import tensorflow as tf

# 1) data: a [0, 4] double tensor (all zeros)
data = tf.zeros([0, 4], dtype=tf.double)

# 2) indices: a [4] int32 tensor with values [48, 0, 116, 0]
indices = tf.constant([48, 0, 116, 0], dtype=tf.int32)

# 3) segment_ids: a [4] int32 tensor (all zeros)
segment_ids = tf.zeros([4], dtype=tf.int32)

# 4) Run the SparseSegmentSum op on GPU
with tf.device('/GPU:0'):
    result = tf.raw_ops.SparseSegmentSum(
        data=data,
        indices=indices,
        segment_ids=segment_ids
    )

tf.print("SparseSegmentSum result:", result)

========= Invalid __global__ read of size 16 bytes
=========     at void tensorflow::SegmentReduceVectorKernel<tensorflow::AlignedVector<double, (int)2>, tensorflow::AlignedVector<double, (int)2>, int, int, int, tensorflow::functor::Sum, double, double>(T3, T3, T5, T6, T7, T7, bool, bool, const T2 *, const T3 *, const T4 *, const T8 *, T2 *)+0x620
=========     by thread (0,0,0) in block (0,0,0)
=========     Address 0x600 is out of bounds