replicationbench@1.0

ReplicationBench - A benchmark for evaluating AI agents on reproducing computational results from astrophysics research papers. Adapted from Christine8888/replicationbench-release.

← Back to Registry

Run this task

CLI:

inspect eval inspect_harbor/replicationbench_1_0 --model openai/gpt-5

Python:

from inspect_ai import eval
from inspect_harbor import replicationbench_1_0

eval(replicationbench_1_0(), model="openai/gpt-5")

Dataset information

Harbor registry	replicationbench@1.0
Inspect task	replicationbench_1_0
Version	1.0
Samples	90
Paper	arxiv

See Task Parameters for the parameter set shared across all Harbor tasks.