# From Pixels and Words to Waves: A Unified Framework for Spectral Dictionary vLLMs

arXiv: https://arxiv.org/abs/2506.18943 · HTML: https://mlsanity.com/paper/2506.18943/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in VLM benchmarks and visual reasoning, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 22 of its 31 references that are in the map.
- First posted on arXiv 22 Jun 2025.

## Placed in

- Multimodal LLMs / vision-language models / [VLM benchmarks and visual reasoning](https://mlsanity.com/mllm/vlm-benchmarks-and-visual-reasoning/)

## Builds on (most cited first, 22 of 22)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://mlsanity.com/paper/2103.14030/) (2021)
- [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://mlsanity.com/paper/2301.12597/) (2023)
- [Fully Convolutional Networks for Semantic Segmentation](https://mlsanity.com/paper/1411.4038/) (2014)
- [Mamba: Linear-Time Sequence Modeling with Selective State Spaces](https://mlsanity.com/paper/2312.00752/) (2023)
- [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://mlsanity.com/paper/2104.09864/) (2021)
- [Improved Baselines with Visual Instruction Tuning](https://mlsanity.com/paper/2310.03744/) (2023)
- [Flamingo: a Visual Language Model for Few-Shot Learning](https://mlsanity.com/paper/2204.14198/) (2022)
- [Efficiently Modeling Long Sequences with Structured State Spaces](https://mlsanity.com/paper/2111.00396/) (2021)
- [Rethinking Attention with Performers](https://mlsanity.com/paper/2009.14794/) (2020)
- [Linformer: Self-Attention with Linear Complexity](https://mlsanity.com/paper/2006.04768/) (2020)
- [Oscar: Object-Semantics Aligned Pre-training for Vision-Language Tasks](https://mlsanity.com/paper/2004.06165/) (2020)
- [Kosmos-2: Grounding Multimodal Large Language Models to the World](https://mlsanity.com/paper/2306.14824/) (2023)
- [OFA: Unifying Architectures, Tasks, and Modalities Through a Simple Sequence-to-Sequence Learning Framework](https://mlsanity.com/paper/2202.03052/) (2022)
- [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://mlsanity.com/paper/2107.14795/) (2021)
- [Hyena Hierarchy: Towards Larger Convolutional Language Models](https://mlsanity.com/paper/2302.10866/) (2023)
- [Global Filter Networks for Image Classification](https://mlsanity.com/paper/2107.00645/) (2021)
- [PaLI-3 Vision Language Models: Smaller, Faster, Stronger](https://mlsanity.com/paper/2310.09199/) (2023)
- [Reducing Deep Network Complexity via Sparse Hierarchical Fourier Interaction Networks](https://mlsanity.com/paper/1801.01451/) (2017)
- [Spectral Dictionary Learning for Generative Image Modeling](https://mlsanity.com/paper/2504.17804/) (2025)
- [Graph Laplacian Wavelet Transformer via Learnable Spectral Decomposition](https://mlsanity.com/paper/2505.07862/) (2025)

## Built on by (most cited first, 1 of 1)

- [Operator-Based Machine Intelligence: A Hilbert Space Framework for Spectral Learning and Symbolic Reasoning](https://mlsanity.com/paper/2507.21189/) (2025)

## Abstract (arXiv)

Vision-language models (VLMs) unify computer vision and natural language processing in a single architecture capable of interpreting and describing images. Most state-of-the-art systems rely on two computationally intensive components: convolutions in the vision encoder and quadratic self-attention for multimodal fusion. This work removes both by introducing a spectral dictionary token mixer, which represents each image patch or wordpiece as a sparse combination of learnable frequency atoms. Our 1.1B-parameter prototype, SDict-VLM, achieves BLEU-4 of 39.2, CIDEr of 127.5, and SPICE of 27.0 on MS-COCO captioning, along with 50.3 percent accuracy on VQAv2. These results close approximately 85 percent of the performance gap to BLIP-2 while using 60 percent fewer parameters, 2.3 times less peak GPU memory, and 2.2 times faster inference than PaLI-3. To our knowledge, this is the first VLM to eliminate both convolutions and self-attention while matching mid-scale transformer baselines. In addition to its O(L log L) complexity, the shared frequency dictionary enables transparent cross-modal alignment and offers a tunable trade-off between accuracy and compute, paving the way for efficient and interpretable VLMs.