【代码】SupConLoss

eg1: 相等的元素为True

torch.eq(torch.arange(5).view(5,1),torch.arange(5).view(5,1))
Out[28]: 
tensor([[True],
        [True],
        [True],
        [True],
        [True]])
torch.eq(torch.arange(5).view(5,1),torch.arange(5).view(5,1).T)
Out[29]: 
tensor([[ True, False, False, False, False],
        [False,  True, False, False, False],
        [False, False,  True, False, False],
        [False, False, False,  True, False],
        [False, False, False, False,  True]])
torch.eq(torch.arange(5).view(5,1).T,torch.arange(5).view(5,1))
Out[30]: 
tensor([[ True, False, False, False, False],
        [False,  True, False, False, False],
        [False, False,  True, False, False],
        [False, False, False,  True, False],
        [False, False, False, False,  True]])

eg2: 解开指定的维度

torch.unbind(torch.tensor([[1],[2],[3]]),1)
Out[36]: (tensor([1, 2, 3]),)
torch.unbind(torch.tensor([[1],[2],[3]]),0)
Out[37]: (tensor([1]), tensor([2]), tensor([3]))

eg3:

torch.unbind(features,dim=1)
Out[44]: 
(tensor([[ 0.0165, -0.1257,  0.0335,  ...,  0.0430,  0.0588,  0.0256],
         [ 0.0581, -0.0996, -0.0443,  ..., -0.0111,  0.1081, -0.0078],
         [ 0.0172, -0.1306, -0.0858,  ..., -0.0411,  0.0833,  0.0013],
         ...,
         [ 0.0601, -0.1264, -0.0413,  ...,  0.0127,  0.1198, -0.0309],
         [-0.0102, -0.1497,  0.0010,  ..., -0.0122,  0.1112, -0.0583],
         [ 0.0758, -0.1189, -0.0197,  ...,  0.0220,  0.0872, -0.0166]],
        device='cuda:0', grad_fn=<UnbindBackward>),
 tensor([[ 0.0165, -0.1257,  0.0335,  ...,  0.0430,  0.0588,  0.0256],
         [ 0.0581, -0.0996, -0.0443,  ..., -0.0111,  0.1081, -0.0078],
         [ 0.0172, -0.1306, -0.0858,  ..., -0.0411,  0.0833,  0.0013],
         ...,
         [ 0.0601, -0.1264, -0.0413,  ...,  0.0127,  0.1198, -0.0309],
         [-0.0102, -0.1497,  0.0010,  ..., -0.0122,  0.1112, -0.0583],
         [ 0.0758, -0.1189, -0.0197,  ...,  0.0220,  0.0872, -0.0166]],
        device='cuda:0', grad_fn=<UnbindBackward>))

eg4: t.repeat()

eg5: torch.div()

eq6:

# tile mask
mask = mask.repeat(anchor_count, contrast_count)
# mask-out self-contrast cases
logits_mask = torch.scatter(
    torch.ones_like(mask),
    1,
    torch.arange(batch_size * anchor_count).view(-1, 1).to(device),
    0
)

logits_mask

tensor([[0., 1., 1.,  ..., 1., 1., 1.],
        [1., 0., 1.,  ..., 1., 1., 1.],
        [1., 1., 0.,  ..., 1., 1., 1.],
        ...,
        [1., 1., 1.,  ..., 0., 1., 1.],
        [1., 1., 1.,  ..., 1., 0., 1.],
        [1., 1., 1.,  ..., 1., 1., 0.]], device='cuda:0')

----

torch.scatter(
            torch.ones_like(mask),
            0,
            torch.arange(batch_size * anchor_count).view(-1, 1).to(device),
            0)

tensor([[0., 1., 1.,  ..., 1., 1., 1.],
        [0., 1., 1.,  ..., 1., 1., 1.],
        [0., 1., 1.,  ..., 1., 1., 1.],
        ...,
        [0., 1., 1.,  ..., 1., 1., 1.],
        [0., 1., 1.,  ..., 1., 1., 1.],
        [0., 1., 1.,  ..., 1., 1., 1.]], device='cuda:0')

