I am trying to run the cross-sentence distance experiment.
I created the dataset by setting CSD to 3 in the data processing notebook. I ran the models and checked for the results. I think something is wrong because the results go from 90 to 9 for CSD 1, which seems extremely unlikely.
As I checked the eval script, I saw the relations are fetched from all of the gold standard test files and scores are calculated. So the CSD 1 experiment's result is compared with all of the gold standard datasets, I think.
I am a little confused about this experiment. Could you please help me if I am missing anything or what I am doing wrong?
Hi Team, @bugface @liangchh
I am trying to run the cross-sentence distance experiment.
I created the dataset by setting CSD to 3 in the data processing notebook. I ran the models and checked for the results. I think something is wrong because the results go from 90 to 9 for CSD 1, which seems extremely unlikely. As I checked the eval script, I saw the relations are fetched from all of the gold standard test files and scores are calculated. So the CSD 1 experiment's result is compared with all of the gold standard datasets, I think.
I am a little confused about this experiment. Could you please help me if I am missing anything or what I am doing wrong?
Any help would be greatly appreciated.
Thanks