Inconsistency in evaluator_relationship

[borgr]

People report first and third party for things that look similar. The distinction is not clear enough.
What is an evaluator relationship, and with what. Do we mean relationship between the one running the evals and the one sharing the data? The one uploading to EEE (I think not)? or the one training the model? (If so, maybe better call it model-side, evaluator, third party or something? Because data creator is a different first party and might also creat self-disambiguation).
Need to do any of rename, better document, add another label that means the other one, rename fields, or otherwise simplify to make intuitive what we mean there.
"clean code" self documents.

See more here
https://evalevalcoalition.slack.com/archives/C09D9RTJABS/p1777062652058319?thread_ts=1777021800.437299&cid=C09D9RTJABS

Example:
Apex agents is reporting as first party: https://huggingface.co/datasets/evaleval/EEE_datastore/blob/main/data/apex-agents/anthropic/opus-4.6/ddb6b96d-345f-4731-b62b-29e75c91f8a7.json

Fibble arena is reporting as third party: https://huggingface.co/datasets/evaleval/EEE_datastore/blob/main/data/fibble_arena/a[…]pic/claude-opus-4.6/bcc02a26-e09d-4242-b012-5fd1b14c045a.json

[snakefood3232]

I see the confusion about "evaluator relationship" relating to data-sharing and training roles. I'd handle this by renaming fields for clarity and adding detailed documentation to distinguish between them properly. Can have a PR up in 2 days. I've built similar pipelines for clear categorization in data evaluation contexts before.

[borgr]

Sounds good, can you tag me when you do it? Also, note that schema changes (e.g. names) are still to be done in the dev as long as the shared task freeze is on.