Reve-AI-Halfmoon_t2i_human_preference

<style>

.vertical-container {
display: flex;
flex-direction: column;
gap: 60px;
}

.image-container img {
max-height: 250px; /* Set the desired height */
margin:0;
object-fit: contain; /* Ensures the aspect ratio is maintained */
width: auto; /* Adjust width automatically based on height */
box-sizing: content-box;
}

.image-container {
display: flex; /* Aligns images side by side */
justify-content: space-around; /* Space them evenly */
align-items: center; /* Align them vertically */
gap: .5rem
}

.container {
width: 90%;
margin: 0 auto;
}

.text-center {
text-align: center;
}

.score-amount {
margin: 0;
margin-top: 10px;
}

.score-percentage {Score:
font-size: 12px;
font-weight: semi-bold;
}

</style>

# Rapidata Reve AI Halfmoon Preference

<a href="https://www.rapidata.ai">
<img src="https://cdn-uploads.huggingface.co/production/uploads/66f5624c42b853e73e0738eb/jfxR79bOztqaC6_yNNnGU.jpeg" width="400" alt="Dataset visualization">
</a>

This T2I dataset contains over 195k human responses from over 51k individual annotators, collected in just ~1 Day using the [Rapidata Python API](https://docs.rapidata.ai), accessible to anyone and ideal for large scale evaluation.
Evaluating Reve AI Halfmoon across three categories: preference, coherence, and alignment.

Explore our latest model rankings on our [website](https://www.rapidata.ai/benchmark).

If you get value from this dataset and would like to see more in the future, please consider liking it ❤️

## Overview

This T2I dataset contains over 195k human responses from over 51k individual annotators, collected in just ~1 Day.
Evaluating Halfmoon-4-4-2025 across three categories: preference, coherence, and alignment.

The evaluation consists of 1v1 comparisons between Halfmoon-4-4-2025 and 13 other models: OpenAI 4o-26-3-25, Ideogram V2, Recraft V2, Lumina-15-2-25, Frames-23-1-25, Imagen-3, Flux-1.1-pro, Flux-1-pro, DALL-E 3, Midjourney-5.2, Stable Diffusion 3, Aurora and Janus-7b.
> **Note:** The number following the model name (e.g., Halfmoon-4-4-2025) represents the date (April 4, 2025) on which the images were generated to give an understanding of what model version was used.
## Alignment

The alignment score quantifies how well an video matches its prompt. Users were asked: "Which image matches the description better?".

<div class="vertical-container">
<div class="container">
<div class="text-center">
<q>A black colored banana.</q>
</div>
<div class="image-container">
<div>
<h3 class="score-amount">Halfmoon-4-4-2025 </h3>
<div class="score-percentage">Score: 100%</div>
<img style="border: 3px solid #18c54f;" src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/rysxNO-VYHYGjloCy8uHr.jpeg" width=500>
</div>
<div>
<h3 class="score-amount">Midjourney-5.2 </h3>
<div class="score-percentage">Score: 0%</div>
<img src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/UmQ2HdNUuh-7zILudKCC8.jpeg" width=500>
</div>
</div>
</div>

<div class="container">
<div class="text-center">
<q>A bird scaring a scarecrow.</q>
</div>
<div class="image-container">
<div>
<h3 class="score-amount">Halfmoon-4-4-2025</h3>
<div class="score-percentage">Score: 20%</div>
<img src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/BSQTSdYt-a_ePVu8Bc79W.jpeg" width=500>
</div>
<div>
<h3 class="score-amount">DALL-E 3</h3>
<div class="score-percentage">Score: 80%</div>
<img style="border: 3px solid #18c54f;" src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/lelElAS7Lf8re7WMdnMfN.jpeg" width=500>
</div>
</div>
</div>
</div>

## Coherence

The coherence score measures whether the generated video is logically consistent and free from artifacts or visual glitches. Without seeing the original prompt, users were asked: "Which image has **more** glitches and is **more** likely to be AI generated?"

<div class="vertical-container">
<div class="container">
<div class="image-container">
<div>
<h3 class="score-amount">Halfmoon-4-4-2025 </h3>
<div class="score-percentage">Glitch Rating: 7.3%</div>
<img style="border: 3px solid #18c54f;" src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/auqOer3bjzyzDjTXddk2s.jpeg" width=500>
</div>
<div>
<h3 class="score-amount">Janus-7B </h3>
<div class="score-percentage">Glitch Rating: 92.7%</div>
<img src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/zSlaNCAVIbiAudr6_hSmU.jpeg" width=500>
</div>
</div>
</div>

<div class="container">
<div class="image-container">
<div>
<h3 class="score-amount">Halfmoon-4-4-2025 </h3>
<div class="score-percentage">Glitch Rating: 100%</div>
<img src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/0cb4-CTGY-I4cIaliQ1My.jpeg" width=500>
</div>
<div>
<h3 class="score-amount">Flux-1.1 Pro</h3>
<div class="score-percentage">Glitch Rating: 0%</div>
<img style="border: 3px solid #18c54f;" src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/vm7_r3CIGe5cy5kiEHCZW.jpeg" width=500>
</div>
</div>
</div>
</div>

## Preference

The preference score reflects how visually appealing participants found each image, independent of the prompt. Users were asked: "Which image do you prefer?"

