majumderb / pabst

Code for "Unsupervised Enrichment of Persona-grounded Dialog with Background Stories", ACL 2021

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

train

xiao-ming-code opened this issue · comments

Hello, could you please tell me the steps and commands of training and prediction

Hi,

Our method doesn't require training a dialog model. You should use any transformer-based generative model as your dialog model.

To train the dialog NLI classifier, see here: https://github.com/majumderb/pabst/blob/main/pabst/run_pplm_discrim_train.py
--help from the command line should give you all the arguments you can pass.

Once you have both the model (dialog model of your choice, and the dnli model), you can run inference using this code: https://github.com/majumderb/pabst/blob/main/pabst/run_pplm.py
Similarly, --help from the command line should give you all the arguments you can pass.