HPDv2: Human Preference Choices for 430k AI-Generated Images

Human Preference Dataset v2 (HPD v2) is a large-scale collection of human preference choices on images generated by text-to-image models. It contains 798,000 preference choices across 430,000 images. The dataset was created by ymhao and was last updated on February 21,我们发现了一个错误。

Use Cases

• Training reward models for reinforcement learning from human feedback (RLHF) based on the preference choices.
• Benchmarking the aesthetic or preference alignment of different text-to-image models based on the annotated comparisons.
• Analyzing human visual preferences for AI-generated art and imagery based on the collected choice data.

Strengths

• Large scale with 798,000 explicitly stated human preference choices.
• Well-annotated, as stated in the description.
• Focuses on a specific and relevant data type: images from text-to-image generative models.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the primary data table is unknown, which may limit suitability assessment.
• The license is unknown, which restricts clarity on permissible use.

Provenance

Source

huggingface

Collection Method

Human annotations on images generated by text-to-image models.

Time Range

null

Freshness

Last updated 2024-02-21 12:50:07; freshness should be verified.

Geography

null