Dataset · Gryphe
Character card roleplay dialogues
Gryphe/Sonnet3.5-Charcard-Roleplay
9,736 simulated English roleplay dialogues in ShareGPT format, heavily NSFW.
What's inside
The dataset contains 9,736 roleplay dialogues generated from character cards using unfiltered Sonnet 3.5. Each card is supplemented by a simulated user—male or female with four different personalities. The data is in English in ShareGPT format. The card warns that many scenes are strongly NSFW and potentially disturbing.
What you can do with it
- Roleplay chat model or LoRA fine-tuning
- Testing safety filters and NSFW classifiers
- Researching dialogue style based on character cards
- Tasks
- Text generation
- Languages
- English
- Size
- 1K–10K rows
- Downloads this month
- 688
- Likes
- 101
Models trained on it
- Chun121/Qwen3-4B-RPG-Roleplay-V2Text generation 22.8K
- bartowski/EVA-Qwen2.5-14B-v0.2-GGUFText generation 8.3K
- Epiculous/Violet_Twilight-v0.2-GGUFText generation 7.9K
- mradermacher/EVA-Qwen2.5-14B-v0.0-i1-GGUF 5.2K
- mradermacher/EVA-Qwen2.5-14B-v0.2-i1-GGUF 4K
- mradermacher/EVA-Qwen2.5-7B-v0.1-i1-GGUF 3.8K
- mradermacher/EVA-Qwen2.5-7B-v0.0-i1-GGUF 3K
- bartowski/EVA-Qwen2.5-32B-v0.2-GGUFText generation 3K
How to load it
from datasets import load_dataset
ds = load_dataset("Gryphe/Sonnet3.5-Charcard-Roleplay")