Tangent Model Composition for Ensembling and Continual Fine-tuning
This addresses the challenge of high computational costs and sequential bias in ensembling and continual fine-tuning for machine learning practitioners, though it is incremental as it builds on existing fine-tuning paradigms.
The paper tackles the problem of efficiently combining and updating fine-tuned models by introducing Tangent Model Composition (TMC), which improves accuracy by 4.2% compared to ensembling non-linearly fine-tuned models while reducing inference cost by 2.5x to 10x.
Tangent Model Composition (TMC) is a method to combine component models independently fine-tuned around a pre-trained point. Component models are tangent vectors to the pre-trained model that can be added, scaled, or subtracted to support incremental learning, ensembling, or unlearning. Component models are composed at inference time via scalar combination, reducing the cost of ensembling to that of a single model. TMC improves accuracy by 4.2% compared to ensembling non-linearly fine-tuned models at a 2.5x to 10x reduction of inference cost, growing linearly with the number of component models. Each component model can be forgotten at zero cost, with no residual effect on the resulting inference. When used for continual fine-tuning, TMC is not constrained by sequential bias and can be executed in parallel on federated data. TMC outperforms recently published continual fine-tuning methods almost uniformly on each setting -- task-incremental, class-incremental, and data-incremental -- on a total of 13 experiments across 3 benchmark datasets, despite not using any replay buffer. TMC is designed for composing models that are local to a pre-trained embedding, but could be extended to more general settings. The code is available at: https://github.com/tianyu139/tangent-model-composition