rename OpExprInterpState to AutoGradCaptureState

[simonJJJ]

No description provided.

[hjchen2]

把这个挪到op_expr_grad.h文件里去？

[simonJJJ]

已改

[lixinqi]

删掉注释

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

PyTorch resnet50 time: 149.8ms (= 7490.2ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 128.6ms (= 6429.6ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
Relative speed: 1.16 (= 149.8ms / 128.6ms)

PyTorch resnet50 time: 86.3ms (= 4312.6ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 74.8ms (= 3741.1ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
Relative speed: 1.15 (= 86.3ms / 74.8ms)

PyTorch resnet50 time: 66.0ms (= 3302.4ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 49.1ms (= 2452.5ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
Relative speed: 1.35 (= 66.0ms / 49.1ms)

PyTorch resnet50 time: 80.8ms (= 4041.0ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 50.1ms (= 2503.3ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
Relative speed: 1.61 (= 80.8ms / 50.1ms)

PyTorch resnet50 time: 64.5ms (= 3226.1ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 53.9ms (= 2695.0ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
Relative speed: 1.20 (= 64.5ms / 53.9ms)