49-uniHD

[PhilipEHausner]

Results for task 1b are available only for runs intersection_(2/3)_field, since we did not include predictions for the other runs in the first place (whoopsie), however, those were also the best of our runs for task 1a.

To fix the issues mentioned by @dennlinger, we followed advice from PR #20 and adopted their folder structure.

[PhilipEHausner]

@cmkumar87 I was not able to run the program without these changes. I just tried to replicate the results from #24 using our changes (it did not work without them here as well) and got the same results for the average precision, recall, and F1 score for task 1a and b (checking run 1 and 2 of their setup), but not for the Rouge scores, since they were not available in their scores file. There is also something about that in their error.log, identical to part of our error.log if we run the script without modifications, so it seems that this part of the evaluation script did not work for them either?

[cmkumar87]

@PhilipEHausner Your PR is not on your evaluation folder. Instead it is made to 2018-evaluation-script. I am not able to merge the PR or able to lookup your scores files for all the runs. Please advise and open a new PR