Dataset · allenai
One million human-ChatGPT dialogues
allenai/WildChat-1M
Real multilingual user conversations with ChatGPT featuring demographics and prompt metadata.
What's inside
The dataset contains 1 million dialogues between humans and ChatGPT in various languages. The card also specifies demographic data such as state, country, hashed IP addresses, and prompt headers. Distributed under the ODC-BY license.
What you can do with it
- Supervised fine-tuning of chat models on real dialogues (SFT)
- Analyzing user behavior and multilingual prompts
- Evaluating assistant response quality
- Tasks
- Text generation, Question answering
- Size
- 100K–1M rows
- Downloads this month
- 37.4K
- Likes
- 486
Models trained on it
- microsoft/UserLM-8bText generation 3.2K
- BSC-LT/ALIA-40b-instruct-2606-GGUFText generation 2.9K
- Mungert/UserLM-8b-GGUFText generation 2.7K
- BSC-LT/salamandra-7b-instruct-2606Text generation 1.8K
- mradermacher/salamandra-7b-instruct-2606-i1-GGUF 1.5K
- mradermacher/UserLM-8b-i1-GGUF 1.2K
- mradermacher/ALIA-40b-instruct-2605-i1-GGUF 1.2K
- mradermacher/ALIA-40b-instruct-2601-i1-GGUF 1.2K
How to load it
from datasets import load_dataset
ds = load_dataset("allenai/WildChat-1M")