Add QMSD-Encoder-v1.1 submission

[d0ng231]

Submitting QMSD-Encoder-v1.1 to the RAID leaderboard.

Thanks to the maintainers for the quick response on the previous PR.

This is an interim version; we'll release more complete results + the model alongside an upcoming paper, and decide then which version to merge into the main leaderboard.

[liamdugan]

Sounds good @d0ng231! Just as a quick note, you can update or add new methods to an existing PR by including them in a new folder. It's also fine to do as separate PRs -- up to you.

Thanks for submitting to RAID!

[github-actions]

Eval run succeeded! Link to run: link

Here are the results of the submission(s):

QMSD-Encoder-v1.1

Release date: 2026-04-26

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 99.68 and a TPR of 99.52% at FPR=5% and 97.82% at FPR=1%.
Without adversarial attacks, it achieved AUROC of 99.76 and a TPR of 99.91% at FPR=5% and 99.15% 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!

[github-actions]

Eval run succeeded! Link to run: link

Here are the results of the submission(s):

QMSD-Encoder-v1.2

Release date: 2026-04-27

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 99.90 and a TPR of 99.74% at FPR=5% and 98.53% at FPR=1%.
Without adversarial attacks, it achieved AUROC of 99.97 and a TPR of 99.89% at FPR=5% and 99.53% at FPR=1%.

QMSD-Encoder-v1.2-mini

Release date: 2026-04-27

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 99.30 and a TPR of 99.46% at FPR=5% and 86.78% at FPR=1%.
Without adversarial attacks, it achieved AUROC of 99.35 and a TPR of 99.75% at FPR=5% and 88.08% 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!