Submission/seed456

[MohamedMady19]

No description provided.

[liamdugan]

Hi @MohamedMady19 can you please refrain from submitting the same submission with different random seeds.

If you would like to conduct this type of stability analysis, please do so on a held out split of the RAID training set with cross validation rather than submitting to the official leaderboard. The leaderboard should be used only as a final submission.

I'm going to run this submission but from now on please only submit your final version.

[github-actions]

Eval run succeeded! Link to run: link

Here are the results of the submission(s):

DeBERTa-ConPara-v2.2-Seed42

Release date: 2026-05-01

I've committed detailed results of this detector's performance on the test set to this PR.

On the RAID dataset as a whole (aggregated across all generation models, domains, decoding strategies, repetition penalties, and adversarial attacks), it achieved an AUROC of 96.86 and a TPR of 97.29% at FPR=5% and 93.55% at FPR=1%.
Without adversarial attacks, it achieved AUROC of 97.05 and a TPR of 97.02% at FPR=5% and 93.39% at FPR=1%.

DeBERTa-ConPara-v2.2-Seed456

Release date: 2026-05-03

I've committed detailed results of this detector's performance on the test set to this PR.

On the RAID dataset as a whole (aggregated across all generation models, domains, decoding strategies, repetition penalties, and adversarial attacks), it achieved an AUROC of 97.13 and a TPR of 97.29% at FPR=5% and 91.36% at FPR=1%.
Without adversarial attacks, it achieved AUROC of 97.24 and a TPR of 97.08% at FPR=5% and 92.04% at FPR=1%.

If all looks well, a maintainer will come by soon to merge this PR and your entry/entries will appear on the leaderboard. If you need to make any changes, feel free to push new commits to this PR. Thanks for submitting to RAID!

[MohamedMady19]

Hi @MohamedMady19 can you please refrain from submitting the same submission with different random seeds.

If you would like to conduct this type of stability analysis, please do so on a held out split of the RAID training set with cross validation rather than submitting to the official leaderboard. The leaderboard should be used only as a final submission.

I'm going to run this submission but from now on please only submit your final version.

Hi @liamdugan Thank you so much, I will consider this.