【笔记】scatter_函数:用法如 torch.zeros(target.size(0), 2).scatter_(1,target,1).to(self.device)_程序猿的探索之路的博客-CSDN博客

eg7: 消除对角元素

mask = mask.repeat(anchor_count, contrast_count)
# mask-out self-contrast cases
logits_mask = torch.scatter(
    torch.ones_like(mask),
    1,
    torch.arange(batch_size * anchor_count).view(-1, 1).to(device),
    0
)
mask = mask * logits_mask

mask
Out[20]: 
tensor([[1., 0., 1.,  ..., 1., 0., 0.],
        [0., 1., 0.,  ..., 0., 0., 0.],
        [1., 0., 1.,  ..., 1., 0., 0.],
        ...,
        [1., 0., 1.,  ..., 1., 0., 0.],
        [0., 0., 0.,  ..., 0., 1., 0.],
        [0., 0., 0.,  ..., 0., 0., 1.]], device='cuda:0')
mask*logits_mask
Out[21]: 
tensor([[0., 0., 1.,  ..., 1., 0., 0.],
        [0., 0., 0.,  ..., 0., 0., 0.],
        [1., 0., 0.,  ..., 1., 0., 0.],
        ...,
        [1., 0., 1.,  ..., 0., 0., 0.],
        [0., 0., 0.,  ..., 0., 0., 0.],
        [0., 0., 0.,  ..., 0., 0., 0.]], device='cuda:0')

eg8:

                      2           64
loss = loss.view(anchor_count, batch_size)

tensor([[4.7642, 4.7700, 4.9274, 4.8298, 5.0163, 4.8054, 4.9344, 4.6297, 5.0533,
         4.8786, 4.8808, 4.8946, 4.5965, 4.9085, 4.6794, 4.9939, 4.8648, 4.8382,
         4.5422, 4.7529, 4.6383, 4.7940, 4.7202, 4.9732, 4.5696, 4.7187, 4.8346,
         4.8804, 4.5355, 4.7395, 4.8884, 4.7580, 5.0020, 4.9140, 5.2952, 4.7402,
         4.8660, 4.9400, 4.9015, 4.8370, 5.0518, 4.8339, 5.0241, 4.8498, 5.0187,
         4.6112, 4.6124, 4.7228, 4.8453, 4.6810, 4.7281, 4.7040, 4.8005, 5.0514,
         5.0573, 4.2868, 4.9171, 4.5031, 4.7733, 4.8827, 4.7193, 4.9463, 4.8855,
         4.9188],
        [4.7642, 4.7700, 4.9274, 4.8298, 5.0163, 4.8054, 4.9344, 4.6297, 5.0533,
         4.8786, 4.8808, 4.8946, 4.5965, 4.9085, 4.6794, 4.9939, 4.8648, 4.8382,
         4.5422, 4.7529, 4.6383, 4.7940, 4.7202, 4.9732, 4.5696, 4.7187, 4.8346,
         4.8804, 4.5355, 4.7395, 4.8884, 4.7580, 5.0020, 4.9140, 5.2952, 4.7402,
         4.8660, 4.9400, 4.9015, 4.8370, 5.0518, 4.8339, 5.0241, 4.8498, 5.0187,
         4.6112, 4.6124, 4.7228, 4.8453, 4.6810, 4.7281, 4.7040, 4.8005, 5.0514,
         5.0573, 4.2868, 4.9171, 4.5031, 4.7733, 4.8827, 4.7193, 4.9463, 4.8855,
         4.9188]], device='cuda:0', grad_fn=<ViewBackward>)

loss = loss.view(anchor_count, batch_size).mean()

tensor(4.8208, device='cuda:0', grad_fn=<MeanBackward0>)
 

loss.view(anchor_count, batch_size).mean(0)

tensor([4.7642, 4.7700, 4.9274, 4.8298, 5.0163, 4.8054, 4.9344, 4.6297, 5.0533,
        4.8786, 4.8808, 4.8946, 4.5965, 4.9085, 4.6794, 4.9939, 4.8648, 4.8382,
        4.5422, 4.7529, 4.6383, 4.7940, 4.7202, 4.9732, 4.5696, 4.7187, 4.8346,
        4.8804, 4.5355, 4.7395, 4.8884, 4.7580, 5.0020, 4.9140, 5.2952, 4.7402,
        4.8660, 4.9400, 4.9015, 4.8370, 5.0518, 4.8339, 5.0241, 4.8498, 5.0187,
        4.6112, 4.6124, 4.7228, 4.8453, 4.6810, 4.7281, 4.7040, 4.8005, 5.0514,
        5.0573, 4.2868, 4.9171, 4.5031, 4.7733, 4.8827, 4.7193, 4.9463, 4.8855,
        4.9188], device='cuda:0', grad_fn=<MeanBackward1>)

