[KongchangAI]
Benchmark

ReproBridge-Bench

A benchmark for evaluating the ability to reproduce machine learning papers, comprising 1,029 tasks from 36 papers, anchored by human-verified reference scores, and designed with dual-condition controls and independent auditing.

Timeline (last 90 days)

Nov 6

ReproBridge-Bench contains 1,029 tasks from 387 experiments in 36 papers, with each task anchored by five human-verified reference scores.

Unverified50%
Nov 6

In the evaluation design of ReproBridge-Bench, agents attempt tasks in fresh sessions using one of two conditions—original code or refined artifact—and an independent Auditor is responsible for checking execution evidence.

Unverified50%
Nov 6

Across the four evaluated models, refined artifacts raised the average reproduction success rate from 76.4% to 92.6%.

Unverified50%

All Facts (3)

Source Articles