Spaces:
Running on Zero
Running on Zero
File size: 1,495 Bytes
01e29bd 88c08a8 01e29bd 88c08a8 01e29bd 88c08a8 4586d4a 88c08a8 4586d4a 88c08a8 01e29bd 88c08a8 4586d4a 88c08a8 4586d4a 88c08a8 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 | ---
title: Qwen-Image 2.1 Rewriter Arena
emoji: ⚖️
colorFrom: indigo
colorTo: pink
sdk: gradio
sdk_version: 6.28.0
python_version: "3.12"
app_file: app.py
startup_duration_timeout: 1h
short_description: 9B vs 0.8B vs 2B prompt rewriters, rendered side by side
license: other
license_name: qwen-research
models:
- Qwen/Qwen-Image-2.1
- Qwen/Qwen-Image-2.1-PE-T2I
- ML-Intern-lab/Qwen-Image-2.1-PE-T2I-Pocket-0.8B
- ML-Intern-lab/Qwen-Image-2.1-PE-T2I-Pocket-2B
datasets:
- ML-Intern-lab/Qwen-Image-2.1-rewriter-distill
tags:
- text-to-image
- prompt-rewriting
- distillation
- benchmark
---
# Qwen-Image 2.1 Rewriter Arena
One request goes through the official 9B prompt rewriter
([Qwen/Qwen-Image-2.1-PE-T2I](https://huggingface.co/Qwen/Qwen-Image-2.1-PE-T2I)), the two pocket distillations
([0.8B](https://huggingface.co/ML-Intern-lab/Qwen-Image-2.1-PE-T2I-Pocket-0.8B), [2B](https://huggingface.co/ML-Intern-lab/Qwen-Image-2.1-PE-T2I-Pocket-2B)),
and optionally no rewriter at all. Every rewrite is rendered by [Qwen-Image 2.1](https://huggingface.co/Qwen/Qwen-Image-2.1) with a
shared seed and shown in a grid with its prompt, token count and rewrite time. A second tab shows the 40-request evaluation
contact sheets from [ML-Intern-lab/Qwen-Image-2.1-rewriter-distill](https://huggingface.co/datasets/ML-Intern-lab/Qwen-Image-2.1-rewriter-distill)
without using any GPU.
Built with Qwen. All models are under the Qwen Research License (non-commercial research use).
|