# Beyond Routing Saturation: A Long-Horizon Class-Incremental Perspective on Expert Routing in Multimodal Continual Instruction Tuning

arXiv: https://arxiv.org/abs/2608.01437 · HTML: https://mlsanity.com/paper/2608.01437/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Prompt learning and adaptation of CLIP, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 9 of its 114 references that are in the map.
- First posted on arXiv 2 Aug 2026.

## Placed in

- Multimodal LLMs / vision-language models / [Prompt learning and adaptation of CLIP](https://mlsanity.com/mllm/prompt-learning-and-adaptation-of-clip/)
- Mixture-of-experts / [Mixture of LoRA experts](https://mlsanity.com/moe/mixture-of-lora-experts/)

## Builds on (most cited first, 9 of 9)

- [Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer](https://mlsanity.com/paper/1701.06538/) (2017)
- [Decoupled Weight Decay Regularization](https://mlsanity.com/paper/1711.05101/) (2017)
- [PathVQA: 30000+ Questions for Medical Visual Question Answering](https://mlsanity.com/paper/2003.10286/) (2020)
- [PMC-VQA: Visual Instruction Tuning for Medical Visual Question Answering](https://mlsanity.com/paper/2305.10415/) (2023)
- [Don't Stop Learning: Towards Continual Learning for the CLIP Model](https://mlsanity.com/paper/2207.09248/) (2022)
- [MLLM-CL: Continual Learning for Multimodal Large Language Models](https://mlsanity.com/paper/2506.05453/) (2025)
- [PlantExpertVQA: A Visual Question Answering Dataset for Benchmarking Vision-Language Models in Plant Science](https://mlsanity.com/paper/2508.17117/) (2025)
- [Prism: A Plug-in Reproducible Infrastructure for Scalable Multimodal Continual Instruction Tuning](https://mlsanity.com/paper/2605.26110/) (2026)
- [Beyond Point-wise Neural Collapse: A Topology-Aware Hierarchical Classifier for Class-Incremental Learning](https://mlsanity.com/paper/2605.11904/) (2026)

## Abstract (arXiv)

Multimodal Continual Instruction Tuning (MCIT) enables multimodal large language models to acquire new tasks sequentially while retaining previously learned capabilities. Many recent methods maintain task-specific LoRA experts and route each input to one or more experts at inference. Yet the task-identification problem underlying expert routing remains under-explored. We show that routing is nearly saturated on widely used MCIT benchmarks. Textual fingerprints that leak task identity and short 4--10-task sequences with few competing experts jointly obscure the long-horizon routing problem. To expose this challenge, we introduce FLEX (Fingerprint-reduced Long-horizon Expert eXamination), a 34-task long-horizon MCIT benchmark with weakened textual fingerprints. FLEX groups tasks with similar instruction and answer formats but diverse visual and knowledge domains, normalizes their outer templates, and evaluates routing over a substantially larger expert pool. Crucially, we formulate progressive-LoRA routing as soft task-as-class Multimodal Class-Incremental Learning (MCIL): each task defines an incremental routing class, whose complete score distribution supplies the LoRA mixture weights, with hard routing as a discrete special case. FLEX exposes this expanding task-identification challenge, while the MCIL formulation provides a principled interface for transferring CIL methods to expert routing. We instantiate PureLoRA as a controlled baseline and adapt four CIL methods to four MCIT frameworks without modifying their LoRA experts or generation pipelines. Our plug-in routers improve strict LoRA matching by up to 16.3 percentage points and overall MacroScore by up to 4.6 points. Code is available at: https://github.com/RINC-CL/FLEX