CoolFace
Datasetpublic

caiotheodoro/plumb

Plumb Gold tasks and the held-out eval. The study is on the collection. Config n What benchmark 1000 Held-out eval, seed 777. Never in train. train_handseeded 223 Mix matched to the eval, including PASS. train_ornith 58 Ornith-1.5 proposals that passed the oracle. train_blended 281 Both of the above. Leakprobe vs benchmark: exact signature overlap 0. curriculum train n sw-recall precision exact hand-seeded 223 0.318 [0.290, 0.347] 0.308 [0.279… See the full description on the dataset page: https://huggingface.co/datasets/caiotheodoro/plumb.

sourceHugging Faceapache-2.0updated 1mo agoView on Hugging Face
0likes106downloads
7 commits on main
d45dcb81mo ago

evals: flat published rows + CIs (reproduces the post tables)

caiotheodoro
c10ace41mo ago

dataset card: add evals config

caiotheodoro
b9452891mo ago

Cut repeated lecture; cards are TL;DR

caiotheodoro
1a6a1b31mo ago

Rewrite card: thesis, domain, and standalone context

caiotheodoro
36825971mo ago

Upload PATH_A_RESULTS.json with huggingface_hub

caiotheodoro
375c5041mo ago

Upload folder using huggingface_hub

caiotheodoro
47d2b381mo ago

initial commit

caiotheodoro