STAMP :param hidden_size: embedding size :param item_num: the number of item :param padding_idx: padding idx

calculate similarity between current session and neighbourhood session :param s: current session :param m: candidate neighbourhood sessio

get DCG@k :param y_pre: numpy (batch_size,x) :param y_true: y_truth: list[batch_size][ground_truth_num] :param k: k :return: DCG@

divid and extend negative samples

get DCG value, the k is determined by the size of pred :param pred: numpy (batch_size,k) :param y_truth: list[batch_size][ground_truth_nu

pred: numpy type(batch_size,k) y_truth: list type (batch_size,groudtruth_num)

calculate hit rate :param predict: predict, (batch_size, k) :param ground_truth: ground truth, (batch_size, ) :return: hr

calculate mrr :param predict: predict, (batch_size, k) :param ground_truth: ground truth, (batch_size, ) :return: mrr score

calculate ndcg :param predict: predict, (batch_size, k) :param ground_truth: ground truth, (batch_size, ) :return: ndcg score

pred: numpy type(batch_size,k) y_truth: list type (batch_size,groudtruth_num)

get hit num, HR@k the k is determined by the size of pred :param pred: numpy (batch_size,k) :param y_truth: list[batch_size][ground_truth

get NDCG value,NDCG@k the k is determined by the size of pred :param pred: numpy (batch_size,k) :param y_truth: list[batch_size][ground_t

get MRR value, MRR@k the k is determined by the size of pred :param pred: numpy (batch_size,k) :param y_truth: list[batch_size][ground_tr

padding