# OASIS: Online Sample Selection for Continual Visual Instruction Tuning

arXiv: https://arxiv.org/abs/2506.02011 · HTML: https://mlsanity.com/paper/2506.02011/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- Builds on 62 of its 87 references that are in the map.
- First posted on arXiv 27 May 2025.

## Builds on (most cited first, 30 of 62)

- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding](https://mlsanity.com/paper/1804.07461/) (2018)
- [Overcoming catastrophic forgetting in neural networks](https://mlsanity.com/paper/1612.00796/) (2016)
- [Improved Baselines with Visual Instruction Tuning](https://mlsanity.com/paper/2310.03744/) (2023)
- [Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answering](https://mlsanity.com/paper/2209.09513/) (2022)
- [SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systems](https://mlsanity.com/paper/1905.00537/) (2019)
- [ChartQA: A Benchmark for Question Answering about Charts with Visual and Logical Reasoning](https://mlsanity.com/paper/2203.10244/) (2022)
- [DocVQA: A Dataset for VQA on Document Images](https://mlsanity.com/paper/2007.00398/) (2020)
- [Experience Replay for Continual Learning](https://mlsanity.com/paper/1811.11682/) (2018)
- [mPLUG-Owl: Modularization Empowers Large Language Models with Multimodality](https://mlsanity.com/paper/2304.14178/) (2023)
- [Shikra: Unleashing Multimodal LLM's Referential Dialogue Magic](https://mlsanity.com/paper/2306.15195/) (2023)
- [MagicBrush: A Manually Annotated Dataset for Instruction-Guided Image Editing](https://mlsanity.com/paper/2306.10012/) (2023)
- [Deep Learning on a Data Diet: Finding Important Examples Early in Training](https://mlsanity.com/paper/2107.07075/) (2021)
- [Beyond neural scaling laws: beating power law scaling via data pruning](https://mlsanity.com/paper/2206.14486/) (2022)
- [Overfitting in adversarially robust deep learning](https://mlsanity.com/paper/2002.11569/) (2020)
- [PathVQA: 30000+ Questions for Medical Visual Question Answering](https://mlsanity.com/paper/2003.10286/) (2020)
- [DVQA: Understanding Data Visualizations via Question Answering](https://mlsanity.com/paper/1801.08163/) (2018)
- [Selection via Proxy: Efficient Data Selection for Deep Learning](https://mlsanity.com/paper/1906.11829/) (2019)
- [Not All Samples Are Created Equal: Deep Learning with Importance Sampling](https://mlsanity.com/paper/1803.00942/) (2018)
- [From Quantity to Quality: Boosting LLM Performance with Self-Guided Data Selection for Instruction Tuning](https://mlsanity.com/paper/2308.12032/) (2023)
- [SemDeDup: Data-efficient learning at web-scale through semantic deduplication](https://mlsanity.com/paper/2303.09540/) (2023)
- [Rainbow Memory: Continual Learning with a Memory of Diverse Samples](https://mlsanity.com/paper/2103.17230/) (2021)
- [IconQA: A New Benchmark for Abstract Diagram Understanding and Visual Language Reasoning](https://mlsanity.com/paper/2110.13214/) (2021)
- [Prioritized Training on Points that are Learnable, Worth Learning, and Not Yet Learnt](https://mlsanity.com/paper/2206.07137/) (2022)
- [Progressive Prompts: Continual Learning for Language Models](https://mlsanity.com/paper/2301.12314/) (2023)
- [HQ-Edit: A High-Quality Dataset for Instruction-based Image Editing](https://mlsanity.com/paper/2404.09990/) (2024)
- [Coverage-centric Coreset Selection for High Pruning Rates](https://mlsanity.com/paper/2210.15809/) (2022)
- [NumGLUE: A Suite of Fundamental yet Challenging Mathematical Reasoning Tasks](https://mlsanity.com/paper/2204.05660/) (2022)
- [MultiInstruct: Improving Multi-Modal Zero-Shot Learning via Instruction Tuning](https://mlsanity.com/paper/2212.10773/) (2022)
- [Adapting Auxiliary Losses Using Gradient Similarity](https://mlsanity.com/paper/1812.02224/) (2018)

## Built on by (most cited first, 9 of 9)

- [Reinforcement Fine-Tuning Naturally Mitigates Forgetting in Continual Post-Training](https://mlsanity.com/paper/2507.05386/) (2025)
- [Continual Learning for Generative AI: From LLMs to MLLMs and Beyond](https://mlsanity.com/paper/2506.13045/) (2025)
- [SAME: Stabilized Mixture-of-Experts for Multimodal Continual Instruction Tuning](https://mlsanity.com/paper/2602.01990/) (2026)
- [Spectral Imbalance Causes Forgetting in Low-Rank Continual Adaptation](https://mlsanity.com/paper/2602.00722/) (2026)
- [CL-VISTA: Benchmarking Continual Learning in Video Large Language Models](https://mlsanity.com/paper/2604.00677/) (2026)
- [MAny: Merge Anything for Multimodal Continual Instruction Tuning](https://mlsanity.com/paper/2604.14016/) (2026)
- [StrLoRA: Towards Streaming Continual Visual Instruction Tuning for MLLMs](https://mlsanity.com/paper/2605.16353/) (2026)
- [CRAM: Centroid-Routing and Adaptive MoE for Multimodal Continual Instruction Tuning](https://mlsanity.com/paper/2606.02502/) (2026)
- [ProtoAda: Prototype-Guided Adaptive Adapter Expansion and Geometric Consolidation for Multimodal Continual Instruction Tuning](https://mlsanity.com/paper/2606.02576/) (2026)

## Abstract (arXiv)

In continual instruction tuning (CIT) scenarios, where new instruction tuning data continuously arrive in an online streaming manner, training delays from large-scale data significantly hinder real-time adaptation. Data selection can mitigate this overhead, but existing strategies often rely on pretrained reference models, which are impractical in CIT setups since future data are unknown. Recent reference model-free online sample selection methods address this, but typically select a fixed number of samples per batch (e.g., top-k), making them vulnerable to distribution shifts where informativeness varies across batches. To address these limitations, we propose OASIS, an adaptive online sample selection approach for CIT that (1) selects informative samples by estimating each sample's informativeness relative to all previously seen data, beyond batch-level constraints, and (2) minimizes informative redundancy of selected samples through iterative selection score updates. Experiments on various large foundation models show that OASIS, using only 25 percent of the data, achieves comparable performance to full-data training and outperforms the state-of-the-art sampling methods.