Datasets:

---

aggr8

/

Percept-V

Like 0

Fixes the ground truth for the colours_present task.

Problem. The generator built each 20-entry yes/no vector by iterating a Python set, so data.json listed the answers in the set's hash order (blue, fuchsia, green, ivory, …) instead of the order given in rules.txt (black, gray, brown, maroon, …). 183 of 200 records were wrong, and every model scored 0–7.5% regardless of its perception.

Changes

• colours_present/data.json: labels reordered to match rules.txt. Images are unchanged. Each corrected label was checked against the colours actually present in its image (200/200).
• colours_present/eval.py: gold labels are now read from data.json (matched on id) instead of from the answer file. Existing answer files, which carry the old labels, are therefore scored correctly without re-running inference, and the script prints a note when it sees them. A new optional --gold flag points at another data.json; output format is unchanged. Verified on 58 existing answer files: stale and corrected answer files give identical scores.
• Removed the 200 unused colours_present/data/first{N}.png. The task is single-image and the generator never produced these files.
• README.md: colours_present is listed as single-image, the loader snippet no longer pairs it with a first image, water_image is added to the two-image list, and a changelog entry is added.