import tensorflow as tf

a = tf.random.shuffle(tf.range(5))
a

tf.sort(a, direction='DESCENDING')

# 返回索引
tf.argsort(a, direction='DESCENDING')

idx = tf.argsort(a, direction='DESCENDING')
tf.gather(a, idx)

idx = tf.argsort(a, direction='DESCENDING')
tf.gather(a, idx)

a = tf.random.uniform([3, 3], maxval=10, dtype=tf.int32)
a

tf.sort(a)

tf.sort(a, direction='DESCENDING')

idx = tf.argsort(a)
idx

# 返回前2个值
res = tf.math.top_k(a, 2)
res

res.values

res.indices

prob = tf.constant([[0.1, 0.2, 0.7], [0.2, 0.7, 0.1]])
target = tf.constant([2, 0])

# 概率最大的索引在最前面
k_b = tf.math.top_k(prob, 3).indices
k_b

k_b = tf.transpose(k_b, [1, 0])
k_b

# 对真实值broadcast，与prod比较
target = tf.broadcast_to(target, [3, 2])
target

def accuracy(output, target, topk=(1, )):
    maxk = max(topk)
    batch_size = target.shape[0]

    pred = tf.math.top_k(output, maxk).indices
    pred = tf.transpose(pred, perm=[1, 0])
    target_ = tf.broadcast_to(target, pred.shape)
    correct = tf.equal(pred, target_)

    res = []
    for k in topk:
        correct_k = tf.cast(tf.reshape(correct[:k], [-1]), dtype=tf.float32)
        correct_k = tf.reduce_sum(correct_k)
        acc = float(correct_k / batch_size)
        res.append(acc)

    return res

# 10个样本6类
output = tf.random.normal([10, 6])
# 使得所有样本的概率加起来为1
output = tf.math.softmax(output, axis=1)
# 10个样本对应的标记
target = tf.random.uniform([10], maxval=6, dtype=tf.int32)
print(f'prob: {output.numpy()}')
pred = tf.argmax(output, axis=1)
print(f'pred: {pred.numpy()}')
print(f'label: {target.numpy()}')

acc = accuracy(output, target, topk=(1, 2, 3, 4, 5, 6))
print(f'top-1-6 acc: {acc}')