P-Bench and Fisher-R1 show that small LLM agents trained with outcome-grounded reinforcement learning on synthetic hypothesis-testing tasks can outperform frontier models at statistically valid p-value reporting and decisions.
Infiagent-dabench: Evaluating agents on data analysis tasks,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
dataset 1
citation-polarity summary
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1roles
dataset 1polarities
unclear 1representative citing papers
citing papers explorer
-
Fisher-R1: Training LLM Agents for Reliable Hypothesis Testing
P-Bench and Fisher-R1 show that small LLM agents trained with outcome-grounded reinforcement learning on synthetic hypothesis-testing tasks can outperform frontier models at statistically valid p-value reporting and decisions.