Login or sign up
For business
Average rating
0.0/ 5

Based on 0 reviews

50%
40%
30%
20%
10%
Ask AI

Ask AI about any% labs

Generated with AI on Trustburn

AI-generated from reviews on this page. Verify anything important with the company directly.

No reviews yet

Be the first to review any% labs.

Write a review

News

No news yet. This is where the business shares its own updates and press.

Photos

No photos yet. Photos shared by reviewers will show up here.

Widgets

Boost your business trust by displaying a widget on your website — a simple way to convert your visitors into customers.

Get the widget for your site →

About any% labs

any% labs is a small research team that studies reward hacking in reinforcement learning with verifiable rewards (RLVR) environments. The team investigates failures in real training runs and develops tools to detect, analyze, and prevent reward exploits, such as cases where models achieve high reward scores without actually solving the intended task. Its current focus is auditing open-source RLVR environments from Prime Intellect's Hub, identifying shortcuts that earn reward without genuine task completion. The group invites contact from others who have training runs or environments that may be scoring well for the wrong reasons, and is also recruiting collaborators interested in working on reward hacking detection and prevention.