Source code for PlanetAlign.algorithms.meaformer.main
import time
import torch
import torch.nn.functional as F
from torch_geometric.utils import to_dense_adj
from PlanetAlign.data import Dataset
from PlanetAlign.utils import get_anchor_pairs
from PlanetAlign.metrics import hits_ks_scores, mrr_score
from PlanetAlign.algorithms.base_model import BaseModel
[docs]
class MEAformer(BaseModel):
"""
Embedding-based method MEAformer for pairwise entity alignment. DualMatch is proposed by the paper
"`MEAformer: Multi-modal Entity Alignment Transformer for Meta Modality Hybrid <https://arxiv.org/pdf/2212.14454>`_"
in ACM MM 2023.
"""
def __init__(self,
dtype: torch.dtype = torch.float32):
super(MEAformer, self).__init__(dtype=dtype)
def train(self,
dataset: Dataset,
gid1: int,
gid2: int,
use_attr: bool = True,
save_log: bool = True,
verbose: bool = True):
pass