无法修复此错误：ValueError：数据基数不明确

如何解决无法修复此错误：ValueError：数据基数不明确

我正在构建一个简单的序列分类模型，但我有一个错误，我找不到 :(

ValueError: Data cardinality is ambiguous:
  x sizes: 21,21,21
  y sizes: 615

我正在从 2 个目录 Positive 和 Negative 中的文本文件中读取序列，并准备将数据发送到模型进行学习，但可以使其工作。

Python 程序在这里失败：

history = model.fit(X_train,y_train,validation_data=(X_test,y_test),epochs=10,batch_size=64)

我的猜测是它来自 one_hot 函数。这一步以前做过很多次，想利用Keras的内部功能来简化一下。

# Get all sequences from raw data
poz_files = GetTextFilePathsInDirectory(f"{base_directory}/_POZ")
sequences_poz = []
for i in range(0,len(poz_files)):
    sequences_poz.extend(GetLinesFromTextFile(poz_files[i]))
label_poz = np.ones(len(sequences_poz))

neg_files = GetTextFilePathsInDirectory(f"{base_directory}/_NEG")
sequences_neg = []
for i in range(0,len(neg_files)):
    sequences_neg.extend(GetLinesFromTextFile(neg_files[i]))
label_neg = np.zeros(len(sequences_neg))

# Assign labels
sequences = []
labels    = []
sequences.extend(sequences_poz)
sequences.extend(sequences_neg)
labels.extend(label_poz)
labels.extend(label_neg)

# get info out of our sequence list
max_sequence_length,vocab_size = getUniqueWordsSentenceLength(sequences) # fit to match exact number of unique words in sequence

# transform sequences into vectors
encoded_sequences = [one_hot(s,vocab_size) for s in sequences] # use vocab x 4 - much larger than needed to reduce the probability of collisions from the hash function
padded_sequences  = pad_sequences(encoded_sequences,maxlen=max_sequence_length,padding='pre') # padd sequences so they are all the same length

# label sequences
labeled_sequences = list(zip(padded_sequences,labels))

# split train and test data
train,test = train_test_split(labeled_sequences,test_size=0.05,shuffle=True)
print(test)
X_train,y_train = list(zip(*train))
X_test,y_test = list(zip(*test))
y_train = np.array(y_train)
y_test = np.array(y_test)

model = Sequential()
model.add(Embedding(vocab_size,2,input_length=max_sequence_length)) #The model is a simple binary classification model. Importantly,the output from the Embedding layer will be 4 vectors of 8 dimensions each,one for each word.
model.add(LSTM(1000))
model.add(Dense(1,activation='sigmoid'))
model.compile(loss='binary_crossentropy',optimizer='adam',metrics=['accuracy'])
print(model.summary())
history = model.fit(X_train,batch_size=64)

输出

print(test)

如下：

[(array([7,4,6,7,1,4],dtype=int32),1.0),(array([0,3,0.0),(array([7,(array([3,1.0)]