Created
November 21, 2015 18:37
-
-
Save rhaps0dy/3717737ab28f2ef7c3c3 to your computer and use it in GitHub Desktop.
NaNoGenMo character-based RNN, based on Karpathy's blog and code.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
#!/usr/bin/env python2 | |
import tensorflow as tf | |
import numpy as np | |
import time | |
# Hyperparameters | |
learning_rate = 1e-1 | |
n_hidden = 100 | |
seq_length = 25 | |
n_show = 2500 // seq_length | |
f = open('corpus.txt', 'r') | |
corpus = f.read() | |
f.close() | |
charset = list(set(corpus)) | |
char_to_ix = { ch:i for i,ch in enumerate(charset) } | |
ix_to_char = { i:ch for i,ch in enumerate(charset) } | |
n_chars = len(charset) | |
print "Corpus has %d characters, %d unique" % (len(corpus), n_chars) | |
def weight_variable(shape): | |
initial = tf.truncated_normal(shape, stddev=0.1) | |
return tf.Variable(initial) | |
def bias_variable(shape): | |
initial = tf.constant(0.1, shape=shape) | |
return tf.Variable(initial) | |
Wxh = weight_variable([n_chars, n_hidden]) | |
Whh = weight_variable([n_hidden, n_hidden]) | |
bh = bias_variable([1, n_hidden]) | |
Why = weight_variable([n_hidden, n_chars]) | |
by = bias_variable([1, n_chars]) | |
x = list(tf.placeholder("float", shape=[1, n_chars]) for _ in xrange(seq_length)) | |
y_ = list(tf.placeholder("float", shape=[1, n_chars]) for _ in xrange(seq_length)) | |
h0 = tf.placeholder("float", shape=[1, n_hidden]) | |
h = h0 | |
prob = range(seq_length) | |
for i in xrange(seq_length): | |
h = tf.nn.tanh(tf.matmul(x[i], Wxh) + tf.matmul(h, Whh) + bh) | |
y = tf.matmul(h, Why) + by | |
prob[i] = tf.nn.softmax(y) | |
loss = -tf.reduce_sum(y_[0] * tf.log(prob[0])) | |
for i in xrange(1, seq_length): | |
loss += -tf.reduce_sum(y_[i] * tf.log(prob[i])) | |
train_step = tf.train.AdagradOptimizer(learning_rate).minimize(loss) | |
print time.clock() | |
prev_time = time.clock() | |
h_prev = [[0]*n_hidden] | |
with tf.Session() as sess: | |
sess.run(tf.initialize_all_variables()) | |
p = 0 | |
while p < len(corpus): | |
inputs = corpus[p:p+seq_length] | |
targets = corpus[p+1:p+seq_length+1] | |
inp = dict() | |
i = 0 | |
for c, t in zip(inputs, targets): | |
inp[x[i]] = [[(1 if j==char_to_ix[c] else 0) for j in xrange(n_chars)]] | |
inp[y_[i]] = [[(1 if j==char_to_ix[t] else 0) for j in xrange(n_chars)]] | |
i += 1 | |
inp[h0] = h_prev | |
if p // seq_length % n_show == 0: | |
print "Iteration %d:" % (p // seq_length) | |
print time.clock() - prev_time | |
prev_time = time.clock() | |
result = sess.run(prob, inp) | |
s = "" | |
for r in result: | |
c = np.random.choice(range(n_chars), p=r.ravel()) | |
s += ix_to_char[c] | |
print s | |
_, h_prev = sess.run([train_step, h], inp) | |
p += seq_length |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
#!/usr/bin/env python2 | |
""" | |
Minimal character-level Vanilla RNN model. Written by Andrej Karpathy (@karpathy) | |
BSD License | |
""" | |
import numpy as np | |
import time | |
# data I/O | |
data = open('corpus.txt', 'r').read() # should be simple plain text file | |
chars = list(set(data)) | |
data_size, vocab_size = len(data), len(chars) | |
print 'data has %d characters, %d unique.' % (data_size, vocab_size) | |
char_to_ix = { ch:i for i,ch in enumerate(chars) } | |
ix_to_char = { i:ch for i,ch in enumerate(chars) } | |
# hyperparameters | |
hidden_size = 100 # size of hidden layer of neurons | |
seq_length = 25 # number of steps to unroll the RNN for | |
learning_rate = 1e-1 | |
# model parameters | |
Wxh = np.random.randn(hidden_size, vocab_size)*0.01 # input to hidden | |
Whh = np.random.randn(hidden_size, hidden_size)*0.01 # hidden to hidden | |
Why = np.random.randn(vocab_size, hidden_size)*0.01 # hidden to output | |
bh = np.zeros((hidden_size, 1)) # hidden bias | |
