[CPU/GPU] paddle.tile rejects repeat_times containing 0

bug描述 Describe the Bug

paddle.tile rejects repeat_times values that contain 0. However, the documented output-size rule says that the size of each output dimension is x.shape[i] * repeat_times[i], which makes a zero repeat well-defined as a zero-sized output dimension.

Minimal reproducing example:

import traceback
import paddle

print("paddle:", paddle.__version__)

for device in ["cpu", "gpu:0"]:
    paddle.device.set_device(device)
    print("\nDevice:", device)
    x = paddle.empty([0, 5], dtype="float32")
    try:
        out = paddle.tile(x, repeat_times=[1, 0])
        print(out.shape, out.dtype, out.place)
    except Exception:
        traceback.print_exc()

Expected result:

Device: cpu
[0, 0] paddle.float32 Place(cpu)

Device: gpu:0
[0, 0] paddle.float32 Place(gpu:0)

The key point is that repeat_times=[1, 0] should produce a tensor whose second dimension is 5 * 0 == 0, instead of rejecting the repeat factor before shape construction.

Actual result:

Traceback (most recent call last):
  File "<string>", line 11, in <module>
  File ".../site-packages/paddle/tensor/manipulation.py", line 3481, in tile
    return _C_ops.tile(x, repeat_times)
ValueError: (InvalidArgument) Every element of the input 'repeat_times' for tile op must be greater than 0, but the value given is 0.
  [Hint: Expected repeat_times_data[i] > 0, but received repeat_times_data[i]:0 <= 0:0.]
  (at ../paddle/phi/infermeta/unary.cc:4424)

For reference, TensorFlow handles zero multiples and returns the expected empty shape:

import tensorflow as tf

with tf.device("/GPU:0"):
    x = tf.zeros([0, 5], dtype=tf.float32)
    out = tf.tile(x, multiples=[1, 0])
print(out.shape, out.dtype, out.device)

(0, 0) <dtype: 'float32'> /job:localhost/replica:0/task:0/device:GPU:0

其他补充信息 Additional Supplementary Information

Reproduced in fresh Python processes with CUDA_LAUNCH_BLOCKING=1.
Environment:

• Python: 3.10.20
• PaddlePaddle: 2.6.1
• GPU: NVIDIA GeForce RTX 3090
• NVIDIA driver: 595.58.03
• Paddle CUDA runtime: 11.7