Vision-Language Model Token Pruning Diagram
Overview
Four color-coded panels for two-phase visual-token compression in a VLM. It is an editable PowerPoint diagram adapted from the original paper figure under CC BY 4.0.
What the diagram shows
(a) Joint two-phase compression with a compute / memory / accuracy trade-off triangle. (b) Spatial density signal: patchify into H×W tokens, 3×3 / 5×5 / 7×7 neighborhoods, the density-score formula ρ_ij and a score map. (c) Prefill: density-guided bipartite merge from source set A to destination set B, giving N′ = N(1−r) tokens. (d) Decode: rank visual KV by ρ, keep top-k and evict the rest before the LLM decoder. Every box, arrow, label and math term is a native shape you can recolor, resize and relabel for your own slides, posters or papers.
File format
The download contains .pptx (1 slide, 14.83 × 7.37 in; opens in PowerPoint, Keynote, WPS and LibreOffice Impress; Google Slides works too, but complex freeforms may shift slightly), .svg and .drawio. Includes preview.png (full resolution) and a 1200 px web preview.
Shapes and editing
The file has 354 native editable shapes: 242 auto-shapes, 69 text boxes, 6 freeforms, 37 lines/connectors. 2 embedded images (see below). You can edit all text and the density-score formula (fraction, sum, ⟨𝐱_ij, 𝐱_i′j′⟩, 𝒩_w(i, j)), every token / KV cell, the neighborhood grids, score map and color bar, the bipartite merge arcs, bar chart, triangle and callout boxes.
Embedded images and fonts
2 items are embedded images: the example food photo and its patchified copy. Text uses Times New Roman, which is preinstalled on Windows and macOS (LibreOffice uses the metric-compatible Liberation Serif). A few calligraphic math letters (such as 𝒩, ℰ, 𝒮) use Cambria Math, which ships with Microsoft Office; other apps substitute a math font. Sub- and superscripts are real formatted text runs, not equation objects. The SVG keeps text as text; the draw.io file is a shape-by-shape conversion and may need small nudges. This is a faithful redraw for reuse, not the authors' original source file.
Source & license Adapted from Figure 3 of TwinPrune: Density-Aware Two-Phase Token Pruning for Vision-Language Models by Kai Liu, Anqi Li, Junxian Li, Zhixin Wang, Zhikai Chen, Renjing Pei, Yulun Zhang (arXiv:2610.17530, https://arxiv.org/abs/2610.17530; code: https://github.com/Kai-Liu001/TwinPrune). Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Redrawn as editable shapes; changes made by 叫我Alonzo就好了/alonzolab. Not endorsed by the original authors. Attribution to the original paper is required when you publish or share the figure; the exact credit line is in LICENSE-ATTRIBUTION.txt inside each download.