by = np.zeros((vocab_size, 1)) # output bias | |
def lossFun(inputs, targets, hprev): | |
""" | |
inputs,targets are both list of integers. | |
hprev is Hx1 array of initial hidden state | |
returns the loss, gradients on model parameters, and last hidden state | |
""" | |
xs, hs, ys, ps = {}, {}, {}, {} | |
hs[-1] = np.copy(hprev) | |
loss = 0 | |
# forward pass | |
for t in xrange(len(inputs)): | |
xs[t] = np.zeros((vocab_size,1)) # encode in 1-of-k representation | |
xs[t][inputs[t]] = 1 | |
hs[t] = np.tanh(np.dot(Wxh, xs[t]) + np.dot(Whh, hs[t-1]) + bh) # hidden state | |
ys[t] = np.dot(Why, hs[t]) + by # unnormalized log probabilities for next chars | |
ps[t] = np.exp(ys[t]) / np.sum(np.exp(ys[t])) # probabilities for next chars | |
loss += -np.log(ps[t][targets[t],0]) # softmax (cross-entropy loss) | |
# backward pass: compute gradients going backwards | |
dWxh, dWhh, dWhy = np.zeros_like(Wxh), np.zeros_like(Whh), np.zeros_like(Why) | |
dbh, dby = np.zeros_like(bh), np.zeros_like(by) | |
dhnext = np.zeros_like(hs[0]) | |
for t in reversed(xrange(len(inputs))): | |
dy = np.copy(ps[t]) | |
dy[targets[t]] -= 1 # backprop into y | |
dWhy += np.dot(dy, hs[t].T) | |
dby += dy | |
dh = np.dot(Why.T, dy) + dhnext # backprop into h | |
dhraw = (1 - hs[t] * hs[t]) * dh # backprop through tanh nonlinearity | |
dbh += dhraw | |
dWxh += np.dot(dhraw, xs[t].T) | |
dWhh += np.dot(dhraw, hs[t-1].T) | |
dhnext = np.dot(Whh.T, dhraw) | |
for dparam in [dWxh, dWhh, dWhy, dbh, dby]: | |
np.clip(dparam, -5, 5, out=dparam) # clip to mitigate exploding gradients | |
return loss, dWxh, dWhh, dWhy, dbh, dby, hs[len(inputs)-1] | |
def sample(h, seed_ix, n): | |
""" | |
sample a sequence of integers from the model | |
h is memory state, seed_ix is seed letter for first time step | |
""" | |
x = np.zeros((vocab_size, 1)) | |
x[seed_ix] = 1 | |
ixes = [] | |
for t in xrange(n): | |
h = np.tanh(np.dot(Wxh, x) + np.dot(Whh, h) + bh) | |
y = np.dot(Why, h) + by | |
p = np.exp(y) / np.sum(np.exp(y)) | |
ix = np.random.choice(range(vocab_size), p=p.ravel()) | |
x = np.zeros((vocab_size, 1)) | |
x[ix] = 1 | |
ixes.append(ix) | |
return ixes | |
n, p = 0, 0 | |
mWxh, mWhh, mWhy = np.zeros_like(Wxh), np.zeros_like(Whh), np.zeros_like(Why) | |
mbh, mby = np.zeros_like(bh), np.zeros_like(by) # memory variables for Adagrad | |
smooth_loss = -np.log(1.0/vocab_size)*seq_length # loss at iteration 0 | |
prev_time = time.clock() | |
while True: | |
# prepare inputs (we're sweeping from left to right in steps seq_length long) | |
if p+seq_length+1 >= len(data) or n == 0: | |
hprev = np.zeros((hidden_size,1)) # reset RNN memory | |
p = 0 # go from start of data | |
inputs = [char_to_ix[ch] for ch in data[p:p+seq_length]] | |
targets = [char_to_ix[ch] for ch in data[p+1:p+seq_length+1]] | |
# sample from the model now and then | |
if n % 100 == 0: | |
print time.clock() - prev_time | |
prev_time = time.clock() | |
sample_ix = sample(hprev, inputs[0], 200) | |
txt = ''.join(ix_to_char[ix] for ix in sample_ix) | |
print '----\n %s \n----' % (txt, ) | |
# forward seq_length characters through the net and fetch gradient | |
loss, dWxh, dWhh, dWhy, dbh, dby, hprev = lossFun(inputs, targets, hprev) | |
smooth_loss = smooth_loss * 0.999 + loss * 0.001 | |
if n % 100 == 0: print 'iter %d, loss: %f' % (n, smooth_loss) # print progress | |
# perform parameter update with Adagrad | |
for param, dparam, mem in zip([Wxh, Whh, Why, bh, by], | |
[dWxh, dWhh, dWhy, dbh, dby], | |
[mWxh, mWhh, mWhy, mbh, mby]): | |
mem += dparam * dparam | |
param += -learning_rate * dparam / np.sqrt(mem + 1e-8) # adagrad update | |
p += seq_length # move data pointer | |
n += 1 # iteration counter |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment