Fine-tuning Multi-modal LLMs with ART: Art-based Reinforcement Training
For practitioners deploying fine-tuned MLLMs in high-throughput engines like vLLM, ART enables parameter-efficient fine-tuning without computational graph modifications.
ART (Art-based Reinforcement Training) fine-tunes multimodal LLMs by optimizing raw visual input pixels instead of modifying model weights, achieving accuracy competitive with LoRA on mathematics and structured-tool-use benchmarks.
There are two main Parameter-Efficient Fine-Tuning (PEFT) techniques for Large Language Models (LLMs). While Low-Rank Adaptation (LoRA) introduces additional weights between the LLM layers, Soft Prompting introduces additional fine-tuning-specific raw tokens to an LLM input. However, both require modification to the computational graphs of precompiled, preoptimized LLMs. As a result, neither is fully supported in high-throughput engines like vLLM. We propose fine-tuning with ART (Art-based Reinforcement Training). The method injects information into a frozen Multimodal Large Language Model (MLLM) by optimizing only its raw visual input, thus enabling the soft-token approach on pre-compiled computational graphs. It relies on backpropagation of gradients back into a plain pixel array and thus supports any fine-tuning objective. Moreover, the optimized visual input can be stylized as task-relevant computational artworks. The approach's effectiveness is confirmed for different sizes of a popular open Qwen architecture and for several textual benchmarks. Specifically, ART reaches accuracy competitive with LoRA across mathematics and structured-tool-use benchmarks.