Fix one_hot api limit

[leaves-zwx]

No description provided.

[yuanms2]

这个是修复谢暄说的wide and deep 不收敛的问题吗？

[leaves-zwx]

这个是修复谢暄说的wide and deep 不收敛的问题吗？

不是的，这个是在 nn.Graph 里面不能调用 one_hot 的问题。

[MARD1NO]

这个是修复谢暄说的wide and deep 不收敛的问题吗？

修复的是nn.Graph下使用one_hot的问题

[github-actions]

CI failed, removing label automerge

[MARD1NO]

应该是if num_classes == -1

[leaves-zwx]

最好别用 -1 代表参数 optional 的语义

[leaves-zwx]

我看了 torch.nn.functional.one_hot，它这就是 -1 代表需要从 input 求值，改为与它对齐

[github-actions]

CI failed, removing label automerge

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

PyTorch resnet50 time: 142.2ms (= 7108.1ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 128.4ms (= 6420.5ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
Relative speed: 1.11 (= 142.2ms / 128.4ms)

PyTorch resnet50 time: 81.3ms (= 4065.1ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 74.5ms (= 3723.4ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
Relative speed: 1.09 (= 81.3ms / 74.5ms)

PyTorch resnet50 time: 59.1ms (= 2955.8ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 47.8ms (= 2389.1ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
Relative speed: 1.24 (= 59.1ms / 47.8ms)

PyTorch resnet50 time: 50.2ms (= 2510.8ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 40.4ms (= 2018.5ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
Relative speed: 1.24 (= 50.2ms / 40.4ms)

PyTorch resnet50 time: 44.4ms (= 2219.9ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 42.2ms (= 2109.0ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
Relative speed: 1.05 (= 44.4ms / 42.2ms)