hongshaorou/Tianshou
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

modified the network

Browse Source
This commit is contained in:
Tongzheng Ren 2017-11-05 15:30:35 +08:00
parent 5f923f565e
commit 493d361022
1 changed files with 51 additions and 39 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

90
AlphaGo/Network.py
View File

@ -4,13 +4,6 @@ import time
import multi_gpu import multi_gpu
import tensorflow.contrib.layers as layers import tensorflow.contrib.layers as layers
data = np.load("data.npz")
boards = data["boards"]
wins = data["wins"]
ps = data["ps"]
print (boards.shape)
print (wins.shape)
print (ps.shape)
def residual_block(input, is_training): def residual_block(input, is_training):
normalizer_params = {'is_training': is_training, normalizer_params = {'is_training': is_training,
'updates_collections': None} 'updates_collections': None}
@ -44,7 +37,6 @@ def value_heads(input, is_training):
h = layers.fully_connected(h, 1, activation_fn=tf.nn.tanh, weights_regularizer=layers.l2_regularizer(1e-4)) h = layers.fully_connected(h, 1, activation_fn=tf.nn.tanh, weights_regularizer=layers.l2_regularizer(1e-4))
return h return h
x = tf.placeholder(tf.float32,shape=[None,19,19,17]) x = tf.placeholder(tf.float32,shape=[None,19,19,17])
is_training = tf.placeholder(tf.bool, shape=[]) is_training = tf.placeholder(tf.bool, shape=[])
z = tf.placeholder(tf.float32, shape=[None, 1]) z = tf.placeholder(tf.float32, shape=[None, 1])
@ -62,34 +54,54 @@ train_op = tf.train.RMSPropOptimizer(1e-2).minimize(total_loss)
var_list = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES) var_list = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES)
saver = tf.train.Saver(max_to_keep=10, var_list=var_list) saver = tf.train.Saver(max_to_keep=10, var_list=var_list)
epochs = 100 def train():
batch_size = 32 data = np.load("data.npz")
batch_num = boards.shape[0] // batch_size boards = data["boards"]
result_path = "./results/" wins = data["wins"]
with multi_gpu.create_session() as sess: ps = data["ps"]
sess.run(tf.global_variables_initializer()) print (boards.shape)
ckpt_file = tf.train.latest_checkpoint(result_path) print (wins.shape)
if ckpt_file is not None: print (ps.shape)
print('Restoring model from {}...'.format(ckpt_file)) epochs = 100
saver.restore(sess, ckpt_file) batch_size = 32
for epoch in range(epochs): batch_num = boards.shape[0] // batch_size
time_train = -time.time() result_path = "./results/"
index = np.arange(boards.shape[0]) with multi_gpu.create_session() as sess:
np.random.shuffle(index) sess.run(tf.global_variables_initializer())
losses = [] ckpt_file = tf.train.latest_checkpoint(result_path)
regs = [] if ckpt_file is not None:
for iter in range(batch_num): print('Restoring model from {}...'.format(ckpt_file))
_, l, r, value, prob = sess.run([train_op, loss, reg, v, p], feed_dict={x:boards[index[iter*batch_size:(iter+1)*batch_size]], saver.restore(sess, ckpt_file)
z:wins[index[iter*batch_size:(iter+1)*batch_size]], for epoch in range(epochs):
pi:ps[index[iter*batch_size:(iter+1)*batch_size]], time_train = -time.time()
is_training:True}) index = np.arange(boards.shape[0])
losses.append(l) np.random.shuffle(index)
regs.append(r) losses = []
if iter % 1 == 0: regs = []
print("Epoch: {}, Iteration: {}, Time: {}, Loss: {}, Reg: {}".format(epoch, iter, time.time()+time_train, np.mean(np.array(losses)), np.mean(np.array(regs)))) for iter in range(batch_num):
time_train=-time.time() _, l, r, value, prob = sess.run([train_op, loss, reg, v, p], feed_dict={x:boards[index[iter*batch_size:(iter+1)*batch_size]],
losses = [] z:wins[index[iter*batch_size:(iter+1)*batch_size]],
regs = [] pi:ps[index[iter*batch_size:(iter+1)*batch_size]],
if iter % 20 == 0: is_training:True})
save_path = "Epoch{}.Iteration{}.ckpt".format(epoch, iter) losses.append(l)
saver.save(sess, result_path + save_path) regs.append(r)
if iter % 1 == 0:
print("Epoch: {}, Iteration: {}, Time: {}, Loss: {}, Reg: {}".format(epoch, iter, time.time()+time_train, np.mean(np.array(losses)), np.mean(np.array(regs))))
time_train=-time.time()
losses = []
regs = []
if iter % 20 == 0:
save_path = "Epoch{}.Iteration{}.ckpt".format(epoch, iter)
saver.save(sess, result_path + save_path)
def forward(board):
result_path = "./results/"
with multi_gpu.create_session() as sess:
sess.run(tf.global_variables_initializer())
ckpt_file = tf.train.latest_checkpoint(result_path)
if ckpt_file is not None:
print('Restoring model from {}...'.format(ckpt_file))
saver.restore(sess, ckpt_file)
else:
raise ValueError("No model loaded")
return sess.run([p,v], feed_dict={x:board})