TensorFlow:使用索引从张量中获取子张量

Question

A 是一个 tensorflow.tensor，形状为 (2261,)

我想从 A 的以下索引中获取一个新的张量:[10,20,30]

我尝试了以下所有方法，但均无效:

A[[10,20,30]]
# *** ValueError: Index out of range using input dim 1; input has only 1 dims for 'strided_slice' (op: 'StridedSlice') with input shapes: [2261], [3], [3], [3].

A[10,20,30]
# same error as above 

A[numpy.array([10,20,30])]
# *** ValueError: Shape must be rank 1 but is rank 2 for 'strided_slice' (op: 'StridedSlice') with input shapes: [2261], [1,3], [1,3], [1].

A[10]
# <tf.Tensor 'strided_slice:0' shape=() dtype=float32> - not an error but a shapeless tensor

A[tensorflow.constant(10)]
# same problem as above

为什么这些不起作用，我该怎么办？

Answer 1

C = tf.nn.embedding_lookup(A, B)

其中 B 是一个值为 [10,20,30] 的张量

供引用:https://www.tensorflow.org/api_docs/python/nn/embeddings