This is the code for the paper Personalized Dialogue Response Generation from Monologues by Feng-Guang Su*, Aliyah R. Hsu*, Yi-Lin Tuan and Hung-Yi Lee (* indicates equal contribution). See the project page for details and generated sentence examples.
Please configure train.sh first with your own settings. (e.g. model path and paprameters, data path, feature path and glove model path) Once train.sh is complete, run the following:
bash train.sh
For single sentence testing (1 prompt -> 1 response from your desired persona), please run:
bash test.sh
For multiple sentence testing (1 prompt -> multiple responses from different personas), please run:
bash seriestest.sh
For automatic testing by generating sentence files, please run:
bash filetest.sh
If you are using your own dataset, please refer to our provided data directory for the correct format of conversations, monologues and features.