TokLIP: Marry Visual Tokens to CLIP for Multimodal Comprehension and Generation
This work addresses computational and performance bottlenecks in multimodal AI systems, offering an incremental improvement over existing token-based approaches like Chameleon and Emu3.
The paper tackles the challenge of high training computational overhead and limited comprehension performance in token-based multimodal models by introducing TokLIP, a visual tokenizer that enhances comprehension with CLIP-level semantics and enables efficient end-to-end multimodal autoregressive training. It achieves exceptional data efficiency and improves both high-level semantic understanding and low-level generative capacity for autoregressive Transformers.
Pioneering token-based works such as Chameleon and Emu3 have established a foundation for multimodal unification but face challenges of high training computational overhead and limited comprehension performance due to a lack of high-level semantics. In this paper, we introduce TokLIP, a visual tokenizer that enhances comprehension by semanticizing vector-quantized (VQ) tokens and incorporating CLIP-level semantics while enabling end-to-end multimodal autoregressive training with standard VQ tokens. TokLIP integrates a low-level discrete VQ tokenizer with a ViT-based token encoder to capture high-level continuous semantics. Unlike previous approaches (e.g., VILA-U) that discretize high-level features, TokLIP disentangles training objectives for comprehension and generation, allowing the direct application of advanced VQ tokenizers without the need for tailored quantization operations. Our empirical results demonstrate that TokLIP achieves exceptional data efficiency, empowering visual tokens with high-level semantic understanding while enhancing low-level generative capacity, making it well-suited for autoregressive Transformers in both comprehension and generation tasks. The code and models are available at https://github.com/TencentARC/TokLIP.