Hugging Face's logo

Datasets:
ulamai
/
STEM-RL-Tasks

Modalities:
Tabular
Text
Formats:
json
Languages:
English
Size:
< 1K
Tags:
stem
scientific-reasoning
reinforcement-learning
rlvr
verifiable-rewards
numerical-reasoning
Libraries:
Datasets
pandas
Polars
Dataset card Data Studio Files
xet
 Community
1
STEM-RL-Tasks / tasks /EARTH5 /solution /solution.json
przemekch's picture
przemekch
Publish STEM RL Tasks: 40 tasks across eight domains with verifiers and recorded attempts
6395f35 verified
Raw History Blame Contribute Delete
154 Bytes
{
"problem_id": "EARTH5",
"J_ClN_top": 4.481562,
"J_nat_base": 1.299111,
"I_psc": 3.170627,
"lambda_0_psc": 11.818555,
"u_node_1": 0.632615
}