This dataset is roughly 27k examples of erotica stories which I've fed through GPT-3.5-turbo-16k to obtain a summary, writing prompt, and tags as a response. I've filtered out all the refusals, and deleted a fair ammount of "GPT-isms". I'd still like to go through this again to prune any remaining low quality responses I've missed, but I think this is a good start. Most of the context size comes from the stories themselves, not the responses.
Please consider supporting my Patreon… See the full description on the dataset page:
https://huggingface.co/datasets/openerotica/erotica-analysis.
