ValueError:形状(无,1)和(无,64)是不兼容的序列模型

我正在根据文本的情感基调编写文本摘要。为了训练模型,我从 Here .这是我的密码

df = pd.read_csv("emotion-dataset.csv")
df['Clean_Text'] = df['Text'].apply(nfx.remove_userhandles)
df['Clean_Text'] = df['Clean_Text'].apply(nfx.remove_stopwords)

texts = df["Clean_Text"]
emotions = df["Emotion"]


# Step 3: Split the dataset into training and testing sets
X_train, X_test, y_train, y_test = train_test_split(padded_sequences, emotions, test_size=0.2)


model = Sequential()
model.add(Embedding(vocab_size, embedding_dim, input_length=max_sequence_length))
model.add(LSTM(hidden_units))
model.add(Dense(64, activation='relu', input_shape=(86,)))  # Assuming 5 emotional labels

model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])

model.fit(X_train, y_train, epochs=10, batch_size=32, validation_data=(X_test, y_test))

# Step 5: Text Summarization
def summarize_text(text):
    sequence = tokenizer.texts_to_sequences([text])
    sequence = pad_sequences(sequence, maxlen=max_sequence_length)
    predicted_emotion = model.predict(sequence)[0]
    emotion_labels = ['joy', 'sadness', 'surprise', 'anger', 'neutral'] 
    summary = [(label, prob) for label, prob in zip(emotion_labels, predicted_emotion)]
    summary.sort(key=lambda x: x[1], reverse=True)
    return summary

# Example usage
input_text = "I had a great day with my friends. We laughed a lot and enjoyed ourselves."
summary = summarize_text(input_text)
print(summary)

当我尝试模型拟合时,我出现了以下错误

ValueError: Shapes (None, 1) and (None, 64) are incompatible