Pytorch

Question

我一直运行进入这个错误：

RuntimeError: Trying to backward through the graph a second time, but the buffers have already been freed. Specify retain_graph=True when calling backward the first time.

我已经在Pytorch论坛上搜索过了，但仍然找不到我的自定义损失函数做错了什么。我的模型是 nn.GRU，这是我的自定义损失函数：

def _loss(outputs, session, items):  # `items` is a dict() contains embedding of all items
    def f(output, target):
        pos = torch.from_numpy(np.array([items[target["click"]]])).float()
        neg = torch.from_numpy(np.array([items[idx] for idx in target["suggest_list"] if idx != target["click"]])).float()
        if USE_CUDA:
            pos, neg = pos.cuda(), neg.cuda()
        pos, neg = Variable(pos), Variable(neg)

        pos = F.cosine_similarity(output, pos)
        if neg.size()[0] == 0:
            return torch.mean(F.logsigmoid(pos))
        neg = F.cosine_similarity(output.expand_as(neg), neg)

        return torch.mean(F.logsigmoid(pos - neg))

    loss = map(f, outputs, session)
return -torch.mean(torch.cat(loss))

训练代码：

    # zero the parameter gradients
    model.zero_grad()

    # forward + backward + optimize
    outputs, hidden = model(inputs, hidden)
    loss = _loss(outputs, session, items)
    acc_loss += loss.data[0]

    loss.backward()
    # Add parameters' gradients to their values, multiplied by learning rate
    for p in model.parameters():
        p.data.add_(-learning_rate, p.grad.data)

Answer 1

问题出在我的训练循环中：它不会在批次之间分离或重新打包隐藏状态？如果是这样，那么 loss.backward() 会尝试一直反向传播到时间的开始，这适用于第一批但不适用于第二批，因为第一批的图表已被丢弃。

有两种可能的解决方案。

1) detach/repackage 批次之间的隐藏状态。有（在至少）三种方法（我选择了这个解决方案）：

 hidden.detach_()
 hidden = hidden.detach()

2) 将 loss.backward() 替换为 loss.backward(retain_graph=True) 但要知道每个连续的批次都将比前一个批次花费更多的时间，因为它必须一直反向传播到开始第一批。

Example

Pytorch - RuntimeError: Trying to backward through the graph a second time, but the buffers have already been freed

Pytorch - RuntimeError: Trying to backward through the graph a second time, but the buffers have already been freed

python

deep-learning

recurrent-neural-network

autograd