Dear LM-KBC Organizers,
I am planning to submit a short paper to the LM-KBC Shared Task and have a few technical and logistical questions:
1. Submission Process & Deadlines:
Could you confirm that the system description short paper is a Direct Submission due on August 15, and does not require any prior ARR review process?
2. ACL Reviewer Mandate:
If we are submitting exclusively to the workshop via Direct Submission, does the general EMNLP/ACL mandate requiring all authors to register as reviewers still apply to our team?
3. Few-Shot Prompting:
Given that RAG is strictly prohibited, is it permissible to use few-shot prompting by injecting examples directly from the provided train.jsonl dataset into the prompt?
4. Handling Unknown Relations:
If the model determines that it does not know a specific relation, what is the expected output format? Should we output the row with an empty prediction, or can we omit the row entirely from the test submission? The evaluation script seems to handle missing
rows, but I want to ensure strict compliance with the test phase requirements.
Thank you for your time and clarifications.
Best regards,
Mikołaj Tym
|
|
mgr Mikołaj Tym
Asystent, Katedra Informatyki Ekonomicznej
Uniwersytet Ekonomiczny w Poznaniu
|
|
Teaching Assistant, Department of Information Systems
Poznań University of Economics and Business
|
|
|
Nadawcą korespondencji i jednocześnie administratorem Państwa danych osobowych jest Uniwersytet Ekonomiczny w Poznaniu, al. Niepodległości 10, 61-875 Poznań, z którym można skontaktować się listownie lub poprzez wiadomość e-mail -
ro...@ue.poznan.pl Państwa
dane osobowe przetwarzane będą w celu prowadzenia niniejszej korespondencji i przez czas jej trwania. Przysługuje Państwu prawo dostępu do danych, ich sprostowania, ograniczenia przetwarzania, usunięcia, żądania sprzeciwu wobec przetwarzania oraz wniesienia
skargi do Prezesa Urzędu Ochrony Danych Osobowych. Szczegółowe informacje na temat ochrony danych osobowych dostępne są w Polityce prywatności:
https://ue.poznan.pl/uniwersytet/polityka-prywatnosci/