mlsanityData to 14 Oct 2026

GeoStack: A Framework for Quasi-Abelian Knowledge Composition in VLMs

arXiv 2605.06477 ↗First posted 7 May 2026cs.CVcited by 0

On the map: Multimodal LLMs / vision-language models → Embodied and 3D vision-language models

Abstract (arXiv)

We address the challenge of knowledge composition in Vision-Language Models (VLMs), where accumulating expertise across multiple domains or tasks typically leads to catastrophic forgetting. We introduce GeoStack (Geometric Stacking), a modular framework that allows independently trained domain experts to be composed into a unified model. By imposing geometric and structural constraints on the adapter manifold, GeoStack ensures the foundational knowledge of the base model is preserved. Furthermore, we mathematically demonstrate a weight-folding property that achieves constant-time inference complexity (O(1)), regardless of the number of integrated experts. Experimental results across multi-domain adaptation and class-incremental learning show that GeoStack provides an efficient mechanism for long-term knowledge composition while significantly mitigating catastrophic forgetting. Code is available at https://github.com/QuantitativeImagingLaboratory/GeoStack.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Embodied and 3D vision-language models, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 18 of its 28 references that are in the map. First posted on arXiv 7 May 2026.

Builds on

18 of 28 references.

References · 18

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2015Distilling the Knowledge in a Neural Networkcited by 10,324
  3. 2016Overcoming catastrophic forgetting in neural networkscited by 4,787
  4. 2019Parameter-Efficient Transfer Learning for NLPcited by 4,236
  5. 2016Learning without Forgettingcited by 2,496
  6. 2016iCaRL: Incremental Classifier and Representation Learningcited by 2,333
  7. 2021Learning to Prompt for Vision-Language Modelscited by 2,281
  8. 2013Describing Textures in the Wildcited by 2,125
  9. 2017EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classificationcited by 1,496
  10. 2022Conditional Prompt Learning for Vision-Language Modelscited by 1,433
  11. 2022Editing Models with Task Arithmeticcited by 1,120
  12. 2021CLIP-Adapter: Better Vision-Language Models with Feature Adapterscited by 1,026
6 more
  1. 2020AdapterFusion: Non-Destructive Task Composition for Transfer Learningcited by 788
  2. 2022Mind the Gap: Understanding the Modality Gap in Multi-modal Contrastive Representation Learningcited by 561
  3. 2021Tip-Adapter: Training-free CLIP-Adapter for Better Vision-Language Modelingcited by 431
  4. 2016Recurrent Neural Network for Text Classification with Multi-Task Learningcited by 144
  5. 2025Decom-Renorm-Merge: Model Merging on the Right Space Improves Multitaskingcited by 2
  6. 2026BiCLIP: Domain Canonicalization via Structured Geometric Transformationcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026SpatialStack: Layered Geometry-Language Fusion for 3D VLM Spatial Reasoning
  2. 2026Skill-Conditioned Visual Geolocation for Vision-Language Models
  3. 2026Self-Evolving Spatial Reasoning in Vision Language Models via Geometric Logic Consistency
  4. 2025Dynamic Knowledge Integration for Enhanced Vision-Language Reasoning
  5. 2026GeoPID: Decomposing and Steering Visual Information in Vision-Language Models
  6. 2026GeoWorld-VLM: Geometry from World Models for Vision-Language Models
  7. 2023Compositional Semantics for Open Vocabulary Spatio-semantic Representations
  8. 2025Unveiling the Compositional Ability Gap in Vision-Language Reasoning Model