📊💡Dataset of interactions with ChatGPT
Wild Chat is a dataset of 1 million real user interactions with ChatGPT, characterized by a wide range of languages and a variety of prompts.
It was collected by providing free access to everyone to ChatGPT and GPT-4 in exchange for collecting chat history.
Using this dataset, the developers created the Llama-2-based WildLlama-7b-user-assistant bot WildLlama-7b-user-assistant, which is capable of predicting both the user's prompts and the responses that ChatGPT might choose.
The following script can also be used to load a dataset:
from datasets import load_dataset
dataset = load_dataset(“allenai/WildChat-1M”)