def __init__(self, mark, epsilon=0.1, step_size=0.1, td_lambda=0.6):
'''
初期化
arguments:
どちらのマークのプレイヤーとして行動するか
ソフト方策用のepsilon(前回の実装ではstaticに定義していた)
TD(λ)のステップ数
TD(λ)のλ
'''
self.mark = mark
self.epsilon = epsilon
self.step_size = step_size
self.td_lambda = td_lambda
# tensorflow Session
self.sess = tf.InteractiveSession()
# input placeholder
self.x = tf.placeholder(tf.float32, shape=[None, 9])
self.logits = model.inference(self.x, str(mark.to_int()))
# initialize
init_op = tf.initialize_all_variables()
self.sess.run(init_op)
# 価値テーブルを保持(Comを保存できるようにする)
# stateはhashableにしてvalueのkeyにする
self.value = defaultdict(lambda: 0.0)
# TD(λ)で使う直前の状態
self.previous_state = None
# TD(λ)で使う現在の状態
self.current_state = None
# 各状態の適格度
self.accumulated_weights = defaultdict(lambda: 0.0)
self.training = True
self.verbose = False
import model_mlp
NUM_CLASSES = 1
IMAGE_SIZE = 8
IMAGE_PIXELS = IMAGE_SIZE*IMAGE_SIZE*3
# 学習時のbatch size
BATCH_SIZE = 1
if __name__ == '__main__':
filename_queue = tf.train.string_input_producer(["data/airquality.csv"])
feature, data = load.mini_batch(filename_queue, 1)
data_placeholder = tf.placeholder("float", shape=(None, 5))
logits_maru = model_mlp.inference(data_placeholder, 'maru')
logits_batsu = model_mlp.inference(data_placeholder, 'batsu')
sess = tf.InteractiveSession()
saver = tf.train.Saver()
sess.run(tf.initialize_all_variables())
# restore trained model
ckpt = tf.train.get_checkpoint_state('train')
print(ckpt.model_checkpoint_path)
if ckpt and ckpt.model_checkpoint_path:
print("Load checkpint.")
saver.restore(sess, ckpt.model_checkpoint_path)
global_step = ckpt.model_checkpoint_path.split('/')[-1].split('-')[-1]
else:
print('No checkpoint file found.')
import model_mlp
NUM_CLASSES = 1
IMAGE_SIZE = 8
IMAGE_PIXELS = IMAGE_SIZE * IMAGE_SIZE * 3
# 学習時のbatch size
BATCH_SIZE = 1
if __name__ == '__main__':
filename_queue = tf.train.string_input_producer(["data/airquality.csv"])
feature, data = load.mini_batch(filename_queue, 1)
data_placeholder = tf.placeholder("float", shape=(None, 5))
logits_maru = model_mlp.inference(data_placeholder, 'maru')
logits_batsu = model_mlp.inference(data_placeholder, 'batsu')
sess = tf.InteractiveSession()
saver = tf.train.Saver()
sess.run(tf.initialize_all_variables())
# restore trained model
ckpt = tf.train.get_checkpoint_state('train')
print(ckpt.model_checkpoint_path)
if ckpt and ckpt.model_checkpoint_path:
print("Load checkpint.")
saver.restore(sess, ckpt.model_checkpoint_path)
global_step = ckpt.model_checkpoint_path.split('/')[-1].split('-')[-1]
else:
print('No checkpoint file found.')
def train():
'''
Train CNN_tiny for a number of steps.
'''
with tf.Graph().as_default():
# globalなstep数
global_step = tf.Variable(0, trainable=False)
# 教師データ
filename_queue = tf.train.string_input_producer(
["data/airquality.csv"])
datas, targets = load.mini_batch(filename_queue, BATCH_SIZE)
# placeholder
x = tf.placeholder(tf.float32, shape=[None, 5])
y = tf.placeholder(tf.float32, shape=[None, 1])
# graphのoutput
logits = model.inference(x)
debug_value = model.debug(logits)
# loss graphのoutputとlabelを利用
loss = model.loss(logits, y)
# 学習オペレーション
train_op = op.train(loss, global_step)
# saver
saver = tf.train.Saver(tf.all_variables())
# サマリー
summary_op = tf.merge_all_summaries()
# 初期化オペレーション
init_op = tf.initialize_all_variables()
# Session
sess = tf.Session(config=tf.ConfigProto(
log_device_placement=LOG_DEVICE_PLACEMENT))
sess.run(init_op)
print("settion start.")
