SAEScientist-Bench

What is SAEScientist-Bench?

Evaluates AI agents performing sparse autoencoder feature interpretation on Gemma-2-9B-IT, measuring feature rank, activation separation, and steering ability…

Released

2026-09-08

Evaluates

Agents & Tool Use, General AI, cs.AI

Openness

unknown

Importer

Claire Radar

Review status

ai-reviewed

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2609.09113

SAEScientist-Bench paper and code

• Paper https://arxiv.org/abs/2609.09113
• Code https://github.com/Trae1ounG/SAEScientist

Which sources cite SAEScientist-Bench?

• SAEScientist-Bench

Related Agents & Tool Use, General AI, cs.AI benchmarks

• $\tau^\tau$-Bench: An Environment for End-To-End, Realistic Agent Construction
• 3DCodeBench
• 5-Dialects-BN
• A3S-Bench
• AARRI-Bench
• ABC-Bench
• ActBench
• ActiveSciBench