---
base_model:
- ByteDance-Seed/BAGEL-7B-MoT
datasets:
- WeiChow/WEAVE
language:
- en
- zh
tags:
  - image
  - image-editing
  - multimodal
license: apache-2.0
pipeline_tag: any-to-any
library_name: transformers
---

## Bagel-weave

<div align="center">

[![arXiv](https://img.shields.io/badge/arXiv-2511.11434-b31b1b.svg)](https://arxiv.org/abs/2511.11434)
[![Dataset](https://img.shields.io/badge/🤗%20Huggingface-Dataset-yellow)](https://huggingface.co/datasets/WeiChow/Weave/)
[![Checkpoint](https://img.shields.io/badge/🤗%20Huggingface-CKPT-blue)](https://huggingface.co/WeiChow/Bagel-weave)
[![GitHub](https://img.shields.io/badge/GitHub-Repo-181717?logo=github)](https://github.com/weichow23/weave)
[![Page](https://img.shields.io/badge/Home-Page-b3.svg)](https://weichow23.github.io/weave/)

</div>

> 📌 This is the official Bagel implementation of 
> **Weave: A Benchmark for Evaluating Multimodal Editing Models**

Bagel-weave is finetuned by [WEAVE-100k](https://huggingface.co/datasets/WeiChow/WEAVE). 

<img src="quality.png" alt="Test Set Categories" width="700" />

## License
BAGEL-weave is licensed under the Apache 2.0 license. It is finetuned from [Bagel-7B-MOT](https://huggingface.co/ByteDance-Seed/BAGEL-7B-MoT), [Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct) and [siglip-so400m-14-384-flash-attn2](https://huggingface.co/HuggingFaceM4/siglip-so400m-14-384-flash-attn2) model, and uses the [FLUX.1-schnell VAE model](https://huggingface.co/black-forest-labs/FLUX.1-schnell), all under Apache 2.0.

## ✍️ Citation
```bibtex

```