Dataset · DJRHails
Fresh public LLM dialogues from 9 platforms
DJRHails/sharetrawl
Over 256 thousand publicly shared chats with language models, updated weekly.
What's inside
The dataset contains 256,218 public LLM dialogues gathered from 9 platforms, including examples with toxic responses. About 29% of the links (75,000) are sourced from the ShareChat project list, while the rest were discovered independently by the authors. The corpus is updated weekly and provided in English and other multilingual formats, though the exact file format is not specified in the card.
What you can do with it
- SFT training of conversational models on current user queries
- Training and testing moderation filters and toxicity detectors
- Tasks
- Text generation, Question answering
- Languages
- English, Multilingual
- Size
- 100K–1M rows
- Downloads this month
- 38.2K
- Likes
- 3
- Access
- on request on Hugging Face
How to load it
from datasets import load_dataset
ds = load_dataset("DJRHails/sharetrawl")