Language Models are General-Purpose Interfaces

TL;DR

This paper introduces a versatile framework using language models as universal interfaces to various foundation models, enabling multi-modal processing and combining benefits of causal and non-causal modeling for improved generalization.

Contribution

It proposes a semi-causal language modeling approach that unifies diverse modality encoders with language models, enhancing flexibility and performance across tasks.

Findings

Outperforms specialized models on multiple benchmarks.

Enables in-context learning and instruction following with finetuned encoders.

Shows strong zero-shot and few-shot generalization capabilities.

Abstract

Foundation models have received much attention due to their effectiveness across a broad range of downstream applications. Though there is a big convergence in terms of architecture, most pretrained models are typically still developed for specific tasks or modalities. In this work, we propose to use language models as a general-purpose interface to various foundation models. A collection of pretrained encoders perceive diverse modalities (such as vision, and language), and they dock with a language model that plays the role of a universal task layer. We propose a semi-causal language modeling objective to jointly pretrain the interface and the modular encoders. We subsume the advantages and capabilities from both causal and non-causal modeling, thereby combining the best of two worlds. Specifically, the proposed method not only inherits the capabilities of in-context learning and open-ended generation from causal language modeling, but also is conducive to finetuning because of the bidirectional encoders. More importantly, our approach seamlessly unlocks the combinations of the above capabilities, e.g., enabling in-context learning or instruction following with finetuned encoders. Experimental results across various language-only and vision-language benchmarks show that our model outperforms or is competitive with specialized models on finetuning, zero-shot generalization, and few-shot learning.