RL_reasoning

Authors:Haotian Cui, Yukun Yan, Xianggen Liu

data preparing

• preprocess.py: replace the end char "." and add "|||"
• stream.py: 1.add "" to the phrase 2.build the process class, then map to the index, make up the vector representation
  finally,return a list of vectors. namely, for facts, question, label in data_class.data_stream():#facts,question:length->2,label -> 1,such as: facts[0]:(shape:(5,10)) [[ 9 21 0 2 14 20 21 16 11 5] [ 9 21 19 2 17 20 21 0 11 5] [ 9 21 18 2 10 20 21 3 11 5] [ 9 21 0 2 10 20 21 18 11 5] [ 9 21 6 2 17 20 21 18 11 5]] question[0]:shape:(1,13) print("-------------------")

fda