mf.py

#!/usr/bin/env python2

from __future__ import print_function
import tensorflow as tf
import argparse
from antk.core import config
from antk.core import generic_model
from antk.core import loader
from antk.models import mfmodel


def return_parser():
    parser = argparse.ArgumentParser(description="For testing")
    parser.add_argument("datadir", metavar="DATA_DIRECTORY", type=str,
                        help="The directory where train, dev, and test data resides. ")
    parser.add_argument("config", metavar="CONFIG", type=str,
                        help="The config file for building the ant architecture.")
    return parser

if __name__ == '__main__':

    args = return_parser().parse_args()
    data = loader.read_data_sets(args.datadir, hashlist=['item', 'user', 'ratings'])
    data.train.labels['ratings'] = loader.center(data.train.labels['ratings'])
    data.dev.labels['ratings'] = loader.center(data.dev.labels['ratings'])
    x = mfmodel.mf(data, args.config, lamb=0.01, kfactors=10, verbose=True)