TensorFlow | 如何实现10倍交叉验证？

Question

如何在这段代码中实现 10 倍交叉验证？

(train_ds, val_ds, test_ds), metadata = tfds.load(

    'tf_flowers',

    split=['train[:60%]', 'train[60%:90%]', 'train[90%:]'],

    with_info=True,

    as_supervised=True)

聚苯乙烯

也许我做了 10 倍交叉验证，但我不确定。

(train_ds, test_ds), metadata = tfds.load(

    'tf_flowers',

    split=['train[:90%]', 'train[90%:]'],

    with_info=True,

    as_supervised=True

)

val_ds = train_ds.split = [

  f'train[{k}%:{k+10}%]' for k in range(0, 100, 10)

]

Answer 1

对我有什么帮助！(train_ds, test_ds), metadata = tfds.load(    'tf_flowers',    split=['train[:90%]', 'train[90%:]'],    with_info=True,    as_supervised=True)val_ds = train_ds.split = [  f'train[{k}%:{k+10}%]' for k in range(0, 100, 10)]