Add MultiScale-Hybrid-mpnet submission

[Vigora]

This PR adds the MultiScale-Hybrid-mpnet detector submission to the RAID leaderboard.

The detector was trained on the RAID training split and should be categorized in the "trained on RAID" section.

Submission files:

• metadata.json
• predictions.json

[Vigora]

Hi, just a gentle follow-up on this leaderboard submission. The evaluation workflow is still awaiting maintainer approval. Please let me know if anything is needed or should be changed on my side. Thank you!

[github-actions]

It looks like this eval run failed. Please check the workflow logs to see what went wrong, then push a new commit to your PR to rerun the eval.

[liamdugan]

Hi @Vigora it looks like there was an error evaluating your submission. Here's the full traceback

Traceback (most recent call last):
  File "/home/runner/work/raid/raid/leaderboard/hydrate.py", line 74, in hydrate_all
    result_fp = eval_submission(submission)
  File "/home/runner/work/raid/raid/leaderboard/hydrate.py", line 163, in eval_submission
    eval_results = run_evaluation(d, df, target_fpr=(0.05, 0.01))
  File "/home/runner/work/raid/raid/raid/evaluate.py", line 233, in run_evaluation
    df = load_detection_result(df, results)
  File "/home/runner/work/raid/raid/raid/evaluate.py", line 11, in load_detection_result
    scores_df = pd.DataFrame.from_records(results)
  File "/opt/hostedtoolcache/Python/3.10.20/x64/lib/python3.10/site-packages/pandas/core/frame.py", line 2542, in from_records
    mgr = arrays_to_mgr(arrays, columns, result_index, typ=manager)
  File "/opt/hostedtoolcache/Python/3.10.20/x64/lib/python3.10/site-packages/pandas/core/internals/construction.py", line 114, in arrays_to_mgr
    index = _extract_index(arrays)
  File "/opt/hostedtoolcache/Python/3.10.20/x64/lib/python3.10/site-packages/pandas/core/internals/construction.py", line 667, in _extract_index
    raise ValueError("If using all scalar values, you must pass an index")
ValueError: If using all scalar values, you must pass an index
Error: If using all scalar values, you must pass an index
Done. Wrote 0 results files.
Error: Process completed with exit code 1.

Can you make sure that you have a single floating point value in score for every single id present in the test data? Please check the other submissions for examples of the formatting needed for predictions.json

[Vigora]

Hi Liam, thank you. I fixed predictions.json to the expected list-of-records format with one {"id": ..., "score": ...} record per prediction and pushed the correction to this PR. The new evaluation workflow is now awaiting approval.

[github-actions]

Eval run succeeded! Link to run: link

Here are the results of the submission(s):

MultiScale-Hybrid-mpnet

Release date: 2026-08-07

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 90.66 and a TPR of 77.48% at FPR=5% and 59.90% at FPR=1%.
Without adversarial attacks, it achieved AUROC of 95.26 and a TPR of 88.46% at FPR=5% and 78.16% 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!