Mechanistic Interpretability Benchmark

university

https://mib-bench.github.io

AI & ML interests

Principled evaluation of mechanistic interpretability methods.

Recent Activity

amueller updated a Space 21 days ago

mib-bench/leaderboard

hij authored a paper 4 months ago

Blackbox Model Provenance via Palimpsestic Membership Inference

amueller updated a Space 5 months ago

mib-bench/leaderboard

View all activity

Collections 1

spaces 1

MIB Leaderboard

Leaderboard for the Mechanistic Interpretability Benchmark

models 3

mib-bench/mib-circuits-example

Updated Jul 23, 2025

mib-bench/mib-causalvariable-example

Updated May 29, 2025

mib-bench/interpbench

Updated May 17, 2025

datasets 7

mib-bench/ravel

Viewer • Updated May 31, 2025 • 117k • 31

mib-bench/arithmetic_subtraction

Viewer • Updated May 31, 2025 • 20.9k • 168

mib-bench/arithmetic_addition

Viewer • Updated May 31, 2025 • 40.4k • 225

mib-bench/ioi

Viewer • Updated May 29, 2025 • 21k • 1.11k

mib-bench/arc_easy

Viewer • Updated Jan 25, 2025 • 4.01k • 469

mib-bench/arc_challenge

Viewer • Updated Jan 25, 2025 • 2k • 289

mib-bench/copycolors_mcqa

Viewer • Updated Jan 16, 2025 • 1.89k • 308