机器学习碰到MemoryError: Unable to allocate1.5 . MiB for an array with shape解决办法

详细代码如下

# 训练集
train_gen = train_val_generator(
    data_dir='../train',
    target_size=(64, 64),
    batch_size=32,
    class_model='categorical',
    subset='training'
)
# 验证集
val_gen = train_val_generator(
    data_dir='../test',
    target_size=(64, 64),
    batch_size=32,
    class_model='categorical',
    subset='validation'
)

train_batch, train_label_batch = train_gen.next()

plot_images(train_batch, train_label_batch)

val_batch, val_label_batch = val_gen.next()

plot_images(val_batch, val_label_batch)

model = ConvModel()

model.compile(loss='categorical_crossentropy', optimizer=tf.keras.optimizers.SGD(learning_rate=0.01),
              metrics=['accuracy'])

history = model.fit(x=train_gen, steps_per_epoch=351, epochs=100, validation_data=val_gen, validation_steps=88,
                    shuffle=True)

在训练到第50轮的时候报错
MemoryError: Unable to allocate 1.50 MiB for an array with shape (32, 64, 64, 3) and data type float32
一直到现在都没有解决，有没有大神有好的解决办法，求助

最后通过监控GPU，内存，磁盘容量，发现是由于C盘空间不够导致的！！！！
应该是某个库或者插件在运算过程中会产生临时数据，怎么找这个库就不太清楚了，我是清理了C盘，然后做了扩容