AI Model Modulation with Logits Redistribution

TL;DR

AIM introduces a training data-agnostic, retraining-free logits redistribution method for versatile model modulation, enabling dynamic utility and focus control across various tasks and architectures.

Contribution

It proposes a novel logits redistribution strategy for model modulation that does not require retraining and is applicable across multiple model types and tasks.

Findings

Effective utility modulation for varying output quality

Precise focus control on input features achieved

Applicable to diverse architectures and tasks

Abstract

Large-scale models are typically adapted to meet the diverse requirements of model owners and users. However, maintaining multiple specialized versions of the model is inefficient. In response, we propose AIM, a novel model modulation paradigm that enables a single model to exhibit diverse behaviors to meet the specific end requirements. AIM enables two key modulation modes: utility and focus modulations. The former provides model owners with dynamic control over output quality to deliver varying utility levels, and the latter offers users precise control to shift model's focused input features. AIM introduces a logits redistribution strategy that operates in a training data-agnostic and retraining-free manner. We establish a formal foundation to ensure AIM's regulation capability, based on the statistical properties of logits ordering via joint probability distributions. Our evaluation confirms AIM's practicality and versatility for Al model modulation, with tasks spanning image classification, semantic segmentation and text generation, and prevalent architectures including ResNet, SegFormer and Llama.