Problem of the scale of the reward

lancopku / DPGAN

Diversity-Promoting Generative Adversarial Network for Generating Informative and Diversified Text (EMNLP2018)

146 stars 38 forks source link

Problem of the scale of the reward #6

Open RedBeanCake opened 6 years ago

RedBeanCake commented 6 years ago

Great thanks for sharing your code! It is not clear for me why do you scale the reward in the following way: if reward['y_pred_auc'][i][j][k] > 12: reward['y_pred_auc'][i][j][k] = 12/ 10000.0 else: reward['y_pred_auc'][i][j][k] = reward['y_pred_auc'][i][j][k] / 10000.0

Could you please help?