Functions — github.com/StatsDLMathsRecomSys/Inductive-representation-learning-on-temporal-graphs

CodeHub MCPcopy Create free account

hub / github.com/StatsDLMathsRecomSys/Inductive-representation-learning-on-temporal-graphs / functions

Functions41 in github.com/StatsDLMathsRecomSys/Inductive-representation-learning-on-temporal-graphs

⨍ Functions 41 ◇ Types & classes 15

↓ 10 callersMethod__init__

(self, ngh_finder, n_feat, e_feat, attn_mode='prod', use_time='time', agg_method='attn', node

↓ 7 callersMethodtem_conv

(self, src_idx_l, cut_time_l, curr_layers, num_neighbors=20)

↓ 4 callersFunctioneval_one_epoch

(hint, tgan, sampler, src, dst, ts, label)

learn_edge.py:86

↓ 3 callersFunctioneval_epoch

(src_l, dst_l, ts_l, label_l, batch_size, lr_model, tgan, num_layer=NODE_LAYER)

learn_node.py:205

↓ 3 callersMethodget_temporal_neighbor

Params ------ src_idx_l: List[int] cut_time_l: List[float], num_neighbors: int

↓ 3 callersMethodsample

↓ 2 callersMethodcontrast

(self, src_idx_l, target_idx_l, background_idx_l, cut_time_l, num_neighbors=20)

↓ 1 callersMethodearly_stop_check

(self, curr_val)

↓ 1 callersMethodfind_before

Params ------ src_idx: int cut_time: float

↓ 1 callersMethodforward

(self, src_idx_l, target_idx_l, cut_time_l, num_neighbors=20)

↓ 1 callersMethodinit_off_set

Params ------ adj_list: List[List[int]]

↓ 1 callersFunctionpreprocess

↓ 1 callersFunctionreindex

↓ 1 callersFunctionrun

(self, max_round=3, higher_better=True, tolerance=1e-3)

(self, src_list, dst_list)

(self, dim, drop=0.3)

learn_node.py:20

Params ------ node_idx_l: List[int] node_ts_l: List[int] off_set_l: List[int], such that node_idx_l[off_set_l

(self, dim1, dim2, dim3, dim4)

(self, temperature, attn_dropout=0.1)

(self, n_head, d_model, d_k, d_v, dropout=0.1)

(self, n_head, d_model, d_k, d_v, dropout=0.1)

(self, expand_dim, factor=5)

(self, expand_dim, seq_len)

(self, expand_dim)

(self, feat_dim, edge_dim, time_dim)

(self, feat_dim, edge_dim)

args: feat_dim: dim for the node features edge_dim: dim for the temporal edge features time_dim: dim for the ti

Functionexpand_last_dim

Methodfind_k_hop

Sampling the k-hop sub graph

learn_node.py:28

(self, q, k, v, mask=None)

(self, q, k, v, mask=None)

(self, q, k, v, mask=None)

(self, src, src_t, seq, seq_t, seq_e, mask)

(self, src, src_t, seq, seq_t, seq_e, mask)

Attention based temporal attention forward pass args: src: float Tensor of shape [B, D] src_t: float Tensor of shape [B, D