Text Generation

Text generation from a Recurrent Neural Network trained on English translated version of "The Ramayana of Valmiki, translated by Hari Prasad Shastri"

---

Epoch 5/20

2532/2532 [==============================] - 331s 131ms/step - loss: 1.1127

####################

Temperature: 0.2

####################

on of the sages, the titans and the powerful Ravana and the son of Dasaratha, who was able to disport the son of Dasaratha and the son of Dasaratha and the monkeys and the son of Dasaratha and the foremost of the Rakshasas and the son of Ravana's heart and the sun and the son of the Wind-god

of Dasaratha and the son of Vayu, the son of Dasaratha, who was subject to the sun and the son of Dasaratha and Shri Ramachandra and the son of the Gods, the King of the Monkeys, who was able to change the sun and the sun in the presence of the forest and the head of the powerful Ravana and

and the son of Dasaratha and the sun and the son of Dasaratha and the son of Vasava, the King of the Monkeys, having set out from the sun and arrows and the son of the Wind-god and the titans with the army of the sun and the son of Dasaratha, the son of Dasaratha and having slain the monkeys

---

Epoch 20/20

2532/2532 [==============================] - 332s 131ms/step - loss: 0.9149

####################

Temperature: 1.0

####################

able to deep like a lion, made calamity bore the terror of his anger. Having accepted these words of Raghava, many deceiting flowering battle. On none hanging his instruction and great intellect with thy grief! It was formerly. Now Bharata and cross their lives four departure. The prowess of roya

icate tones, Manu whose boon must take heart, this is help to the titans in seas for shouting, lay possessed of himself with con-litulat ever bestowed on him; therefore, provoked youth and in words with joined palms, said :- cc This matter none were mine own abode. None were mastered with their

s, ascended a celestial partike, cut forth, whether he bears the scent of this toward and again, the aged titan among the sacrificial fires of my opposi- ity, would not listen, between the Goddess Poulastya and their leaders, in a brahmin of which I will destroy the kine and the third was fil

---

Model, library and parameters used:

textgen = textgenrnn(name='Ramayan RNN')

train_function = textgen.train_from_largetext_file

train_function(file_path='ram.txt',
              new_model=True,
              num_epochs=train_cfg['num_epochs'],
              gen_epochs=train_cfg['gen_epochs'],
              batch_size=1024,
              train_size=train_cfg['train_size'],
              validation=train_cfg['validation'],
              rnn_layers=model_cfg['rnn_layers'],
              rnn_size=model_cfg['rnn_size'],
              rnn_bidirectional=model_cfg['rnn_bidirectional'],
              max_length=model_cfg['max_length'],
              dim_embeddings=100)

model_cfg = {'rnn_size': 128,
            'rnn_layers': 3,
            'rnn_bidirectional': True,
            'max_length': 30}

train_cfg = {'num_epochs': 20,
            'gen_epochs': 5,
            'train_size': 0.9,
            'validation': True}