Utils

setup_parser

neuralkg_ind.utils.setup_parser.setup_parser()

Set up Python’s ArgumentParser with data, model, trainer, and other arguments.

tools

neuralkg_ind.utils.tools.import_class(module_and_class_name: str) → type

Import class from a module, e.g. ‘model.TransE’

neuralkg_ind.utils.tools.save_config(args)

neuralkg_ind.utils.tools.load_config(args, config_path)

neuralkg_ind.utils.tools.get_param(*shape)

neuralkg_ind.utils.tools.deserialize(data)

neuralkg_ind.utils.tools.deserialize_RMPI(data)

neuralkg_ind.utils.tools.set_logger(args)

Write logs to checkpoint and console

neuralkg_ind.utils.tools.log_metrics(epoch, metrics)

Print the evaluation logs

neuralkg_ind.utils.tools.log_step_metrics(step, metrics)

Print the evaluation logs for check_per_step

neuralkg_ind.utils.tools.override_config(args)

Override model and data configuration

neuralkg_ind.utils.tools.reidx_withr_ande(tri, rel_reidx, ent_reidx)

neuralkg_ind.utils.tools.reidx(tri)

neuralkg_ind.utils.tools.reidx_withr(tri, rel_reidx)

neuralkg_ind.utils.tools.data2pkl(dataset_name)

Store data in pickle

neuralkg_ind.utils.tools.gen_subgraph_datasets(args, splits=['train', 'valid'], saved_relation2id=None, max_label_value=None)

neuralkg_ind.utils.tools.load_ind_data_grail(args)

neuralkg_ind.utils.tools.load_data_grail(args, add_traspose_rels=False)

neuralkg_ind.utils.tools.get_average_subgraph_size(sample_size, links, A, params)

neuralkg_ind.utils.tools.serialize(data)

neuralkg_ind.utils.tools.sample_neg(adj_list, edges, num_neg_samples_per_link=1, max_size=1000000, constrained_neg_prob=0)

neuralkg_ind.utils.tools.get_edge_count(adj_list)

neuralkg_ind.utils.tools.intialize_worker(A, params, max_label_value)

neuralkg_ind.utils.tools.extract_save_subgraph(args_)

neuralkg_ind.utils.tools.links2subgraphs(A, graphs, params, max_label_value=None, testing=False)

extract enclosing subgraphs, write map mode + named dbs

neuralkg_ind.utils.tools.subgraph_extraction_labeling(ind, rel, A_list, h=1, enclosing_sub_graph=False, max_nodes_per_hop=None, max_node_label_value=None)

neuralkg_ind.utils.tools.node_label(subgraph, max_distance=1, enclosing_flag=False)

neuralkg_ind.utils.tools.remove_nodes(A_incidence, nodes)

neuralkg_ind.utils.tools.get_neighbor_nodes(roots, adj, h=1, max_nodes_per_hop=None)

neuralkg_ind.utils.tools.incidence_matrix(adj_list)

adj_list: List of sparse adjacency matrices

neuralkg_ind.utils.tools.bfs_relational(adj, roots, max_nodes_per_hop=None)

BFS for graphs. Modified from dgl.contrib.data.knowledge_graph to accomodate node sampling

neuralkg_ind.utils.tools.get_neighbors(adj, nodes)

Takes a set of nodes and a graph adjacency matrix and returns a set of neighbors. Directly copied from dgl.contrib.data.knowledge_graph

neuralkg_ind.utils.tools.sp_row_vec_from_idx_list(idx_list, dim)

Create sparse vector of dimensionality dim from a list of indices.

neuralkg_ind.utils.tools.ssp_multigraph_to_dgl(graph, n_feats=None)

Converting ssp multigraph (i.e. list of adjs) to dgl multigraph.

neuralkg_ind.utils.tools.gen_meta_subgraph_datasets(args)

neuralkg_ind.utils.tools.sample_one_subgraph(args, bg_train_g)

neuralkg_ind.utils.tools.get_average_meta_subgraph_size(args, sample_size, bg_train_g)

neuralkg_ind.utils.tools.get_g(tri_list)

neuralkg_ind.utils.tools.get_g_bidir(triples, args)

neuralkg_ind.utils.tools.get_hr2t_rt2h(tris)

neuralkg_ind.utils.tools.get_hr2t_rt2h_sup_que(sup_tris, que_tris)

neuralkg_ind.utils.tools.get_indtest_test_dataset_and_train_g(args)