示例#1
0
def graph_rewards_seq_len(filepaths):
    initrewards = []
    min_len = 10000000
    for f in filepaths:
        r = np.load(f)['values']
        mr = logger.moving_average(r, 5)
        if len(mr) < min_len:
            min_len = len(mr)
        initrewards.append(mr)

    rewards = []
    for r in initrewards:
        rewards.append(r[:min_len])

    r2 = plt.plot(rewards[0], label='length 2 sequence', color='orange')   
    r4 = plt.plot(rewards[1], label='length 4 sequence', color='crimson')  
    r8 = plt.plot(rewards[2], label='length 8 sequence', color='cyan')   
    r12 = plt.plot(rewards[3], label='length 12 sequence', color='brown')  
    r16 = plt.plot(rewards[4], label='length 16 sequence', color='blue')   
    r20 = plt.plot(rewards[5], label='length 20 sequence', color='black')  
    r24 = plt.plot(rewards[6], label='length 24 sequence', color='pink')   

    plt.legend(loc='lower right')
    plt.ylabel('Episode Rewards')
    plt.xlabel('Epochs')
    plt.savefig('/Users/wulfe/Dropbox/School/Stanford/winter_2016/cs239/project/hierarchical_rl/results/seqlen_rewards.png')
示例#2
0
def graph_rewards_seq_len(filepaths):
    initrewards = []
    min_len = 10000000
    for f in filepaths:
        r = np.load(f)['values']
        mr = logger.moving_average(r, 5)
        if len(mr) < min_len:
            min_len = len(mr)
        initrewards.append(mr)

    rewards = []
    for r in initrewards:
        rewards.append(r[:min_len])

    r2 = plt.plot(rewards[0], label='length 2 sequence', color='orange')
    r4 = plt.plot(rewards[1], label='length 4 sequence', color='crimson')
    r8 = plt.plot(rewards[2], label='length 8 sequence', color='cyan')
    r12 = plt.plot(rewards[3], label='length 12 sequence', color='brown')
    r16 = plt.plot(rewards[4], label='length 16 sequence', color='blue')
    r20 = plt.plot(rewards[5], label='length 20 sequence', color='black')
    r24 = plt.plot(rewards[6], label='length 24 sequence', color='pink')

    plt.legend(loc='lower right')
    plt.ylabel('Episode Rewards')
    plt.xlabel('Epochs')
    plt.savefig(
        '/Users/wulfe/Dropbox/School/Stanford/winter_2016/cs239/project/hierarchical_rl/results/seqlen_rewards.png'
    )
示例#3
0
def graph_rewards(filepaths):
    rewards = []
    for f in filepaths:
        r = np.load(f)['values']
        mr = logger.moving_average(r, 10)
        rewards.append(mr)

    plt.plot(rewards[0], label='row/col + room', color='r')  
    plt.plot(rewards[1], label='row/col only', color='g') 
    plt.plot(rewards[2], label='tabular', color='b')    
    plt.plot(rewards[3], label='coordinates', color='magenta') 

    plt.legend(loc='upper left')
    plt.ylabel('Episode Rewards')
    plt.xlabel('Epochs')
    plt.savefig('/Users/wulfe/Dropbox/School/Stanford/winter_2016/cs239/project/hierarchical_rl/results/staterep_rewards.png')
示例#4
0
def graph_rewards(filepaths):
    rewards = []
    for f in filepaths:
        r = np.load(f)['values']
        mr = logger.moving_average(r, 10)
        rewards.append(mr)

    plt.plot(rewards[0], label='row/col + room', color='r')
    plt.plot(rewards[1], label='row/col only', color='g')
    plt.plot(rewards[2], label='tabular', color='b')
    plt.plot(rewards[3], label='coordinates', color='magenta')

    plt.legend(loc='upper left')
    plt.ylabel('Episode Rewards')
    plt.xlabel('Epochs')
    plt.savefig(
        '/Users/wulfe/Dropbox/School/Stanford/winter_2016/cs239/project/hierarchical_rl/results/staterep_rewards.png'
    )