lytzv / cs285proj

CS285Proj

Policy Gradient

• To run policy gradient: python3 rl(rnnorbert)+rnn.py
• The main algorithm is in pg.py

Q Learning

• To run Q learning: python3 run_q.py
• The main algorithm is in q_seq2seq.py

Imitation Learning (i.e No teacher forcing) or Supervised Learning (i.e. Teacher forcing)

• Run (rnnorbert)+rnn.py

Any file with (rnnorbert) in it allows you to choose either a Bert or an RNN-based Encoder.

Data

• To parse sgml file csv, refer to sgml2csv.py