kaiyuyue created evaluation artifacts for the paper 'Image Generation with a Sphere Encoder'. The artifacts include data statistic files and reference images used to calculate Fréchet Inception Distance (FID) for generative models. The artifacts cover several datasets, including CIFAR-10, ImageNet, Animal Faces, and Oxford Flowers.
Use Cases
- Calculate Fréchet Inception Distance (FID) scores for generative models based on provided reference images.
- Benchmark model performance across multiple datasets, including CIFAR-10 and ImageNet.
- Standardize evaluation workflows using pre-computed data statistics (fid_stats).
Strengths
- Artifacts are designed for a specific, well-defined evaluation metric (FID).
- Reference images and statistics cover multiple established benchmark datasets.
Limitations
- Description metadata is limited; actual data quality requires manual inspection after download.
- Column-level documentation is absent; field semantics must be inferred after download.
Provenance
- Source
- kaiyuyue
- Collection Method
- Created as evaluation artifacts for a research paper.
- Freshness
- Last updated 2026-02-27 02:18:10; freshness should be verified.