<div class="vertical-container">
<div class="container">
<div class="image-container">
<div>
<h3 class="score-amount">Halfmoon-4-4-2025</h3>
<div class="score-percentage">Score: 63.6%</div>
<img style="border: 3px solid #18c54f;" src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/DCSHfz8-hqVWRzHHZG6dn.jpeg" width=500>
</div>
<div>
<h3 class="score-amount">Frames-23-1-25</h3>
<div class="score-percentage">Score: 36.4%</div>
<img src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/N7rUm0rWG3HF-EkZElqUb.jpeg" width=500>
</div>
</div>
</div>

<div class="container">
<div class="image-container">
<div>
<h3 class="score-amount">Halfmoon </h3>
<div class="score-percentage">Score: 34.0%</div>
<img src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/6mA0bup-qSLSoZwx0Pk4e.jpeg" width=500>
</div>
<div>
<h3 class="score-amount">Flux 1 Pro </h3>
<div class="score-percentage">Score: 76.0%</div>
<img style="border: 3px solid #18c54f;" src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/5cTri7WBRnwaY7ELQ1LkQ.jpeg" width=500>
</div>
</div>
</div>
</div>

## About Rapidata

Rapidata's technology makes collecting human feedback at scale faster and more accessible than ever before. Visit [rapidata.ai](https://www.rapidata.ai/) to learn more about how we're revolutionizing human feedback collection for AI development.

<style>

.vertical-container {
display: flex;
flex-direction: column;
gap: 60px;
}

.image-container img {
max-height: 250px; /* Set the desired height */
margin:0;
object-fit: contain; /* Ensures the aspect ratio is maintained */
width: auto; /* Adjust width automatically based on height */
box-sizing: content-box;
}

.image-container {
display: flex; /* Aligns images side by side */
justify-content: space-around; /* Space them evenly */
align-items: center; /* Align them vertically */
gap: .5rem
}

.container {
width: 90%;
margin: 0 auto;
}

.text-center {
text-align: center;
}

.score-amount {
margin: 0;
margin-top: 10px;
}

.score-percentage {
font-size: 12px;
font-weight: semi-bold;
}

</style>

<h1>Rapidata Reve AI Halfmoon 偏好数据集</h1>

<a href="https://www.rapidata.ai">
<img src="https://cdn-uploads.huggingface.co/production/uploads/66f5624c42b853e73e0738eb/jfxR79bOztqaC6_yNNnGU.jpeg" width="400" alt="数据集可视化">
</a>

<p>本<strong>文本到图像（Text-to-Image, T2I）</strong>数据集收录了来自5.1万余名独立标注者的19.5万余条人类标注反馈，仅用约1天时间便通过<a href="https://docs.rapidata.ai">Rapidata Python应用程序编程接口（API）</a>完成采集，面向所有用户开放，非常适合开展大规模模型评估。</p>
<p>本次评估围绕偏好性、一致性与对齐性三大维度对Reve AI Halfmoon模型展开。</p>
<p>您可访问我们的<a href="https://www.rapidata.ai/benchmark">官方网站</a>查看最新的模型排名榜单。</p>
<p>如果您从本数据集获益并希望后续获取更多同类资源，欢迎为其点赞❤️</p>

<h2>数据集概览</h2>

<p>本T2I数据集收录了来自5.1万余名独立标注者的19.5万余条人类反馈，采集周期仅约1天。</p>
<p>本次评估将从偏好性、一致性与对齐性三个维度对Halfmoon-4-4-2025模型进行评测。</p>
<p>本次评估采用1v1对比的方式，将Halfmoon-4-4-2025与其余13款模型进行对比：OpenAI 4o-26-3-25、Ideogram V2、Recraft V2、Lumina-15-2-25、Frames-23-1-25、Imagen-3、Flux-1.1-pro、Flux-1-pro、DALL-E 3、Midjourney-5.2、Stable Diffusion 3、Aurora以及Janus-7b。</p>
<blockquote><strong>注意：</strong> 模型名称后的数字（如Halfmoon-4-4-2025）代表生成对应图像的日期（2025年4月4日），以此说明所使用的模型版本。</blockquote>

<h2>对齐性</h2>

<p>对齐性评分用于量化生成图像与提示词的匹配程度。本次调研向参与者提出的问题为：“哪张图像更贴合描述内容？”</p>

<div class="vertical-container">
<div class="container">
<div class="text-center">
<q>一根黑色的香蕉。</q>
</div>
<div class="image-container">
<div>
<h3 class="score-amount">Halfmoon-4-4-2025</h3>
<div class="score-percentage">得分：100%</div>
<img style="border: 3px solid #18c54f;" src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/rysxNO-VYHYGjloCy8uHr.jpeg" width="500" alt="Halfmoon-4-4-2025生成的黑色香蕉图像">
</div>
<div>
<h3 class="score-amount">Midjourney-5.2</h3>
<div class="score-percentage">得分：0%</div>
<img src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/UmQ2HdNUuh-7zILudKCC8.jpeg" width="500" alt="Midjourney-5.2生成的黑色香蕉图像">
</div>
</div>
</div>

