CoPE: A Small Language Model for Steerable and Scalable Content Labeling

TL;DR

CoPE is a small, policy-steerable language model that achieves high accuracy in content labeling with minimal size, enabling scalable and fast moderation for online platforms.

Contribution

The paper introduces Contradictory Example Training and Binocular Labeling, novel methods for training and data generation that improve content labeling efficiency and accuracy.

Findings

CoPE matches or exceeds larger models in accuracy across seven harm areas.

A 9-billion-parameter CoPE model runs on a single GPU.

The methods enable rapid, unambiguous dataset creation for content policies.

Abstract

This paper details the methodology behind CoPE, a policy-steerable small language model capable of fast and accurate content labeling. We present a novel training curricula called Contradictory Example Training that enables the model to learn policy interpretation rather than mere policy memorization. We also present a novel method for generating content policies, called Binocular Labeling, which enables rapid construction of unambiguous training datasets. When evaluated across seven different harm areas, CoPE exhibits equal or superior accuracy to frontier models at only 1% of their size. We openly release a 9 billion parameter version of the model that can be run on a single consumer-grade GPU. Models like CoPE represent a paradigm shift for classifier systems. By turning an ML task into a policy writing task, CoPE opens up new design possibilities for the governance of online platforms.