- Paper: Adaptive Natural Language Generation for Task-oriented Dialogue via Reinforcement Learning
- GitHub: nu-dialogue/antor
You can load the GPT2 + RL's weight, which is officially trained and published, by the following code.
from gpt2rl import GPT2RLNLG
nlg = GPT2RLNLG(model_name_or_path='ohashi56225/antor-full_bert_nlu-no_noise', device="cuda:0")