<div class="container">
<div class="text-center">
<q>一只鸟驱赶稻草人。</q>
</div>
<div class="image-container">
<div>
<h3 class="score-amount">Halfmoon-4-4-2025</h3>
<div class="score-percentage">得分：20%</div>
<img src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/BSQTSdYt-a_ePVu8Bc79W.jpeg" width="500" alt="Halfmoon-4-4-2025生成的驱稻草人场景图像">
</div>
<div>
<h3 class="score-amount">DALL-E 3</h3>
<div class="score-percentage">得分：80%</div>
<img style="border: 3px solid #18c54f;" src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/lelElAS7Lf8re7WMdnMfN.jpeg" width="500" alt="DALL-E 3生成的驱稻草人场景图像">
</div>
</div>
</div>
</div>

<h2>一致性</h2>

<p>一致性评分用于衡量生成图像是否具备逻辑自洽性，且未出现视觉伪影或视觉瑕疵。本次调研在不展示原始提示词的前提下，向参与者提出问题：“哪张图像存在<strong>更多</strong>瑕疵，且<strong>更有可能</strong>是AI生成的？”</p>

<div class="vertical-container">
<div class="container">
<div class="image-container">
<div>
<h3 class="score-amount">Halfmoon-4-4-2025</h3>
<div class="score-percentage">瑕疵评级：7.3%</div>
<img style="border: 3px solid #18c54f;" src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/auqOer3bjzyzDjTXddk2s.jpeg" width="500" alt="Halfmoon-4-4-2025生成的图像">
</div>
<div>
<h3 class="score-amount">Janus-7B</h3>
<div class="score-percentage">瑕疵评级：92.7%</div>
<img src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/zSlaNCAVIbiAudr6_hSmU.jpeg" width="500" alt="Janus-7B生成的图像">
</div>
</div>
</div>

<div class="container">
<div class="image-container">
<div>
<h3 class="score-amount">Halfmoon-4-4-2025</h3>
<div class="score-percentage">瑕疵评级：100%</div>
<img src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/0cb4-CTGY-I4cIaliQ1My.jpeg" width="500" alt="Halfmoon-4-4-2025生成的图像">
</div>
<div>
<h3 class="score-amount">Flux-1.1 Pro</h3>
<div class="score-percentage">瑕疵评级：0%</div>
<img style="border: 3px solid #18c54f;" src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/vm7_r3CIGe5cy5kiEHCZW.jpeg" width="500" alt="Flux-1.1 Pro生成的图像">
</div>
</div>
</div>
</div>

<h2>偏好性</h2>

<p>偏好性评分用于反映参与者对每张图像的视觉吸引力评价，与原始提示词无关。本次调研向参与者提出的问题为：“你更偏好哪张图像？”</p>

<div class="vertical-container">
<div class="container">
<div class="image-container">
<div>
<h3 class="score-amount">Halfmoon-4-4-2025</h3>
<div class="score-percentage">得分：63.6%</div>
<img style="border: 3px solid #18c54f;" src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/DCSHfz8-hqVWRzHHZG6dn.jpeg" width="500" alt="Halfmoon-4-4-2025生成的图像">
</div>
<div>
<h3 class="score-amount">Frames-23-1-25</h3>
<div class="score-percentage">得分：36.4%</div>
<img src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/N7rUm0rWG3HF-EkZElqUb.jpeg" width="500" alt="Frames-23-1-25生成的图像">
</div>
</div>
</div>

<div class="container">
<div class="image-container">
<div>
<h3 class="score-amount">Halfmoon</h3>
<div class="score-percentage">得分：34.0%</div>
<img src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/6mA0bup-qSLSoZwx0Pk4e.jpeg" width="500" alt="Halfmoon生成的图像">
</div>
<div>
<h3 class="score-amount">Flux 1 Pro</h3>
<div class="score-percentage">得分：76.0%</div>
<img style="border: 3px solid #18c54f;" src="https://cdn-uploads.huggingface.co/production/uploads/664dcc6296d813a7e15e170e/5cTri7WBRnwaY7ELQ1LkQ.jpeg" width="500" alt="Flux 1 Pro生成的图像">
</div>
</div>
</div>
</div>

<h2>关于Rapidata</h2>

<p>Rapidata的技术使大规模人类反馈采集工作比以往任何时候都更加快捷、易用。请访问<a href="https://www.rapidata.ai/">rapidata.ai官方网站</a>，了解我们如何革新AI开发领域的人类反馈采集流程。</p>