# Wings: Learning Multimodal LLMs without Text-only Forgetting

arXiv: https://arxiv.org/abs/2406.03496 · HTML: https://mlsanity.com/paper/2406.03496/

## Key facts

- Cited by 23 papers in the map (data through 14 Oct 2026).
- Placed in MLLM training and visual instruction tuning, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 123 of its 149 references that are in the map.
- First posted on arXiv 5 Jun 2024.

## Placed in

- Multimodal LLMs / vision-language models / [MLLM training and visual instruction tuning](https://mlsanity.com/mllm/mllm-training-and-visual-instruction-tuning/)

## Builds on (most cited first, 30 of 123)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [Scaling Laws for Neural Language Models](https://mlsanity.com/paper/2001.08361/) (2020)
- [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://mlsanity.com/paper/2301.12597/) (2023)
- [PaLM: Scaling Language Modeling with Pathways](https://mlsanity.com/paper/2204.02311/) (2022)
- [Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge](https://mlsanity.com/paper/1803.05457/) (2018)
- [Flamingo: a Visual Language Model for Few-Shot Learning](https://mlsanity.com/paper/2204.14198/) (2022)
- [BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation](https://mlsanity.com/paper/2201.12086/) (2022)
- [HellaSwag: Can a Machine Really Finish Your Sentence?](https://mlsanity.com/paper/1905.07830/) (2019)
- [Prefix-Tuning: Optimizing Continuous Prompts for Generation](https://mlsanity.com/paper/2101.00190/) (2021)
- [Deep reinforcement learning from human preferences](https://mlsanity.com/paper/1706.03741/) (2017)
- [Finetuned Language Models Are Zero-Shot Learners](https://mlsanity.com/paper/2109.01652/) (2021)
- [Program Synthesis with Large Language Models](https://mlsanity.com/paper/2108.07732/) (2021)
- [OPT: Open Pre-trained Transformer Language Models](https://mlsanity.com/paper/2205.01068/) (2022)
- [Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer](https://mlsanity.com/paper/1701.06538/) (2017)
- [Qwen Technical Report](https://mlsanity.com/paper/2309.16609/) (2023)
- [Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context](https://mlsanity.com/paper/2403.05530/) (2024)
- [Sigmoid Loss for Language Image Pre-Training](https://mlsanity.com/paper/2303.15343/) (2023)
- [Scaling Instruction-Finetuned Language Models](https://mlsanity.com/paper/2210.11416/) (2022)

## Built on by (most cited first, 23 of 23)

- [Devils in Middle Layers of Large Vision-Language Models: Interpreting, Detecting and Mitigating Object Hallucinations via Attention Lens](https://mlsanity.com/paper/2411.16724/) (2024)
- [Self-Rewarding Vision-Language Model via Reasoning Decomposition](https://mlsanity.com/paper/2508.19652/) (2025)
- [PRISM: Self-Pruning Intrinsic Selection Method for Training-Free Multimodal Data Selection](https://mlsanity.com/paper/2502.12119/) (2025)
- [LLaVA Steering: Visual Instruction Tuning with 500x Fewer Parameters through Modality Linear Representation-Steering](https://mlsanity.com/paper/2412.12359/) (2024)
- [A Survey on Multimodal Benchmarks: In the Era of Large AI Models](https://mlsanity.com/paper/2409.18142/) (2024)
- [Capability Instruction Tuning: A New Paradigm for Dynamic LLM Routing](https://mlsanity.com/paper/2502.17282/) (2025)
- [Zoom-Refine: Boosting High-Resolution Multimodal Understanding via Localized Zoom and Self-Refinement](https://mlsanity.com/paper/2506.01663/) (2025)
- [When Continue Learning Meets Multimodal Large Language Model: A Survey](https://mlsanity.com/paper/2503.01887/) (2025)
- [Locate-then-Merge: Neuron-Level Parameter Fusion for Mitigating Catastrophic Forgetting in Multimodal LLMs](https://mlsanity.com/paper/2505.16703/) (2025)
- [Training-Free Mitigation of Language Reasoning Degradation After Multimodal Instruction Tuning](https://mlsanity.com/paper/2412.03467/) (2024)
- [Mitigating Intra- and Inter-modal Forgetting in Continual Learning of Unified Multimodal Models](https://mlsanity.com/paper/2512.03125/) (2025)
- [MLAN: Language-Based Instruction Tuning Preserves and Transfers Knowledge in Multimodal Language Models](https://mlsanity.com/paper/2411.10557/) (2024)
- [GenieBlue: Integrating both Linguistic and Multimodal Capabilities for Large Language Models on Mobile Devices](https://mlsanity.com/paper/2503.06019/) (2025)
- [LF2AR: Accounting for Layerwise Dynamics to Improve Multimodal Adaptation of Language Models](https://mlsanity.com/paper/2503.06211/) (2025)
- [PlaM: Training-Free Plateau-Guided Model Merging for Better Visual Grounding in MLLMs](https://mlsanity.com/paper/2601.07645/) (2026)
- [LaVi: Efficient Large Vision-Language Models via Internal Feature Modulation](https://mlsanity.com/paper/2506.16691/) (2025)
- [A Thousand Words or An Image: Studying the Influence of Persona Modality in Multimodal LLMs](https://mlsanity.com/paper/2502.20504/) (2025)
- [Instruction-Following Evaluation of Large Vision-Language Models](https://mlsanity.com/paper/2512.23572/) (2025)
- [Text Capability Loss in Vision-Language Adaptation: An Attention-Sink Diagnosis](https://mlsanity.com/paper/2609.00746/) (2026)
- [OmniEvalKit: A Modular, Lightweight Toolbox for Evaluating Large Language Model and its Omni-Extensions](https://mlsanity.com/paper/2412.06693/) (2024)
- [Selective LoRA for Visual Tokens and Attention Heads](https://mlsanity.com/paper/2512.19219/) (2025)
- [XPERT: Expert Knowledge Transfer for Effective Training of Language Models](https://mlsanity.com/paper/2605.08842/) (2026)
- [NeuPAT: Neuron-aware Plasticity Allocation Tuning for Language-Preserving MLLMs](https://mlsanity.com/paper/2608.08107/) (2026)

## Abstract (arXiv)

Multimodal large language models (MLLMs), initiated with a trained LLM, first align images with text and then fine-tune on multimodal mixed inputs. However, the MLLM catastrophically forgets the text-only instructions, which do not include images and can be addressed within the initial LLM. In this paper, we present Wings, a novel MLLM that excels in both text-only dialogues and multimodal comprehension. Analyzing MLLM attention in multimodal instructions reveals that text-only forgetting is related to the attention shifts from pre-image to post-image text. From that, we construct extra modules that act as the boosted learner to compensate for the attention shift. The complementary visual and textual learners, like "wings" on either side, are connected in parallel within each layer's attention block. Initially, image and text inputs are aligned with visual learners operating alongside the main attention, balancing focus on visual elements. Textual learners are later collaboratively integrated with attention-based routing to blend the outputs of the visual and textual learners. We design the Low-Rank Residual Attention (LoRRA) to guarantee high efficiency for learners. Our experimental results demonstrate that Wings outperforms equally-scaled MLLMs in both text-only and visual question-answering tasks. On a newly constructed Interleaved Image-Text (IIT) benchmark, Wings exhibits superior performance from text-only-rich to multimodal-rich question-answering tasks.