Image To Video Human Preference Seedance 1 Pro: 6k Human Evaluations

Approximately 6,000 human responses from around 2,000 annotators were collected to evaluate the Seedance 1 Pro video generation model on a benchmark. The data was gathered in roughly 5 minutes using the Rapidata Python API. The dataset was published by Rapidata and last updated on August 11, 2025.

Use Cases

• Benchmarking video generation models based on human preference scores.
• Training reward models for reinforcement learning from human feedback (RLHF) in video generation.
• Analyzing the correlation between model outputs and human evaluator ratings.
• Studying the efficiency of rapid, large-scale data annotation via API.

Strengths

• Contains approximately 6,000 human evaluation responses.
• Data was collected from a pool of around 2,000 human annotators.
• Collection was performed rapidly, reportedly in roughly 5 minutes.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Rapidata

Collection Method

Collected via the Rapidata Python API from human annotators.

Freshness

Last updated 2025-08-11 12:23:22.