make setitem device match

[Flowingsun007]

No description provided.

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

PyTorch resnet50 time: 149.4ms (= 7469.3ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 128.3ms (= 6413.3ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
Relative speed: 1.16 (= 149.4ms / 128.3ms)

PyTorch resnet50 time: 84.3ms (= 4213.5ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 74.4ms (= 3719.5ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
Relative speed: 1.13 (= 84.3ms / 74.4ms)

PyTorch resnet50 time: 58.3ms (= 2914.3ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 47.3ms (= 2364.6ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
Relative speed: 1.23 (= 58.3ms / 47.3ms)

PyTorch resnet50 time: 56.6ms (= 2828.4ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 42.1ms (= 2103.8ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
Relative speed: 1.34 (= 56.6ms / 42.1ms)

PyTorch resnet50 time: 47.9ms (= 2396.0ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 37.9ms (= 1896.5ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
Relative speed: 1.26 (= 47.9ms / 37.9ms)

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

PyTorch resnet50 time: 140.5ms (= 7023.5ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 128.3ms (= 6417.3ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
Relative speed: 1.09 (= 140.5ms / 128.3ms)

PyTorch resnet50 time: 84.3ms (= 4214.5ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 74.4ms (= 3719.4ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
Relative speed: 1.13 (= 84.3ms / 74.4ms)

PyTorch resnet50 time: 58.1ms (= 2903.1ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 48.0ms (= 2399.3ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
Relative speed: 1.21 (= 58.1ms / 48.0ms)

PyTorch resnet50 time: 49.0ms (= 2448.8ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 40.6ms (= 2027.8ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
Relative speed: 1.21 (= 49.0ms / 40.6ms)

PyTorch resnet50 time: 46.7ms (= 2337.3ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 44.1ms (= 2204.9ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
Relative speed: 1.06 (= 46.7ms / 44.1ms)