loss.view(anchor_count, batch_size).mean(1)

tensor([4.8208, 4.8208], device='cuda:0', grad_fn=<MeanBackward1>)
 

class SupConLoss(nn.Module):
    """Supervised Contrastive Learning: https://arxiv.org/pdf/2004.11362.pdf.
    It also supports the unsupervised contrastive loss in SimCLR"""
    def __init__(self, temperature=0.07, contrast_mode='all',
                 base_temperature=0.07):
        super(SupConLoss, self).__init__()
        self.temperature = temperature
        self.contrast_mode = contrast_mode
        self.base_temperature = base_temperature

    def forward(self, features, labels=None, mask=None):
        """Compute loss for model. If both `labels` and `mask` are None,
        it degenerates to SimCLR unsupervised loss:
        https://arxiv.org/pdf/2002.05709.pdf

        Args:
            features: hidden vector of shape [bsz, n_views, ...].
            labels: ground truth of shape [bsz].
            mask: contrastive mask of shape [bsz, bsz], mask_{i,j}=1 if sample j
                has the same class as sample i. Can be asymmetric.
        Returns:
            A loss scalar.
        """
        device = (torch.device('cuda')
                  if features.is_cuda
                  else torch.device('cpu'))

        if len(features.shape) < 3:
            raise ValueError('`features` needs to be [bsz, n_views, ...],'
                             'at least 3 dimensions are required')
        if len(features.shape) > 3:
            features = features.view(features.shape[0], features.shape[1], -1)

        batch_size = features.shape[0]
        if labels is not None and mask is not None:
            raise ValueError('Cannot define both `labels` and `mask`')
        elif labels is None and mask is None:
            mask = torch.eye(batch_size, dtype=torch.float32).to(device)
        elif labels is not None:
            labels = labels.contiguous().view(-1, 1)
            if labels.shape[0] != batch_size:
                raise ValueError('Num of labels does not match num of features')
            mask = torch.eq(labels, labels.T).float().to(device)
        else:
            mask = mask.float().to(device)

        contrast_count = features.shape[1]
        contrast_feature = torch.cat(torch.unbind(features, dim=1), dim=0)
        if self.contrast_mode == 'one':
            anchor_feature = features[:, 0]
            anchor_count = 1
        elif self.contrast_mode == 'all':
            anchor_feature = contrast_feature
            anchor_count = contrast_count
        else:
            raise ValueError('Unknown mode: {}'.format(self.contrast_mode))

        # compute logits
        anchor_dot_contrast = torch.div(
            torch.matmul(anchor_feature, contrast_feature.T),
            self.temperature)
        # for numerical stability
        logits_max, _ = torch.max(anchor_dot_contrast, dim=1, keepdim=True)
        logits = anchor_dot_contrast - logits_max.detach()

        # tile mask
        mask = mask.repeat(anchor_count, contrast_count)
        # mask-out self-contrast cases
        logits_mask = torch.scatter(
            torch.ones_like(mask),
            1,
            torch.arange(batch_size * anchor_count).view(-1, 1).to(device),
            0
        )
        mask = mask * logits_mask

        # compute log_prob
        exp_logits = torch.exp(logits) * logits_mask
        log_prob = logits - torch.log(exp_logits.sum(1, keepdim=True))

        # compute mean of log-likelihood over positive
        mean_log_prob_pos = (mask * log_prob).sum(1) / mask.sum(1)

        # loss
        loss = - (self.temperature / self.base_temperature) * mean_log_prob_pos
        loss = loss.view(anchor_count, batch_size).mean()

        return loss

评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包

打赏作者

程序猿的探索之路

你的鼓励将是我创作的最大动力

¥1 ¥2 ¥4 ¥6 ¥10 ¥20
扫码支付:¥1
获取中
扫码支付

您的余额不足,请更换扫码支付或充值

打赏作者

实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值