tensorflow使用sparse_to_dense方法出現 indices is out of bounds的錯誤

tf.sparse_to_dense()方法有時會出現indices out of bounds的情況，比如：

import tensorflow as tf
# 假設數據標籤有3類
label = tf.expand_dims(tf.constant([1, 2, 2, 3, 3]), 1)
index = tf.expand_dims(tf.range(5), 1)
conact = tf.concat([index, label], 1)

with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    print(sess.run(tf.sparse_to_dense(conact, [5, 3], 1.0, 0.0)))

InvalidArgumentError (see above for traceback): indices[3] = [3,3] is out of bounds: need 0 <= index < [5,3]
	 [[Node: SparseToDense = SparseToDense[T=DT_FLOAT, Tindices=DT_INT32, validate_indices=true, _device="/job:localhost/replica:0/task:0/device:CPU:0"](concat/_1, SparseToDense/output_shape, SparseToDense/sparse_values, SparseToDense/default_value)]]

假設數據標籤爲1，2，3，按照常理輸出的矩陣大小爲[5,3]，但這樣會出錯，可以把輸出矩陣大小改爲[5,4]，或者將1，2，3標籤改爲0，1，2

print(sess.run(tf.sparse_to_dense(conact, [5, 4], 1.0, 0.0)))

[5,4]的輸出結果：
[[0. 1. 0. 0.]
[0. 0. 1. 0.]
[0. 0. 1. 0.]
[0. 0. 0. 1.]
[0. 0. 0. 1.]]

label = tf.expand_dims(tf.constant([0, 1, 2, 2, 2]), 1)

[5,3]的輸出結果：
[[1. 0. 0.]
[0. 1. 0.]
[0. 1. 0.]
[0. 0. 1.]
[0. 0. 1.]]