Add more documentation

[wyg1997]

Check list

autograd.backward ✅

autograd.autograd ✅

Tensor.grad ✅

Tensor.grad_fn ✅

Tensor.is_lazy ✅

Tensor.is_leaf ✅

Tensor.requires_grad ✅

Tensor.requires_grad_ ✅

Tensor.register_hook ✅

retain_grad ✅

Bugfix

使 pybind11 导出类的派生类可以为没有重写的方法添加文档

[doombeaker]

而且不少 C++ 里直接导出的方法，如果第一行不写函数原型的话，可能会导致 sphinx 自动提取的 signature 是错的（是无参无返回值的），或者很复杂（暴露内部类型）

可以参考下这个 PR：https://github.com/Oneflow-Inc/oneflow/pull/7834/files

[github-actions]

CI failed when running job: cuda-module. PR label automerge has been removed

[github-actions]

Speed stats:

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

✔️ OneFlow resnet50 time: 128.3ms (= 12832.0ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 137.4ms (= 13736.1ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.07 (= 137.4ms / 128.3ms)

❌ OneFlow resnet50 time: 80.8ms (= 8079.0ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 84.2ms (= 8418.5ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.04 (= 84.2ms / 80.8ms)

OneFlow resnet50 time: 52.7ms (= 10549.2ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 56.5ms (= 11308.3ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.07 (= 56.5ms / 52.7ms)

OneFlow resnet50 time: 43.6ms (= 8722.4ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 49.1ms (= 9826.0ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.13 (= 49.1ms / 43.6ms)

OneFlow resnet50 time: 39.5ms (= 7899.4ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 38.5ms (= 7702.5ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 0.98 (= 38.5ms / 39.5ms)

md5 /home/ci-user/ci-cache/test_cache/onerec_test/nanodataset.zip 7f5cde8b5a6c411107517ac9b00f29db
md5 /home/ci-user/ci-cache/test_cache/onerec_test/nanodataset.zip 7f5cde8b5a6c411107517ac9b00f29db
build_dataset >>>>>> ImageFolder
build_dataset >>>>>> ImageFolder
✔️ Relative speed: 1.00 (= 51.0s / 51.1s)

md5 /home/ci-user/ci-cache/test_cache/onerec_test/nanodataset.zip 7f5cde8b5a6c411107517ac9b00f29db
md5 /home/ci-user/ci-cache/test_cache/onerec_test/nanodataset.zip 7f5cde8b5a6c411107517ac9b00f29db
build_dataset >>>>>> ImageFolder
build_dataset >>>>>> ImageFolder
✔️ Relative speed: 1.01 (= 13.3s / 13.5s)

md5 /home/ci-user/ci-cache/test_cache/onerec_test/nanodataset.zip 7f5cde8b5a6c411107517ac9b00f29db
md5 /home/ci-user/ci-cache/test_cache/onerec_test/nanodataset.zip 7f5cde8b5a6c411107517ac9b00f29db
build_dataset >>>>>> ImageFolder
build_dataset >>>>>> ImageFolder
❌ Relative speed: 0.94 (= 7.9s / 7.4s)

✔️ OneFlow resnet50 time: 135.5ms (= 13549.0ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 156.3ms (= 15630.2ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.15 (= 156.3ms / 135.5ms)

OneFlow resnet50 time: 84.5ms (= 8449.1ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 101.5ms (= 10151.4ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.20 (= 101.5ms / 84.5ms)

OneFlow resnet50 time: 60.9ms (= 12184.0ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 86.2ms (= 17243.5ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.42 (= 86.2ms / 60.9ms)

OneFlow resnet50 time: 54.2ms (= 10831.8ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 66.2ms (= 13241.1ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.22 (= 66.2ms / 54.2ms)

OneFlow resnet50 time: 48.1ms (= 9620.8ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 64.9ms (= 12978.9ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.35 (= 64.9ms / 48.1ms)

[github-actions]

CI failed when running job: cuda-speed-test. PR label automerge has been removed