argparser.py

import os
import torch
import random
import numpy as np
import argparse

def isfloat(value):
  try:
    float(value)
    return True
  except ValueError:
    return False

def isint(value):
  try:
    int(value)
    return True
  except ValueError:
    return False

def argparser(seed = 2019):

    parser = argparse.ArgumentParser()

    # configure file 
    parser.add_argument('--config', default="UNSPECIFIED.json")
    parser.add_argument('--model_subset', type=int, nargs='+', 
            help='Use only a subset of models in config file. Pass a list of numbers starting with 0, like --model_subset 0 1 3 5')
    parser.add_argument('--path_prefix', type=str, default="", help="override path prefix")
    parser.add_argument('--seed', type=int, default=seed)
    parser.add_argument('overrides', type=str, nargs='*',
                                help='overriding config dict')
    
    args = parser.parse_args()

    torch.manual_seed(args.seed)
    torch.cuda.manual_seed(args.seed)
    random.seed(args.seed)
    np.random.seed(args.seed)

    # for dual norm computation, we will have 1 / 0.0 = inf
    np.seterr(divide='ignore')

    overrides_dict = {}
    for o in args.overrides:
        key, val = o.strip().split("=")
        d = overrides_dict
        last_key = key
        if ":" in key:
            keys = key.split(":")
            for k in keys[:-1]:
                if k not in d:
                    d[k] = {}
                d = d[k]
            last_key = keys[-1]
        if val == "true":
            val = True
        elif val == "false":
            val = False
        elif isint(val):
            val = int(val)
        elif isfloat(val):
            val = float(val)
        d[last_key] = val
    args.overrides_dict = overrides_dict

    return args