Hi again,
I am curious about what methods the paper authors used for context with DialogBERT development? Did you use context prepending of input tokens for that? And how many conversational turns for context were used to obtain the DialogBERT research paper results?
Thanks in advance