HaploOmni: Unified Single Transformer for Multimodal Video Understanding and Generation
This work addresses the challenge of efficient cross-modal compatibility for researchers and practitioners in multimodal AI, though it appears incremental as it builds on existing unified model frameworks.
The paper tackles the problem of building a single transformer for unified multimodal understanding and generation by proposing a multimodal warmup strategy, feature pre-scaling, and multimodal AdaLN techniques, resulting in HaploOmni achieving competitive performance across multiple image and video benchmarks with limited training costs.
With the advancement of language models, unified multimodal understanding and generation have made significant strides, with model architectures evolving from separated components to unified single-model frameworks. This paper explores an efficient training paradigm to build a single transformer for unified multimodal understanding and generation. Specifically, we propose a multimodal warmup strategy utilizing prior knowledge to extend capabilities. To address cross-modal compatibility challenges, we introduce feature pre-scaling and multimodal AdaLN techniques. Integrating the proposed technologies, we present the HaploOmni, a new single multimodal transformer. With limited training costs, HaploOmni achieves competitive performance across multiple image and video understanding and generation benchmarks over advanced unified models. All codes will be made public at https://github.com/Tencent/HaploVLM.