Unlocking Pretrained LLMs for Motion-Related Multimodal Generation: A Fine-Tuning Approach to Unify Diffusion and Next-Token Prediction

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Tanaka, Shinichi, Wang, Zhao, Kato, Yoichi, Ohya, Jun
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916646986186752
author Tanaka, Shinichi
Wang, Zhao
Kato, Yoichi
Ohya, Jun
author_facet Tanaka, Shinichi
Wang, Zhao
Kato, Yoichi
Ohya, Jun
contents In this paper, we propose a unified framework that leverages a single pretrained LLM for Motion-related Multimodal Generation, referred to as MoMug. MoMug integrates diffusion-based continuous motion generation with the model's inherent autoregressive discrete text prediction capabilities by fine-tuning a pretrained LLM. This enables seamless switching between continuous motion output and discrete text token prediction within a single model architecture, effectively combining the strengths of both diffusion- and LLM-based approaches. Experimental results show that, compared to the most recent LLM-based baseline, MoMug improves FID by 38% and mean accuracy across seven metrics by 16.61% on the text-to-motion task. Additionally, it improves mean accuracy across eight metrics by 8.44% on the text-to-motion task. To the best of our knowledge, this is the first approach to integrate diffusion- and LLM-based generation within a single model for motion-related multimodal tasks while maintaining low training costs. This establishes a foundation for future advancements in motion-related generation, paving the way for high-quality yet cost-efficient motion synthesis.
format Preprint
id arxiv_https___arxiv_org_abs_2503_06119
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Unlocking Pretrained LLMs for Motion-Related Multimodal Generation: A Fine-Tuning Approach to Unify Diffusion and Next-Token Prediction
Tanaka, Shinichi
Wang, Zhao
Kato, Yoichi
Ohya, Jun
Machine Learning
In this paper, we propose a unified framework that leverages a single pretrained LLM for Motion-related Multimodal Generation, referred to as MoMug. MoMug integrates diffusion-based continuous motion generation with the model's inherent autoregressive discrete text prediction capabilities by fine-tuning a pretrained LLM. This enables seamless switching between continuous motion output and discrete text token prediction within a single model architecture, effectively combining the strengths of both diffusion- and LLM-based approaches. Experimental results show that, compared to the most recent LLM-based baseline, MoMug improves FID by 38% and mean accuracy across seven metrics by 16.61% on the text-to-motion task. Additionally, it improves mean accuracy across eight metrics by 8.44% on the text-to-motion task. To the best of our knowledge, this is the first approach to integrate diffusion- and LLM-based generation within a single model for motion-related multimodal tasks while maintaining low training costs. This establishes a foundation for future advancements in motion-related generation, paving the way for high-quality yet cost-efficient motion synthesis.
title Unlocking Pretrained LLMs for Motion-Related Multimodal Generation: A Fine-Tuning Approach to Unify Diffusion and Next-Token Prediction
topic Machine Learning
url https://arxiv.org/abs/2503.06119