README.md

---
dataset_info:
  features:
  - name: id
    dtype: string
  - name: conversations
    list:
    - name: role
      dtype: string
    - name: content
      dtype: string
  splits:
  - name: train
    num_bytes: 277884785
    num_examples: 160000
  download_size: 126665150
  dataset_size: 277884785
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
task_categories:
- image-to-text
---

<h1 align="center"> Text-Based Reasoning About Vector Graphics </h1>

<p align="center">
<a href="https://mikewangwzhl.github.io/VDLM">🌐 Homepage</a>
•
<a href="https://arxiv.org/abs/2404.06479">📃 Paper</a>
•
<a href="https://huggingface.co/datasets/mikewang/PVD-160K" >🤗 Data (PVD-160k)</a>
•
<a href="https://huggingface.co/mikewang/PVD-160k-Mistral-7b" >🤗 Model (PVD-160k-Mistral-7b)</a>
•
<a href="https://github.com/MikeWangWZHL/VDLM" >💻 Code</a>

</p>


We observe that current *large multimodal models (LMMs)* still struggle with seemingly straightforward reasoning tasks that require precise perception of low-level visual details, such as identifying spatial relations or solving simple mazes. In particular, this failure mode persists in question-answering tasks about vector graphics—images composed purely of 2D objects and shapes.

![Teaser](https://github.com/MikeWangWZHL/VDLM/blob/main/figures/teaser.png?raw=true)

To solve this challenge, we propose **Visually Descriptive Language Model (VDLM)**, a visual reasoning framework that operates with intermediate text-based visual descriptions—SVG representations and learned Primal Visual Description, which can be directly integrated into existing LLMs and LMMs. We demonstrate that VDLM outperforms state-of-the-art large multimodal models, such as GPT-4V, across various multimodal reasoning tasks involving vector graphics. See our [paper](https://arxiv.org/abs/2404.06479) for more details.
![Overview](https://github.com/MikeWangWZHL/VDLM/blob/main/figures/overview.png?raw=true)