Python torch.eq方法代码示例

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def forward(self, feed_dict):
        feed_dict = GView(feed_dict)
        feature_f = self._extract_sent_feature(feed_dict.sent_f, feed_dict.sent_f_length, self.gru_f)
        feature_b = self._extract_sent_feature(feed_dict.sent_b, feed_dict.sent_b_length, self.gru_b)
        feature_img = feed_dict.image
        
        feature = torch.cat([feature_f, feature_b, feature_img], dim=1)
        predict = self.predict(feature)

        if self.training:
            label = self.embedding(feed_dict.label)
            loss = cosine_loss(predict, label).mean()
            return loss, {}, {}
        else:
            output_dict = dict(pred=predict)
            if 'label' in feed_dict:
                dis = cosine_distance(predict, self.embedding.weight)
                _, topk = dis.topk(1000, dim=1, sorted=True)
                for k in [1, 10, 100, 1000]:
                    output_dict['top{}'.format(k)] = torch.eq(topk, feed_dict.label.unsqueeze(-1))[:, :k].float().sum(dim=1).mean()
            return output_dict 

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def _compute_xi(self, s, aug, y):

        # find argmax of augmented scores
        _, y_star = torch.max(aug, 1)
        # xi_max: one-hot encoding of maximal indices
        xi_max = torch.eq(y_star[:, None], self._range).float()

        if MultiClassHingeLoss.smooth:
            # find smooth argmax of scores
            xi_smooth = nn.functional.softmax(s, dim=1)
            # compute for each sample whether it has a positive contribution to the loss
            losses = torch.sum(xi_smooth * aug, 1)
            mask_smooth = torch.ge(losses, 0).float()[:, None]
            # keep only smoothing for positive contributions
            xi = mask_smooth * xi_smooth + (1 - mask_smooth) * xi_max
        else:
            xi = xi_max

        return xi 

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def update(self, output):
        y_pred, y = output
        num_classes = y_pred.size(1)
        indices = torch.max(y_pred, 1)[1]
        correct = torch.eq(indices, y)
        actual_onehot = to_onehot(y, num_classes)
        actual = actual_onehot.sum(dim=0)
        if correct.sum() == 0:
            true_positives = torch.zeros_like(actual)
        else:
            correct_onehot = to_onehot(indices[correct], num_classes)
            true_positives = correct_onehot.sum(dim=0)
        if self._actual is None:
            self._actual = actual
            self._true_positives = true_positives
        else:
            self._actual += actual
            self._true_positives += true_positives 

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def update(self, output):
        y_pred, y = output
        num_classes = y_pred.size(1)
        indices = torch.max(y_pred, 1)[1]
        correct = torch.eq(indices, y)
        pred_onehot = to_onehot(indices, num_classes)
        all_positives = pred_onehot.sum(dim=0)
        if correct.sum() == 0:
            true_positives = torch.zeros_like(all_positives)
        else:
            correct_onehot = to_onehot(indices[correct], num_classes)
            true_positives = correct_onehot.sum(dim=0)
        if self._all_positives is None:
            self._all_positives = all_positives
            self._true_positives = true_positives
        else:
            self._all_positives += all_positives
            self._true_positives += true_positives 

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def test_train(self):
        self._metric.train()
        calls = [[torch.FloatTensor([0.0]), torch.LongTensor([0])],
                 [torch.FloatTensor([0.0, 0.1, 0.2, 0.3]), torch.LongTensor([0, 1, 2, 3])]]
        for i in range(len(self._states)):
            self._metric.process(self._states[i])
        self.assertEqual(2, len(self._metric_function.call_args_list))
        for i in range(len(self._metric_function.call_args_list)):
            self.assertTrue(torch.eq(self._metric_function.call_args_list[i][0][0], calls[i][0]).all)
            self.assertTrue(torch.lt(torch.abs(torch.add(self._metric_function.call_args_list[i][0][1], -calls[i][1])), 1e-12).all)
        self._metric_function.reset_mock()
        self._metric.process_final({})

        self.assertEqual(self._metric_function.call_count, 1)
        self.assertTrue(torch.eq(self._metric_function.call_args_list[0][0][1], torch.LongTensor([0, 1, 2, 3, 4])).all)
        self.assertTrue(torch.lt(torch.abs(torch.add(self._metric_function.call_args_list[0][0][0], -torch.FloatTensor([0.0, 0.1, 0.2, 0.3, 0.4]))), 1e-12).all) 