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(sess=sess, coord=coord)
# サマリーのライターを設定
summary_writer = tf.train.SummaryWriter(TRAIN_DIR,
graph_def=sess.graph_def)
# model名
model_name = '/model%s.ckpt' % (tdatetime.strftime('%Y%m%d%H%M%S'))
# max_stepまで繰り返し学習
for step in xrange(MAX_STEPS):
start_time = time.time()
a, b = sess.run([datas, targets])
_, loss_value, predict_value = sess.run(
[train_op, loss, debug_value], feed_dict={
x: a,
y: b
})
duration = time.time() - start_time
assert not np.isnan(loss_value), 'Model diverged with loss = NaN'
# 100回ごと
if step % 100 == 0:
# stepごとの事例数 = mini batch size
num_examples_per_step = BATCH_SIZE
# 1秒ごとの事例数
examples_per_sec = num_examples_per_step / duration
# バッチごとの時間
sec_per_batch = float(duration)
# time, step数, loss, 1秒で実行できた事例数, バッチあたりの時間
format_str = '$s: step %d, loss = %.2f (%.1f examples/sec; %.3f sec/batch)'
print str(datetime.now()) + ': step' + str(
step) + ', loss= ' + str(loss_value) + ' ' + str(
examples_per_sec) + ' examples/sec; ' + str(
sec_per_batch) + ' sec/batch'
print "x", a
print "ground truth:", b
print "predict: ", predict_value
# 100回ごと
if step % 100 == 0:
pass
#summary_str = sess.run(summary_op)
# サマリーに書き込む
#summary_writer.add_summary(summary_str, step)
if step % 1000 == 0 or (step * 1) == MAX_STEPS:
checkpoint_path = TRAIN_DIR + model_name
saver.save(sess, checkpoint_path, global_step=step)
coord.request_stop()
coord.join(threads)
sess.close()
def train():
'''
Train CNN_tiny for a number of steps.
'''
with tf.Graph().as_default():
# globalなstep数
global_step = tf.Variable(0, trainable=False)
# 教師データ
filename_queue = tf.train.string_input_producer(["data/airquality.csv"])
datas, targets = load.mini_batch(filename_queue, BATCH_SIZE)
# placeholder
x = tf.placeholder(tf.float32, shape=[None, 5])
y = tf.placeholder(tf.float32, shape=[None, 1])
# graphのoutput
logits = model.inference(x)
debug_value = model.debug(logits)
# loss graphのoutputとlabelを利用
loss = model.loss(logits, y)
# 学習オペレーション
train_op = op.train(loss, global_step)
# saver
saver = tf.train.Saver(tf.all_variables())
# サマリー
summary_op = tf.merge_all_summaries()
# 初期化オペレーション
init_op = tf.initialize_all_variables()
# Session
sess = tf.Session(config=tf.ConfigProto(log_device_placement=LOG_DEVICE_PLACEMENT))
sess.run(init_op)
print("settion start.")
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(sess=sess, coord=coord)
# サマリーのライターを設定
summary_writer = tf.train.SummaryWriter(TRAIN_DIR, graph_def=sess.graph_def)
# model名
model_name = '/model%s.ckpt' % (tdatetime.strftime('%Y%m%d%H%M%S'))
# max_stepまで繰り返し学習
for step in xrange(MAX_STEPS):
start_time = time.time()
a, b = sess.run([datas, targets])
_, loss_value, predict_value = sess.run([train_op, loss, debug_value], feed_dict={x: a, y: b})
duration = time.time() - start_time
assert not np.isnan(loss_value), 'Model diverged with loss = NaN'
# 100回ごと
if step % 100 == 0:
# stepごとの事例数 = mini batch size
num_examples_per_step = BATCH_SIZE
# 1秒ごとの事例数
examples_per_sec = num_examples_per_step / duration
# バッチごとの時間
sec_per_batch = float(duration)
# time, step数, loss, 1秒で実行できた事例数, バッチあたりの時間
format_str = '$s: step %d, loss = %.2f (%.1f examples/sec; %.3f sec/batch)'
print str(datetime.now()) + ': step' + str(step) + ', loss= '+ str(loss_value) + ' ' + str(examples_per_sec) + ' examples/sec; ' + str(sec_per_batch) + ' sec/batch'
print "x", a
print "ground truth:", b
print "predict: ", predict_value
# 100回ごと
if step % 100 == 0:
pass
#summary_str = sess.run(summary_op)
# サマリーに書き込む
#summary_writer.add_summary(summary_str, step)
if step % 1000 == 0 or (step * 1) == MAX_STEPS:
checkpoint_path = TRAIN_DIR + model_name
saver.save(sess, checkpoint_path, global_step=step)
coord.request_stop()
coord.join(threads)
sess.close()
import model_mlp
NUM_CLASSES = 1
IMAGE_SIZE = 8
IMAGE_PIXELS = IMAGE_SIZE*IMAGE_SIZE*3
# 学習時のbatch size
BATCH_SIZE = 1
if __name__ == '__main__':
filename_queue = tf.train.string_input_producer(["data/airquality.csv"])
feature, data = load.mini_batch(filename_queue, 1)
data_placeholder = tf.placeholder("float", shape=(None, 5))
logits = model_mlp.inference(data_placeholder)
sess = tf.InteractiveSession()
saver = tf.train.Saver()
sess.run(tf.initialize_all_variables())
# restore trained model
ckpt = tf.train.get_checkpoint_state('train')
print(ckpt.model_checkpoint_path)
if ckpt and ckpt.model_checkpoint_path:
print("Load checkpint.")
saver.restore(sess, ckpt.model_checkpoint_path)
global_step = ckpt.model_checkpoint_path.split('/')[-1].split('-')[-1]
else:
print('No checkpoint file found.')
quit()