代码示例 / 自然语言处理 / IMDB 上的双向 LSTM

IMDB 上的双向 LSTM

作者: fchollet
创建日期 2020/05/03
上次修改 2020/05/03
描述:在 IMDB 电影评论情感分类数据集上训练一个两层双向 LSTM。

ⓘ 此示例使用 Keras 3

在 Colab 中查看 GitHub 源码


设置

import numpy as np
import keras
from keras import layers

max_features = 20000  # Only consider the top 20k words
maxlen = 200  # Only consider the first 200 words of each movie review

构建模型

# Input for variable-length sequences of integers
inputs = keras.Input(shape=(None,), dtype="int32")
# Embed each integer in a 128-dimensional vector
x = layers.Embedding(max_features, 128)(inputs)
# Add 2 bidirectional LSTMs
x = layers.Bidirectional(layers.LSTM(64, return_sequences=True))(x)
x = layers.Bidirectional(layers.LSTM(64))(x)
# Add a classifier
outputs = layers.Dense(1, activation="sigmoid")(x)
model = keras.Model(inputs, outputs)
model.summary()
Model: "functional_1"
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Layer (type)                     Output Shape                  Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ input_layer (InputLayer)        │ (None, None)              │          0 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ embedding (Embedding)           │ (None, None, 128)         │  2,560,000 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ bidirectional (Bidirectional)   │ (None, None, 128)         │     98,816 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ bidirectional_1 (Bidirectional) │ (None, 128)               │     98,816 │
├─────────────────────────────────┼───────────────────────────┼────────────┤
│ dense (Dense)                   │ (None, 1)                 │        129 │
└─────────────────────────────────┴───────────────────────────┴────────────┘
 Total params: 2,757,761 (10.52 MB)
 Trainable params: 2,757,761 (10.52 MB)
 Non-trainable params: 0 (0.00 B)

加载 IMDB 电影评论情感数据

(x_train, y_train), (x_val, y_val) = keras.datasets.imdb.load_data(
    num_words=max_features
)
print(len(x_train), "Training sequences")
print(len(x_val), "Validation sequences")
# Use pad_sequence to standardize sequence length:
# this will truncate sequences longer than 200 words and zero-pad sequences shorter than 200 words.
x_train = keras.utils.pad_sequences(x_train, maxlen=maxlen)
x_val = keras.utils.pad_sequences(x_val, maxlen=maxlen)
Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/imdb.npz
 17464789/17464789 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
25000 Training sequences
25000 Validation sequences

训练和评估模型

您可以使用托管在 Hugging Face Hub 上的训练模型,并在 Hugging Face Spaces 上尝试演示。

model.compile(optimizer="adam", loss="binary_crossentropy", metrics=["accuracy"])
model.fit(x_train, y_train, batch_size=32, epochs=2, validation_data=(x_val, y_val))
Epoch 1/2
 782/782 ━━━━━━━━━━━━━━━━━━━━ 61s 75ms/step - accuracy: 0.7540 - loss: 0.4697 - val_accuracy: 0.8269 - val_loss: 0.4202
Epoch 2/2
 782/782 ━━━━━━━━━━━━━━━━━━━━ 54s 69ms/step - accuracy: 0.9151 - loss: 0.2263 - val_accuracy: 0.8428 - val_loss: 0.3650

<keras.src.callbacks.history.History at 0x7f3efd663850>