Dataset Viewer
The dataset could not be loaded because the splits use different data file formats, which is not supported. Read more about the splits configuration. Click for more details.
Couldn't infer the same data file format for all splits. Got {NamedSplit('train'): ('parquet', {}), NamedSplit('validation'): ('json', {}), NamedSplit('test'): ('parquet', {})}
Error code: FileFormatMismatchBetweenSplitsError
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
NCII Guard — Synthetic Generation Outputs (v1)
LLM-generated realistic editing prompts used for the synthetic slices of yjernite/ncii-guard-train-v1 and yjernite/ncii-guard-eval-v1.
- Generator: Qwen/Qwen2.5-7B-Instruct (bf16, temp 0.9, top_p 0.95), seeded scenario dimensions (relationship, phrasing style, image/video, name pools per census ethnicity×gender cell) — not fill-in templates.
synth_train.parquet: 2,999 rows (800 safe / 499 explicit / 1,700 ncii-risk), columnsprompt,label,source,media.synth_eval.parquet: 950 rows (200 / 150 / 600) with demographic proxy columns; the eval draw uses a disjoint seed stream.seeds_used.json: every seed drawn (train + eval streams) for reproducibility.
These rows are synthetic. NCII-risk examples were generated as realistic abuse-adjacent prompts for training a safety classifier; they describe requests, not people, and no real person's data is involved.
- Downloads last month
- 2,500