[컴][머신러닝] Recurrent Neural Networks 예제

ML / machine learning


Recurrent Neural Networks, RNN


관련글 :  The Unreasonable Effectiveness of Recurrent Neural Networks


원본 소스 : https://gist.github.com/karpathy/d4dee566867f8291f086

python 3 버전

# encoding=utf8

"""
Minimal character-level Vanilla RNN model. Written by Andrej Karpathy (@karpathy)
BSD License
"""
import numpy as np
import codecs

# data I/O
with codecs.open('input.txt', 'r', encoding='utf-8') as fp:
    data = fp.read()
    # data = open('input.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 range(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(range(len(inputs))):
        dy = np.copy(ps[t])
        dy[targets[t]] -= 1 # backprop into y. see http://cs231n.github.io/neural-networks-case-study/#grad if confused here
        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 range(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
    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:
        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 
fdsfds


결과

무한 loop 이라, 적정한 시점에 알아서 멈추면 된다.

아래 결과는 대략 12시간 정도를 돌렸다. 컴퓨터 스펙은 아래와 같다.

  • CPU:  i7-6700, 3.40GHz
  • RAM: 16GB
----
 beahngy
amo k ns aeo?cdse nh a taei.rairrhelardr nela haeiahe.
Ddelnss.eelaishaner” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae abeahngy
amo k ns aeo?cdse nh a taei.rairrhelardr nol e iohahenasen
----
iter 9309400, loss: 0.000086
----
 e nh a taei.rairrhelardr naioa aneaa ayio pe e bhnte ayio pe e h’e btentmuhgehi bcgdltt. gey heho grpiahe.
Ddelnss.eelaishaner” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae abeahngy
amo k ns aeo?cds
----
iter 9309500, loss: 0.000086
----
 jCTCnhoofeoxelif edElobe negnk e iohehasenoldndAmdaI ayio pe e h’e btentmuhgehi bcgdltt. gey heho grpiahe.
Ddelnss.eelaishaner” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae abeahngy
amo k ns aeo?cds
----
iter 9309600, loss: 0.000086
----
 negnk e iohehasenoldndAmdaI ayio pe e h’e btentmuhgehi bcgdltt. gey heho grpiahe.
Ddelnss.eelaishaner” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae abeahngy
amo k ns aeo?cdse nh a taei.rairrhelardr
----
iter 9309700, loss: 0.000086
----
 aI ayio pe e h’e btentmuhgehi bcgdltt. gey heho grpiahe.
Ddelnss.eelaishaner” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae abeahngy
amo k ns aeo?cdse nh a taei.rairrhelardr neli ae e angnI hyho gben
----
iter 9309800, loss: 0.000086
----
 gehi bcgdltt. gey heho grpiahe.
Ddelnss.eelaishaner” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae abeahngy
amo k ns aeo?cdse nh a taei.rairrhelardr nela dr iohecgrpiahe.
Ddelnss.eelaishaner” cot AA
----
iter 9309900, loss: 0.000086
----
 piahe.
Ddelnss.eelaishaner” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae abeahngy
amo k ns aeo?cdse nh a taei.rairrhelardr nol e iohahenasenese hbea bhnte ectrsnae abeahngy
amo k ns aeo?cdse nh a t
----
iter 9310000, loss: 0.000086
----
 er” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae abeahngy
amo k ns aeo?cdse nh a taei.rairrhelardr nela hamnaI ayio pe e h’e btentmuhgnhi beahe
Ddabealohe bee amoi bcgdltt. gey heho grpiahe.
Ddeln
----
iter 9310100, loss: 0.000086
----
 bih a”on bhnte ectrsnae abeahngy
amo k ns aeo?cdse nh a taei.rairrhelardr nol gyio pe e h’e btentmuhgehi bcgdltt. gey heho grpiahe.
Ddelnss.eelaishaner” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae
----
iter 9310200, loss: 0.000086
----
 beahngy
amo k ns aeo?cdse nh a taei.rairrhelardr ntlhnegnns. e
amo k ns aeh?cdse nh a taei.rairrhelardr nol e iohehengrpiahe.
Ddelnss.eelaishaner” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae abeah
----
iter 9310300, loss: 0.000086
----
 e nh a taei.rairrhelardr nol’e btentmuhgehi gcdslatha arenbggcodaeta tehr he ni.rhelaney gehnha e ar i ho  bee amote ectrsnae abeahngy
amo k ns aeo?cdse nh a taei.rairrhelardr nol nyio chge heiohecgr
----
iter 9310400, loss: 0.000086
----
 jCTCnhoofeoxelif edElobe negnk e iohehasenoldndAmdaI ayio pe e h’e btentmuhgehi bcgdltt. gey heho grpiahe.
Ddelnss.eelaishaner” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae abeahngy
amo k ns aeo?cds
----
iter 9310500, loss: 0.000086
----
 negnk e iohehasenoldndAmdaI ayio pe e h’e btentmuhgehi bcgdltt. gey heho grpiahe.
Ddelnss.eelaishaner” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae abeahngy
amo k ns aeo?cdse nh a taei.rairrhelardr
----
iter 9310600, loss: 0.000086
----
 aI ayio pe e h’e btentmuhgehi bcgdltt. gey heho grpiahe.
Ddelnss.eelaishaner” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae abeahngy
amo k ns aeo?cdse nh a taei.rairrhelardr nelardae abeahngy
amo k
----
iter 9310700, loss: 0.000086
----
 gehi bcgdltt. gey heho grpiahe.
Ddelnss.eelaishaner” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae abeahngy
amo k ns aeo?cdse nh a taei.rairrhelardr ntl negnk t hi rsnse nhk br ne” a naeiarairr elirs
----
iter 9310800, loss: 0.000086
----
 piahe.
Ddelnss.eelaishaner” cot AAfhB ht ltny
ehbih a”on bhnte ectrsnae abeahngy
amo k ns aeo?cdse nh a taei.rairrhelardr nelardaenabeahngelareierhi. aif edElobe negrcih gey gey heho grpiahe.
Ddel
----



댓글 없음:

댓글 쓰기