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def test_forward(self, hidden_sizes):
        env_spec = GarageEnv(DummyBoxEnv())
        obs_dim = env_spec.observation_space.flat_dim
        act_dim = env_spec.action_space.flat_dim
        obs = torch.ones(obs_dim, dtype=torch.float32).unsqueeze(0)
        act = torch.ones(act_dim, dtype=torch.float32).unsqueeze(0)

        qf = ContinuousMLPQFunction(env_spec=env_spec,
                                    hidden_nonlinearity=None,
                                    hidden_sizes=hidden_sizes,
                                    hidden_w_init=nn.init.ones_,
                                    output_w_init=nn.init.ones_)

        output = qf(obs, act)
        expected_output = torch.full([1, 1],
                                     fill_value=(obs_dim + act_dim) *
                                     np.prod(hidden_sizes),
                                     dtype=torch.float32)
        assert torch.eq(output, expected_output)

    # yapf: disable 

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def test_is_pickleable(self, hidden_sizes):
        env_spec = GarageEnv(DummyBoxEnv())
        obs_dim = env_spec.observation_space.flat_dim
        act_dim = env_spec.action_space.flat_dim
        obs = torch.ones(obs_dim, dtype=torch.float32).unsqueeze(0)
        act = torch.ones(act_dim, dtype=torch.float32).unsqueeze(0)

        qf = ContinuousMLPQFunction(env_spec=env_spec,
                                    hidden_nonlinearity=None,
                                    hidden_sizes=hidden_sizes,
                                    hidden_w_init=nn.init.ones_,
                                    output_w_init=nn.init.ones_)

        output1 = qf(obs, act)

        p = pickle.dumps(qf)
        qf_pickled = pickle.loads(p)
        output2 = qf_pickled(obs, act)

        assert torch.eq(output1, output2) 

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def test_update_context(self):
        """Test update_context."""
        s = TimeStep(env_spec=self.env_spec,
                     observation=np.ones(self.obs_dim),
                     next_observation=np.ones(self.obs_dim),
                     action=np.ones(self.action_dim),
                     reward=1.0,
                     terminal=False,
                     env_info={},
                     agent_info={})
        updates = 10
        for _ in range(updates):
            self.module.update_context(s)
        assert torch.all(
            torch.eq(self.module.context,
                     torch.ones(updates, self.encoder_input_dim))) 

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def test_encode_param():
    param = torch.rand(256, 128, 3, 3)
    prune_vanilla_elementwise(sparsity=0.7, param=param)
    quantize_linear_fix_zeros(param, k=16)
    huffman = EncodedParam(param=param, method='huffman',
                           encode_indices=True, bit_length_zero_run_length=4)
    stats = huffman.stats
    print(stats)
    assert torch.eq(param, huffman.data).all()
    state_dict = huffman.state_dict()
    huffman = EncodedParam()
    huffman.load_state_dict(state_dict)
    assert torch.eq(param, huffman.data).all()
    vanilla = EncodedParam(param=param, method='vanilla',
                           encode_indices=True, bit_length_zero_run_length=4)
    stats = vanilla.stats
    print(stats)
    assert torch.eq(param, vanilla.data).all()
    quantize_fixed_point(param=param, bit_length=4, bit_length_integer=0)
    fixed_point = EncodedParam(param=param, method='fixed_point',
                               bit_length=4, bit_length_integer=0,
                               encode_indices=True, bit_length_zero_run_length=4)
    stats = fixed_point.stats
    print(stats)
    assert torch.eq(param, fixed_point.data).all() 

