ava-FLUX.1-latents-10k

# Dataset Card for AVA FLUX.1-schnell VAE Latents 10k

<!-- Provide a quick summary of the dataset. -->

9.7k latents from FLUX.1-schnell VAE for the AVA dataset

## Dataset Details

### Dataset Description

<!-- Provide a longer summary of what this dataset is. -->

- **Curated by:** [Dave Lage](https://huggingface.com/rockerBOO)
- **License:** [Apache-2.0](https://huggingface.co/datasets/rockerBOO/ava-FLUX.1-latents-10k)

### Dataset Sources [optional]

<!-- Provide the basic links for the dataset. -->

- **Repository:** [More Information Needed]

## Uses

<!-- Address questions around how the dataset is intended to be used. -->

Latents are a sample from the AVA dataset. These latents were created using the [FLUX.1-schnell](https://github.com/black-forest-labs/FLUX.1-schnell) VAE model. Use of these latents is intended for research purposes only. Useful for Aesthetic Predictions using AVA dataset AVA.txt for aesthetic predictive modeling.

## Dataset Structure

<!-- This section provides a description of the dataset fields, and additional information about the dataset structure such as criteria used to create the splits, relationships between data points, etc. -->

- image_id: image id from the AVA dataset
- latents: flattened list of latents
- shape_channels: channels of the VAE (16)
- shape_height: height of the latents
- shape_width: width of the latents
- original_width: width of the image
- original_height: height of the image
- filename: filename of the image

## Dataset Creation

#### Data Collection and Processing

<!-- This section describes the data collection and processing process such as data selection criteria, filtering and normalization methods, tools and libraries used, etc. -->

Randomized dataset was collected from the AVA dataset. Latents captured from the FLUX.1-schnell VAE. Latents are flattened into a list and dimensions are stored in the dataset parquet file.

#### Personal and Sensitive Information

<!-- State whether the dataset contains data that might be considered personal, sensitive, or private (e.g., data that reveals addresses, uniquely identifiable names or aliases, racial or ethnic origins, sexual orientations, religious beliefs, political opinions, financial or health data, etc.). If efforts were made to anonymize the data, describe the anonymization process. -->

Latents are from the Flux.1-schnell VAE of images in the AVA dataset. Images are not included in this dataset.

## Bias, Risks, and Limitations

<!-- This section is meant to convey both technical and sociotechnical limitations. -->

- Dataset is non randomized and not a complete dataset so there is not enough data to create appropriate results.

# AVA FLUX.1-schnell VAE 隐变量10k 数据集卡片

<!-- 提供该数据集的简要概述。 -->

该数据集包含针对AVA数据集（Aesthetic Visual Analysis）提取的、来自FLUX.1-schnell 变分自编码器（Variational Autoencoder，VAE）的9700个隐变量（latents）。

## 数据集详情

### 数据集描述

<!-- 提供该数据集的详细概述。 -->

- **整理者：** [Dave Lage](https://huggingface.com/rockerBOO)
- **许可证：** [Apache-2.0](https://huggingface.co/datasets/rockerBOO/ava-FLUX.1-latents-10k)

### 数据集来源[可选]

<!-- 提供该数据集的基础链接信息。 -->

- **代码仓库：** [需补充更多信息]

## 用途

<!-- 说明本数据集的预期使用场景。 -->

本数据集的隐变量（latents）取自AVA数据集，通过[FLUX.1-schnell](https://github.com/black-forest-labs/FLUX.1-schnell)的变分自编码器（VAE）模型生成。本数据集仅可用于研究用途，适用于结合AVA数据集的AVA.txt文件开展美学预测建模的美学预测任务。

## 数据集结构

<!-- 本节描述数据集的字段信息，以及数据集结构的其他相关细节，例如划分数据集的标准、数据点之间的关联等。 -->

- **图像ID（image_id）：** 来自AVA数据集的图像标识符
- **隐变量（latents）：** 扁平化处理后的隐变量列表
- **通道维度（shape_channels）：** 变分自编码器的通道数（固定为16）
- **高度维度（shape_height）：** 隐变量的高度
- **宽度维度（shape_width）：** 隐变量的宽度
- **原始图像宽度（original_width）：** 原始图像的宽度
- **原始图像高度（original_height）：** 原始图像的高度
- **文件名（filename）：** 原始图像的文件名

## 数据集构建

#### 数据收集与处理

<!-- 本节描述数据收集与处理流程，包括数据选择标准、过滤与归一化方法、所使用的工具与库等。 -->

本数据集从AVA数据集随机采样得到，隐变量通过FLUX.1-schnell VAE模型提取，并被扁平化为列表形式，其维度信息存储在数据集的Parquet文件中。

#### 个人与敏感信息

<!-- 说明本数据集是否包含可被视为个人、敏感或隐私的数据（例如泄露地址、唯一可识别的姓名或别名、种族或族裔出身、性取向、宗教信仰、政治观点、财务或健康数据等）。若已对数据进行匿名化处理，请描述匿名化流程。 -->

本数据集的隐变量源自AVA数据集内图像对应的FLUX.1-schnell VAE特征，原始图像并未包含在本数据集中。

## 偏差、风险与局限性

<!-- 本节旨在说明技术与社会技术层面的局限性。 -->

- 本数据集并非随机采样所得，且并非完整数据集，数据量不足以生成可靠的实验结果。