run_training.py

import os
import sys
import argparse


if "./lib" not in sys.path:
    sys.path.append("./lib")

from DQL import *





if __name__== "__main__":


    parser = argparse.ArgumentParser(formatter_class=argparse.RawTextHelpFormatter)
    parser = argparse.ArgumentParser(description='Train an object localizer')

    parser.add_argument('-n','--num_episodes', type=int, default=5, help = "Number of episodes that the agect can interact with an image. Default: 5")
    parser.add_argument('-rms','--replay_memory_size', type=int, default=500000, help = "Number of the most recent experiences that would be stored. Default: 500000")
    parser.add_argument('-rmis','--replay_memory_init_size', type=int, default=500, help = "Number of experiences to initialize replay memory. Default: 500")
    parser.add_argument('-u','--update_target_estimator_every', type=int, default=10000, help = "Number of steps after which estimator parameters are copied to target network. Default: 10000")
    parser.add_argument('-d','--discount_factor', type=int, default=0.99, help = "Discount factor. Default: 0.99")
    parser.add_argument('-es','--epsilon_start', type=int, default=1.0, help = "Epsilon decay schedule start point. Default: 1.0")
    parser.add_argument('-ee','--epsilon_end', type=int, default=0.2, help="Epsilon decay schedule end point. Default: 0.2")
    parser.add_argument('-ed','--epsilon_decay_steps', type=int, default=500, help="Epsilon decay step rate. This number indicates epsilon would be decearsed from start to end point after how many steps. Default: 500")
    #parser.add_argument('-b','--batch_size', type=int, default=32, help="Epsilon decay schedule end point. Default: 0.2")
    parser.add_argument('-c','--category', type=str, nargs='+', default=['cat'], help='Indicating the categories are going to be used for training. You can list name of the classes you want to use in training. If you wish to use all classes then you can use *. For instnce <-c cat dog>. Default: cat')
    parser.add_argument('-m','--model_name', type=str, default='default_model', help='The trained model would be saved with this name under the path ../experiments/model_name. Default: default_model')
    #parser.add_argument('-co','--coco', type=bool, default=False)

    args = parser.parse_args()


    DQL(args.num_episodes,
        args.replay_memory_size,
        args.replay_memory_init_size,
        args.update_target_estimator_every,
        args.discount_factor,
        args.epsilon_start,
        args.epsilon_end,
        args.epsilon_decay_steps,
        args.category,
        args.model_name)