Have you used multiple rounds of conversation to test the model?
#7
by
Ranck626
- opened
Thank you for your great open source.
I noticed that in the paper Let's go real talk, you seem to have used multi-round conversations to test and train the model.
However, in the experimental comparison part, speechGPT is involved, and the model does not seem to demonstrate the ability of multi-round conversations from the demo on the official website. How did you test it?
Thanks for your attention！