In the paper, the Twitter 2017 datasets contain 3373/723/723 sentences for train/dev/test respectively. However, in the paper of Lu et al., 2018, they reported that there are 4290/1432/1459 sentences in the Twitter dataset. Could you tell me what is the difference between the two datasets?
Hi,
In the paper, the Twitter 2017 datasets contain 3373/723/723 sentences for train/dev/test respectively. However, in the paper of Lu et al., 2018, they reported that there are 4290/1432/1459 sentences in the Twitter dataset. Could you tell me what is the difference between the two datasets?