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def test_data_pickle_correctness(self):
        # this will create new pickle files for train, valid, test
        dataset = Dataset.create(
            config=self.config, folder=self.dataset_folder, preload_data=True
        )

        # create new dataset which loads the triples from stored pckl files
        dataset_load_by_pickle = Dataset.create(
            config=self.config, folder=self.dataset_folder, preload_data=True
        )
        for split in dataset._triples.keys():
            self.assertTrue(
                torch.all(
                    torch.eq(dataset_load_by_pickle.split(split), dataset.split(split))
                )
            )
        self.assertEqual(dataset._meta, dataset_load_by_pickle._meta) 

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def decode(self, input_word_orig, input_word, input_char, _, target=None, mask=None, length=None, hx=None,
               leading_symbolic=0):
        if len(input_word.size()) == 3:
            # input_word is the packed sents [n_sent, sent_len]
            input_word, input_char, target, sent_mask, length, doc_n_sent = self._doc2sent(
                input_word, input_char, target)
        # output from rnn [batch, length, tag_space]
        output, _, mask, length = self._get_rnn_output(input_word_orig, input_word, input_char, mask=mask,
                                                       length=length, hx=hx)

        if target is None:
            return self.crf.decode(output, mask=mask, leading_symbolic=leading_symbolic), None

        if length is not None:
            max_len = length.max()
            target = target[:, :max_len]

        preds = self.crf.decode(output, mask=mask, leading_symbolic=leading_symbolic)
        if mask is None:
            return preds, torch.eq(preds, target).float().sum()
        else:
            return preds, (torch.eq(preds, target).float() * mask).sum() 

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def query(self, x):
        '''
        :param x: input data CxN tensor
        :return: mask: Nxnode_num
        '''
        # expand as CxNxnode_num
        node = self.node.unsqueeze(1).expand(x.size(0), x.size(1), self.rows * self.cols)
        x_expanded = x.unsqueeze(2).expand_as(node)

        # calcuate difference between x and each node
        diff = x_expanded - node  # CxNxnode_num
        diff_norm = (diff ** 2).sum(dim=0)  # Nxnode_num

        # find the nearest neighbor
        _, min_idx = torch.min(diff_norm, dim=1)  # N
        min_idx_expanded = min_idx.unsqueeze(1).expand(min_idx.size()[0], self.rows * self.cols).float()  # Nxnode_num

        node_idx_list = self.node_idx_list.unsqueeze(0).expand_as(min_idx_expanded)  # Nxnode_num
        mask = torch.eq(min_idx_expanded, node_idx_list).float()  # Nxnode_num
        mask_row_max, _ = torch.max(mask, dim=0)  # node_num, this indicates whether the node has nearby x

        return mask, mask_row_max 

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def update(self, output):

        y_pred, y = self._check_shape(output)
        self._check_type((y_pred, y))

        if self._type == "binary":
            correct = torch.eq(y_pred.type(y.type()), y).view(-1)
        elif self._type == "multiclass":
            indices = torch.max(y_pred, dim=1)[1]
            correct = torch.eq(indices, y).view(-1)
        elif self._type == "multilabel":
            # if y, y_pred shape is (N, C, ...) -> (N x ..., C)
            num_classes = y_pred.size(1)
            last_dim = y_pred.ndimension()
            y_pred = torch.transpose(y_pred, 1, last_dim - 1).reshape(-1, num_classes)
            y = torch.transpose(y, 1, last_dim - 1).reshape(-1, num_classes)
            correct = torch.all(y == y_pred.type_as(y), dim=-1)

        self._num_correct += torch.sum(correct).item()
        self._num_examples += correct.shape[0] 

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def get_accuracy_bin(scores, labels):
    preds = torch.ge(scores, 0).long()
    acc = torch.eq(preds, labels).float()
    return torch.sum(acc) / labels.nelement() 

# 需要导入模块: import torch [as 别名]
# 或者: from torch import eq [as 别名]
def get_accuracy(scores, labels):
    _,preds = torch.max(scores, dim=-1)
    acc = torch.eq(preds, labels).float()
    return torch.sum(acc) / labels.nelement()