合规国际互联网加速 OSASE为企业客户提供高速稳定SD-WAN国际加速解决方案。 广告
# 扩展PyTorch # 扩展PyTorch 本篇文章中包含如何扩展 `torch.nn`, `torch.autograd`和 使用我们的 `C 库`编写自定义的`C`扩展。 ## 扩展 torch.autograd 如果你想要添加一个新的 `Operation` 到`autograd`的话,你的`Operation`需要继承 `class Function`。`autograd`使用`Function`计算结果和梯度,同时编码 `operation`的历史。每个新的 `operation(function)` 都需要实现三个方法: - `__init__ (optional)` - 如果你的`operation`包含非`Variable`参数,那么就将其作为`__init__`的参数传入到`operation`中。例如:`AddConstant Function`加一个常数,`Transpose Function`需要指定哪两个维度需要交换。如果你的`operation`不需要额外的参数,你可以忽略`__init__`。 - `forward()` - 在里面写执行此`operation`的代码。可以有任意数量的参数。如果你对某些参数指定了默认值,则这些参数是可传可不传的。记住:`forward()`的参数只能是`Variable`。函数的返回值既可以是 `Variable`也可以是`Variables`的`tuple`。同时,请参考 `Function`\[function\]的 `doc`,查阅有哪些 方法是只能在`forward`中调用的。 - `backward()` - 梯度计算公式。 参数的个数和`forward`返回值的个数一样,每个参数代表传回到此`operation`的梯度. `backward()`的返回值的个数应该和此`operation`输入的个数一样,每个返回值对应了输入值的梯度。如果`operation`的输入不需要梯度,或者不可导,你可以返回`None`。 如果`forward()`存在可选参数,你可以返回比输入更多的梯度,只是返回的是`None`。 下面是 `Linear` 的实现代码: ``` # Inherit from Function class Linear(Function): # bias is an optional argument def forward(self, input, weight, bias=None): self.save_for_backward(input, weight, bias) output = input.mm(weight.t()) if bias is not None: output += bias.unsqueeze(0).expand_as(output) return output # This function has only a single output, so it gets only one gradient def backward(self, grad_output): # This is a pattern that is very convenient - at the top of backward # unpack saved_tensors and initialize all gradients w.r.t. inputs to # None. Thanks to the fact that additional trailing Nones are # ignored, the return statement is simple even when the function has # optional inputs. input, weight, bias = self.saved_tensors grad_input = grad_weight = grad_bias = None # These needs_input_grad checks are optional and there only to # improve efficiency. If you want to make your code simpler, you can # skip them. Returning gradients for inputs that don't require it is # not an error. if self.needs_input_grad[0]: grad_input = grad_output.mm(weight) if self.needs_input_grad[1]: grad_weight = grad_output.t().mm(input) if bias is not None and self.needs_input_grad[2]: grad_bias = grad_output.sum(0).squeeze(0) return grad_input, grad_weight, grad_bias ``` 现在,为了可以更简单的使用自定义的`operation`,我们建议将其用一个简单的 `helper function` 包装起来。 functions: ``` def linear(input, weight, bias=None): # First braces create a Function object. Any arguments given here # will be passed to __init__. Second braces will invoke the __call__ # operator, that will then use forward() to compute the result and # return it. return Linear()(input, weight, bias) ``` 你可能想知道你刚刚实现的 `backward`方法是否正确的计算了梯度。你可以使用 小的有限的差分进行数值估计。 ``` from torch.autograd import gradcheck # gradchek takes a tuple of tensor as input, check if your gradient # evaluated with these tensors are close enough to numerical # approximations and returns True if they all verify this condition. input = (Variable(torch.randn(20,20).double(), requires_grad=True),) test = gradcheck.gradcheck(Linear(), input, eps=1e-6, atol=1e-4) print(test) ``` ## 扩展 torch.nn `nn` 包含两种接口 - `modules`和他们的`functional`版本。通过这两个接口,你都可以扩展`nn`。但是我们建议,在扩展`layer`的时候,使用`modules`, 因为`modules`保存着参数和`buffer`。如果不需要参数的话,那么建议使用`functional`(激活函数,pooling,这些都不需要参数)。 增加一个`operation`的 `functional`版本已经在上面一节介绍完毕。 增加一个模块(`module`)。 由于`nn`重度使用`autograd`。所以,添加一个新`module`需要实现一个 用来执行 计算 和 计算梯度 的`Function`。从现在开始,假定我们想要实现一个`Linear module`,记得之前我们已经实现了一个`Linear Funciton`。 只需要很少的代码就可以完成这个工作。 现在,我们需要实现两个方法: - `__init__ (optional)` - 输入参数,例如`kernel sizes`, `numbers of features`, 等等。同时初始化 `parameters`和`buffers`。 - `forward()` - 实例化一个执行`operation`的`Function`,使用它执行`operation`。和`functional wrapper(上面实现的那个简单的wrapper)`十分类似。 `Linear module`实现代码: ``` class Linear(nn.Module): def __init__(self, input_features, output_features, bias=True): self.input_features = input_features self.output_features = output_features # nn.Parameter is a special kind of Variable, that will get # automatically registered as Module's parameter once it's assigned # as an attribute. Parameters and buffers need to be registered, or # they won't appear in .parameters() (doesn't apply to buffers), and # won't be converted when e.g. .cuda() is called. You can use # .register_buffer() to register buffers. # nn.Parameters can never be volatile and, different than Variables, # they require gradients by default. self.weight = nn.Parameter(torch.Tensor(input_features, output_features)) if bias: self.bias = nn.Parameter(torch.Tensor(output_features)) else: # You should always register all possible parameters, but the # optional ones can be None if you want. self.register_parameter('bias', None) # Not a very smart way to initialize weights self.weight.data.uniform_(-0.1, 0.1) if bias is not None: self.bias.data.uniform_(-0.1, 0.1) def forward(self, input): # See the autograd section for explanation of what happens here. return Linear()(input, self.weight, self.bias) #注意这个Linear是之前实现过的Linear ``` ## 编写自定义`C`扩展 Coming soon. For now you can find an example at [GitHub](https://github.com/pytorch/extension-ffi).