Thank you for open sourcing your baselines and the awesome dataset :) I have a question about the performance of CD-seq2seq. In the paper and the leaderboard, the result is as follow:
But in the README.md file, I find it higher(almost improve 5% on question match result in dev) than the reported result.
So I want to know which one is correct? Or, does the latest result in the README follow the setting in the paper? Thanks for your reply.
Thank you for open sourcing your baselines and the awesome dataset :) I have a question about the performance of
CD-seq2seq
. In the paper and the leaderboard, the result is as follow:But in the README.md file, I find it higher(almost improve 5% on question match result in dev) than the reported result.
So I want to know which one is correct? Or, does the latest result in the README follow the setting in the paper? Thanks for your reply.