Melatih model dengan Teacher Forcing

Penerjemahan Mesin dengan Keras

Thushan Ganegedara

Data Scientist and Author

Detail pelatihan model

Pelatihan model memerlukan:
- Fungsi loss (mis. categorical crossentropy)
- Optimizer (mis. Adam)

Detail pelatihan model

Untuk menghitung loss, diperlukan:
- Prediksi probabilistik dari input ([batch_size, seq_len, vocab_size])
  - mis. [[0.11,...,0.81,0.04], [0.05,...,0.01, 0.93], ..., [0.78,..., 0.03,0.01]]
- Target bahasa Prancis one-hot aktual ([batch_size, seq_len, vocab_size])
  - mis. [[0, ..., 1, 0], [0, ..., 0, 1],..., [0, ..., 1, 0]]
- Crossentropy: selisih antara target dan kata terprediksi
Loss diteruskan ke optimizer yang mengubah parameter model untuk meminimalkan loss

Melatih model dengan Teacher Forcing

n_epochs, bsize = 3, 250
for ei in range(n_epochs):
  for i in range(0,data_size,bsize):

    # Masukan encoder, masukan dan keluaran decoder
    en_x = sents2seqs('source', en_text[i:i+bsize], onehot=True, reverse=True)
    de_xy = sents2seqs('target', fr_text[i:i+bsize], onehot=True)

    # Memisahkan masukan dan keluaran decoder
    de_x = de_xy[:,:-1,:]    
    de_y = de_xy[:,1:,:]

    # Melatih dan mengevaluasi pada satu batch
    nmt_tf.train_on_batch([en_x,de_x], de_y)    
    res = nmt_tf.evaluate([en_x,de_x], de_y, batch_size=bsize, verbose=0)
    print("{} => Train Loss:{}, Train Acc: {}".format(ei+1,res[0], res[1]*100.0))

Pemotongan array secara detail

    de_x = de_xy[:,:-1,:]    
    de_y = de_xy[:,1:,:]

Pemotongan array

Membuat data pelatihan dan validasi

train_size, valid_size = 800, 200
# Membuat indeks data
inds = np.arange(len(en_text))
np.random.shuffle(inds)


# Memisahkan indeks train dan valid
train_inds = inds[:train_size]
valid_inds = inds[train_size:train_size+valid_size]


# Mengambil data train dan valid
tr_en = [en_text[ti] for ti in train_inds]
tr_fr = [fr_text[ti] for ti in train_inds]

v_en = [en_text[vi] for vi in valid_inds]
v_fr = [fr_text[vi] for vi in valid_inds]

print('Training (EN):\n', tr_en[:2], '\nTraining (FR):\n', tr_fr[:2])
print('\nValid (EN):\n', tr_en[:2], '\nValid (FR):\n', tr_fr[:2])

Valid (EN): ['she saw that rusty blue truck .', 'france is usually snowy during autumn , ...'] 
Valid (FR): ['elle a vu ce camion bleu rouillé .', 'la france est généralement enneigée en automne ...']

Pelatihan dengan validasi

for ei in range(n_epochs):
  for i in range(0,train_size,bsize):    
    en_x = sents2seqs('source', tr_en[i:i+bsize], onehot=True, reverse=True)
    de_xy = sents2seqs('target', tr_fr[i:i+bsize], onehot=True)
    de_x, de_y = de_xy[:,:-1,:], de_xy[:,1:,:]
    nmt_tf.train_on_batch([en_x, de_x], de_y)

  v_en_x = sents2seqs('source', v_en, onehot=True, reverse=True)
  v_de_xy = sents2seqs('target', v_fr, onehot=True)
  v_de_x, v_de_y = v_de_xy[:,:-1,:], v_de_xy[:,1:,:]

  res = nmt_tf.evaluate([v_en_x, v_de_x], v_de_y, batch_size=valid_size, verbose=0)
  print("Epoch {} => Loss:{}, Val Acc: {}".format(ei+1,res[0], res[1]*100.0))

Epoch 1 => Loss:4.784221172332764, Val Acc: 1.4999999664723873
Epoch 2 => Loss:4.716882228851318, Val Acc: 44.458332657814026
Epoch 3 => Loss:4.63267183303833, Val Acc: 47.333332896232605

Ayo berlatih!

Penerjemahan Mesin dengan Keras