(self, dataset, data_path="data", task_type="full")
| 8 | class Sampler: |
| 9 | """Sampling the input graph data.""" |
| 10 | def __init__(self, dataset, data_path="data", task_type="full"): |
| 11 | self.dataset = dataset |
| 12 | self.data_path = data_path |
| 13 | (self.adj, |
| 14 | self.train_adj, |
| 15 | self.features, |
| 16 | self.train_features, |
| 17 | self.labels, |
| 18 | self.idx_train, |
| 19 | self.idx_val, |
| 20 | self.idx_test, |
| 21 | self.degree, |
| 22 | self.learning_type) = data_loader(dataset, data_path, "NoNorm", False, task_type) |
| 23 | |
| 24 | #convert some data to torch tensor ---- may be not the best practice here. |
| 25 | self.features = torch.FloatTensor(self.features).float() |
| 26 | self.train_features = torch.FloatTensor(self.train_features).float() |
| 27 | # self.train_adj = self.train_adj.tocsr() |
| 28 | |
| 29 | self.labels_torch = torch.LongTensor(self.labels) |
| 30 | self.idx_train_torch = torch.LongTensor(self.idx_train) |
| 31 | self.idx_val_torch = torch.LongTensor(self.idx_val) |
| 32 | self.idx_test_torch = torch.LongTensor(self.idx_test) |
| 33 | |
| 34 | # vertex_sampler cache |
| 35 | # where return a tuple |
| 36 | self.pos_train_idx = np.where(self.labels[self.idx_train] == 1)[0] |
| 37 | self.neg_train_idx = np.where(self.labels[self.idx_train] == 0)[0] |
| 38 | # self.pos_train_neighbor_idx = np.where |
| 39 | |
| 40 | |
| 41 | self.nfeat = self.features.shape[1] |
| 42 | self.nclass = int(self.labels.max().item() + 1) |
| 43 | self.trainadj_cache = {} |
| 44 | self.adj_cache = {} |
| 45 | #print(type(self.train_adj)) |
| 46 | self.degree_p = None |
| 47 | |
| 48 | def _preprocess_adj(self, normalization, adj, cuda): |
| 49 | adj_normalizer = fetch_normalization(normalization) |
nothing calls this directly
no test coverage detected