Julien Plu
Julien Plu
Thanks a lot for this clear answer! I fully agree that having the dataset name is meaningless and brings too specific pipelines. I just proposed this as a simple workaround...
Thanks @MichaelRoeder for the nice explanation of the situation. From my side I think that the most tough part will be mostly anybody can define its own KB and set...
Thanks @MichaelRoeder I will check with a single document with more specific details on how to reproduce this with Gerbil and the two other scorers ASAP and share it on...
The results with the conlleval scorer was a happy coincidence, because it does not evaluate by "offset" but by "token", so the way it evaluates the recognition is different. Sorry...
No it should correspond to the first line where ```0,4375 | 0,4375 | 0,4375``` is written. Entity Typing is something else. Basically "strong_typed_mention_match" in neleval == "RT2KB" in GERBIL and...
Thanks @MichaelRoeder! Let me know once the bug will be fixed.
Thanks @MichaelRoeder and @TortugaAttack. I can perfectly understand your concerns about the scoring I raised but it is more to be aligned with the known and popular neleval scorer. Personally...
Hello! Is there any workaround for this issue or we have to wait for a bugfix? Thanks.
Thanks @anuzzolese for the merge! @chabmed let us know if you have any other issue.
Cool! thanks @chabmed @anuzzolese I will make another PR today with those fixes.