tf.gather with tf.Session() as sess:

# python
Python 2.7.12 (default, Oct  8 2019, 14:14:10)
[GCC 5.4.0 20160609] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy as np
>>> import tensorflow as tf
>>> a = np.arange(12).reshape(2,2,3)
>>> a
array([[[ 0,  1,  2],
        [ 3,  4,  5]],

       [[ 6,  7,  8],
        [ 9, 10, 11]]])
>>> b = tf.constant([2, 1, 0])
>>> b
<tf.Tensor 'Const:0' shape=(3,) dtype=int32>
>>> a = tf.gather(a, b , axis=1)
>>> a
<tf.Tensor 'GatherV2:0' shape=(2, 3, 3) dtype=int64>
>>> with tf.Session() as sess:
...   print sess.run(a)
...
...
2020-06-16 09:54:51.947755: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFl                                                                      ow binary was not compiled to use: AVX2 FMA
2020-06-16 09:54:52.243377: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:998] successful NUMA node read from SysFS had n                                                                      egative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-06-16 09:54:52.247393: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x5f3a2a0 executing computations on plat                                                                      form CUDA. Devices:
2020-06-16 09:54:52.247432: I tensorflow/compiler/xla/service/service.cc:158]   StreamExecutor device (0): Tesla V100-SXM2-32GB-LS                                                                      , Compute Capability 7.0
2020-06-16 09:54:52.250564: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2195105000 Hz
2020-06-16 09:54:52.259420: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x60f6f30 executing computations on plat                                                                      form Host. Devices:
2020-06-16 09:54:52.259465: I tensorflow/compiler/xla/service/service.cc:158]   StreamExecutor device (0): <undefined>, <undefined                                                                      >
2020-06-16 09:54:52.261153: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1433] Found device 0 with properties:
name: Tesla V100-SXM2-32GB-LS major: 7 minor: 0 memoryClockRate(GHz): 1.44
pciBusID: 0000:00:0c.0
totalMemory: 31.72GiB freeMemory: 31.27GiB
2020-06-16 09:54:52.261191: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1512] Adding visible gpu devices: 0
2020-06-16 09:54:52.263624: I tensorflow/core/common_runtime/gpu/gpu_device.cc:984] Device interconnect StreamExecutor with streng                                                                      th 1 edge matrix:
2020-06-16 09:54:52.263657: I tensorflow/core/common_runtime/gpu/gpu_device.cc:990]      0
2020-06-16 09:54:52.263668: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1003] 0:   N
2020-06-16 09:54:52.265156: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] Created TensorFlow device (/job:localhost/rep                                                                      lica:0/task:0/device:GPU:0 with 30417 MB memory) -> physical GPU (device: 0, name: Tesla V100-SXM2-32GB-LS, pci bus id: 0000:00:0c                                                                      .0, compute capability: 7.0)
[[[ 0  0  0]
  [ 3  4  5]
  [ 0  1  2]]

 [[ 0  0  0]
  [ 9 10 11]
  [ 6  7  8]]]