How are People getting near Perfect Accuracy? My thought is they are not using either phi-2 or falcon. So if they are using other powerful models, Will they be disqualified because the rules were we either use phi-2 or falcon?
Hi Antonio, do all the participants have to submit the paper write-ups? Or the result from the private leaderboard will be published and only the top 3/5/10 teams need to submit the report in written form?
The final score is based on the private leaderboard (Not the public one) code quality and submitted paper.
But still it we be unfair for people to use more powerful models than what was specified right?
I do not know what people is doing but we will check. Please focus on your contribution :)
Hi Antonio, do all the participants have to submit the paper write-ups? Or the result from the private leaderboard will be published and only the top 3/5/10 teams need to submit the report in written form?
Anyone thinking that he/she has done a good work can continue the competition. Follow the instruction discussed here https://zindi.africa/competitions/specializing-large-language-models-for-telecom-networks/discussions/21743
How can someone get private leaderboard score 0 if he is having public leaderboard score 0.45? Can someone please explain?
the participant has likely not replied to the question from the private leaderboard