops\operation.py:266, in Operation.output(self)
256 @Property
257 def output(self):
258 """Retrieves the output tensor(s) of a layer.
259
260 Only returns the tensor(s) corresponding to the first time
(...)
264 Output tensor or list of output tensors.
265 """
--> 266 return self._get_node_attribute_at_index(0, "output_tensors", "output")

File ~\anaconda3\Lib\site-packages\keras\src\ops\operation.py:285, in Operation._get_node_attribute_at_index(self, node_index, attr, attr_name)
269 """Private utility to retrieves an attribute (e.g. inputs) from a node.
270
271 This is used to implement the properties:
(...)
282 The operation's attribute attr at the node of index node_index.
283 """
284 if not self._inbound_nodes:
--> 285 raise ValueError(
286 f"The layer {self.name} has never been called "
287 f"and thus has no defined {attr_name}."
288 )
289 if not len(self._inbound_nodes) > node_index:
290 raise ValueError(
291 f"Asked to get {attr_name} at node "
292 f"{node_index}, but the operation has only "
293 f"{len(self._inbound_nodes)} inbound nodes."
294 )

ValueError: The layer sequential_2 has never been called and thus has no defined output.

[dragongoeseasy]

same situation but solved by gpt：
original：
`def build_model():
"""
Build the model.
Two inputs for wav_data and div_data (metadata) respectively.
Hyperparameters in the middle are tuned to make sure it runs smoothly on my machine.
You can try changing them in the middle if it achieves better result.
"""
train_shape, div_shape, label_shape = get_data_shape();
model1 = keras.Sequential([
keras.layers.TimeDistributed(keras.layers.Conv2D(16, (2, 2),
data_format='channels_last'),
input_shape=(time_interval, train_shape[1], train_shape[2], train_shape[3])),
keras.layers.TimeDistributed(keras.layers.MaxPool2D((1, 2),
data_format='channels_last')),
keras.layers.TimeDistributed(keras.layers.Activation(activation=tf.nn.relu)),
keras.layers.TimeDistributed(keras.layers.Dropout(0.3)),
keras.layers.TimeDistributed(keras.layers.Conv2D(16, (2, 3),
data_format='channels_last')),
keras.layers.TimeDistributed(keras.layers.MaxPool2D((1, 2),
data_format='channels_last')),
keras.layers.TimeDistributed(keras.layers.Activation(activation=tf.nn.relu)),
keras.layers.TimeDistributed(keras.layers.Dropout(0.3)),
keras.layers.TimeDistributed(keras.layers.Flatten()),
keras.layers.LSTM(64, activation=tf.nn.tanh, return_sequences=True)
])

input2 = keras.layers.InputLayer(input_shape=(time_interval, div_shape[1]));

conc = keras.layers.concatenate([model1.output, input2.output]);
dense1 = keras.layers.Dense(71, activation=tf.nn.tanh)(conc);
dense2 = keras.layers.Dense(71, activation=tf.nn.relu)(dense1);
dense3 = keras.layers.Dense(label_shape[1], activation=tf.nn.tanh)(dense2);


# I think the first is correct but whatever...
try:
    optimizer = tf.optimizers.RMSprop(0.001) #Adamoptimizer?
except:
    optimizer = tf.train.RMSPropOptimizer(0.001) #Adamoptimizer?


final_model = Model(inputs=[model1.input, input2.input], outputs=dense3);
final_model.compile(loss='mse',
            optimizer=optimizer,
            metrics=[keras.metrics.mae])
return final_model`

after modifed：
`def build_model():
"""
Build the model.
Two inputs for wav_data and div_data (metadata) respectively.
Hyperparameters in the middle are tuned to make sure it runs smoothly on my machine.
You can try changing them in the middle if it achieves better result.
"""
train_shape, div_shape, label_shape = get_data_shape();
model1 = keras.Sequential([
keras.layers.TimeDistributed(keras.layers.Conv2D(16, (2, 2),
data_format='channels_last'),
input_shape=(time_interval, train_shape[1], train_shape[2], train_shape[3])),
keras.layers.TimeDistributed(keras.layers.MaxPool2D((1, 2),
data_format='channels_last')),
keras.layers.TimeDistributed(keras.layers.Activation(activation=tf.nn.relu)),
keras.layers.TimeDistributed(keras.layers.Dropout(0.3)),
keras.layers.TimeDistributed(keras.layers.Conv2D(16, (2, 3),
data_format='channels_last')),
keras.layers.TimeDistributed(keras.layers.MaxPool2D((1, 2),
data_format='channels_last')),
keras.layers.TimeDistributed(keras.layers.Activation(activation=tf.nn.relu)),
keras.layers.TimeDistributed(keras.layers.Dropout(0.3)),
keras.layers.TimeDistributed(keras.layers.Flatten()),
keras.layers.LSTM(64, activation=tf.nn.tanh, return_sequences=True)
])
# 修改点1：改用 Input 创建输入张量
input1 = keras.Input(shape=(time_interval, train_shape[1], train_shape[2], train_shape[3])) # 新增
input2 = keras.Input(shape=(time_interval, div_shape[1])) # 替换 InputLayer

# 修改点2：调用 model1 获取输出
x1 = model1(input1)  # 关键调用

# 修改点3：直接使用张量而非 .output
conc = keras.layers.concatenate([x1, input2]);  # 使用张量对象
dense1 = keras.layers.Dense(71, activation=tf.nn.tanh)(conc);
dense2 = keras.layers.Dense(71, activation=tf.nn.relu)(dense1);
dense3 = keras.layers.Dense(label_shape[1], activation=tf.nn.tanh)(dense2);

# I think the first is correct but whatever...
try:
    optimizer = tf.optimizers.RMSprop(0.001) #Adamoptimizer?
except:
    optimizer = tf.train.RMSPropOptimizer(0.001) #Adamoptimizer?

# 修改点4：
final_model = keras.Model(inputs=[input1, input2], outputs=dense3);
final_model.compile(loss='mse',
            optimizer=optimizer,
            metrics=[keras.metrics.mae])
return final_model

`
The core problem is that model1 is not called directly and its .output attribute is accessed
Explicitly call the model through model1(input1) to build the computational graph and thus have a valid .output
Hope useful