Improving Multi-modal Large Language Model through Boosting Vision Capabilities

TL;DR

This work presents Multimodal LoRA (MM-LoRA), a module designed to enhance the decoder, and introduces the Query Ladder adapter (QLadder) to improve the visual encoder.

Abstract

We focus on improving the visual understanding capability for boosting the vision-language models. We propose \textbf{Arcana}, a multiModal language model, which introduces two crucial techniques. First, we present Multimodal LoRA (MM-LoRA), a module designed to enhance the decoder. Unlike traditional language-driven decoders, MM-LoRA consists of two parallel LoRAs -- one for vision and one for language -- each with its own parameters. This disentangled parameters design allows for more specialized learning in each modality and better integration of multimodal information. Second, we introduce the Query Ladder adapter (QLadder) to improve the visual encoder. QLadder employs a learnable ``\textit{ladder}'' structure to deeply aggregates the intermediate representations from the frozen pretrained visual encoder (e.g., CLIP image encoder). This enables the model to learn new and informative visual features, as well as remaining the powerful capabilities of the pretrained visual encoder. These techniques collectively enhance Arcana's visual perception power, enabling it to leverage improved visual information for more accurate and contextually relevant outputs across various multimodal scenarios. Extensive experiments and ablation studies demonstrate the effectiveness and generalization capability of our Arcana. The code and re-annotated data are available at \url{https://arcana-project-page.github.io}.

Figures (8)

• Figure 1: (a) Sampled some VQA examples involving color, quantity, small objects, and localization tasks, showcasing the importance of visual recognition capabilities for multimodal language models (MLLMs). (b) Contrasting Arcana's multimodal decoder with mainstream methods' language driven decoder. The language-driven decoder employs a language decoder (LLMs) directly to handle tokens from different modalities, which may lead to modality interference and performance degradation. In contrast, the multimodal decoder independently processes different token types to avoid modality interference. (c) illustrates the structures of different visual encoders and the resulting number of visual tokens obtained. The bar chart displays the model's performance across various architectures.
• Figure 2: (a) The architecture of the Arcana. (b) The training pipeline of Arcana. MM-LoRA is optional during the pre-training phase.
• Figure 3: (a) The farmework of MM-LoRA vs. LoRA. MM-LoRA introduces two new hyperparameters, $\beta$ and $\gamma$, to control the ranks of the visual and language LoRAs, respectively. Notably, we set $\beta + \gamma = 1$ to ensure that MM-LoRA has the same number of parameters as LoRA. (b) The architecture of the visual encoder includes the QLadder adapter and CLIP. The QLadder adapter consists of cross-attention and FFN layers, with weights initialized from those of CLIP.
• Figure 4: Ablation of $\beta$ and $\gamma$ in MM-LoRA. The default rank is set to 256, while $\beta$ and $\gamma$ are used to control the rank values in visual and language LoRA components, respectively.
• Figure 5: Ablation of query number in QLadder. N$_q$ represents the number of learnable query.