references
Reference sets
What arms are scored against: labels promoted on purpose, at the tier chosen then. Gold is the yardstick a decision rests on; silver is consistent, good for ranking arms against each other, not a claim that any arm is right.
all time, 2026-09-25 to 2026-11-02
01reference sets
20 shown
| reference set | tier | dataset | items | promoted from | created | arms scored | arms scored in period | runs in period |
|---|---|---|---|---|---|---|---|---|
| gold/ag_news_20 | gold | ag_news_20 | 20 | person the dataset's authors | 2026-09-27 | 0 | 0 | 0 |
| gold/ag_news_200 | gold | ag_news_200 | 200 | person the dataset's authors | 2026-09-27 | 33 | 33 | 33 |
| gold/ag_news_200_adj1 | gold | ag_news_200 | 200 | person adjudication of the dataset's labels | 2026-10-19 | 30 | 0 | 0 |
| gold/cnn_dailymail_20 | gold | cnn_dailymail_20 | 20 | person the dataset's authors | 2026-09-25 | 2 | 2 | 2 |
| gold/cnn_dailymail_200 | gold | cnn_dailymail_200 | 200 | person the dataset's authors | 2026-09-25 | 35 | 35 | 35 |
| silver/cnn_dailymail_200 | silver | cnn_dailymail_200 | 200 | model the arm: cnn_anthropic_l_n200_v1 | 2026-10-25 | 29 | 0 | 0 |
| gold/cnn_dailymail_dm200 | gold | cnn_dailymail_dm200 | 200 | person the dataset's authors | 2026-10-14 | 5 | 5 | 5 |
| gold/dbpedia_280 | gold | dbpedia_280 | 280 | person the dataset's authors | 2026-09-27 | 29 | 29 | 29 |
| gold/dbpedia_56 | gold | dbpedia_56 | 56 | person the dataset's authors | 2026-09-27 | 0 | 0 | 0 |
| gold/few_nerd_280 | gold | few_nerd_280 | 280 | person the dataset's authors | 2026-09-28 | 28 | 28 | 28 |
| gold/few_nerd_56 | gold | few_nerd_56 | 56 | person the dataset's authors | 2026-09-28 | 0 | 0 | 0 |
| silver/fleet_dedup_v1 | silver | fleet_dedup_v1 | 100 | model silver_fleet_dedup_v1 the arm: dedup_pinned_f07efa9_v1 | 2026-10-30 | 1 | 1 | 1 |
| gold/orrery_triage_smoke | gold | orrery_triage_smoke | 2 | person gold_orrery_triage_smoke | 2026-10-30 | 0 | 0 | 0 |
| gold/orrery_triage_v1_labelled | gold | orrery_triage_v1_labelled | 65 | person gold_orrery_triage_v1_labelled | 2026-10-30 | 336 | 83 | 91 |
| silver/orrery_triage_v1_labelled_hand | silver | orrery_triage_v1_labelled | 3 | person silver_orrery_triage_v1_labelled_hand | 2026-10-30 | 0 | 0 | 0 |
| silver/orrery_triage_v1_labelled_outcomes | silver | orrery_triage_v1_labelled | 34 | rule silver_orrery_triage_v1_labelled_outcomes | 2026-11-02 | 0 | 0 | 0 |
| gold/scifact_200 | gold | scifact_200 | 200 | person the dataset's authors | 2026-09-28 | 26 | 26 | 26 |
| gold/scifact_40 | gold | scifact_40 | 40 | person the dataset's authors | 2026-09-28 | 0 | 0 | 0 |
| silver/smoke_v1 | silver | smoke_v1 | 5 | model silver_smoke_v1 the arm: reference_opus_v1 | 2026-09-25 | 1 | 1 | 1 |
| gold/summeval_100 | gold | summeval_100 | 100 | person the dataset's authors | 2026-10-15 | 41 | 41 | 41 |