TensorFlow | 如何實現 10 折交叉驗證？

Question

如何在此代碼中實現 10 倍交叉驗證？

(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

也許我做了 10 倍交叉驗證，但我不確定。

(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

什么幫助了我！

(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)
]