def get_feed_dict(self, corpus, data, batch_start, batch_size, phase):
feed_dict = BPR.get_feed_dict(self, corpus, data, batch_start,
batch_size, phase)
real_batch_size = feed_dict['batch_size']
times = data['time'][batch_start:batch_start + real_batch_size].values
time_ids = (times - self.min_time) // self.time_bin_width
feed_dict['time_id'] = utils.numpy_to_torch(time_ids).long()
return feed_dict
def parse_model_args(parser):
parser.add_argument('--layers',
type=str,
default='[64]',
help="Size of each layer.")
return BPR.parse_model_args(parser)
def parse_model_args(parser):
parser.add_argument('--time_bin',
type=int,
default=100,
help='Number of time bins.')
return BPR.parse_model_args(parser)
def _define_params(self):
BPR._define_params(self)
self.u_t_embeddings = torch.nn.Embedding(self.time_bin, self.emb_size)
self.i_t_embeddings = torch.nn.Embedding(self.time_bin, self.emb_size)
self.embeddings.extend(['u_t_embeddings', 'i_t_embeddings'])
def __init__(self, args, corpus):
self.time_bin = args.time_bin
self.min_time = corpus.min_time
self.time_bin_width = (corpus.max_time - self.min_time +
1.) / self.time_bin
BPR.__init__(self, args, corpus)
# -*- coding: utf-8 -*-
from models.BPR import BPR
from readers.naisdataloader import Dataloader
from configs.config import Config
config = Config()
dl = Dataloader(config)
bpr = BPR(config, dl)
bpr.train_and_evaluate()
def __init__(self, args, corpus):
self.layers = eval(args.layers)
self.dropout = args.dropout
BPR.__init__(self, args, corpus)