Some Newer Rp: Curated Roleplay Conversations from Glimmer-Data

1,938 ShareGPT-style roleplay conversations curated from the glimmer-data project. The dataset, created by ToastyPigeon, includes separate files for SFW and NSFW content, with each record tagged by its source sub-dataset. It was last updated on May 31, 2026.

Use Cases

• Fine-tune language models for roleplay dialogue based on the ShareGPT-style conversation format.
• Filter and analyze sub-datasets for specific conversational styles based on the provided source field.
• Train content moderation classifiers using the separated SFW and NSFW conversation files.

Strengths

• 1,938 total records across two curated files.
• Each record includes a source field for filtering individual sub-datasets.
• Content is explicitly separated into SFW (1,121 records) and NSFW (817 records) files.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for individual source sub-datasets beyond the listed examples is unknown.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Curated from the glimmer-data project.

Collection Method

ShareGPT-style conversations aggregated from multiple sources.

Freshness

Last updated 2026-05-31 20:14:23; freshness should be verified.