An English dataset composed of 10,000 real and 10,000 generated dreams.
Real dreams are from DreamBank. Generated dreams were produced with Oneirogen (0.5B, 1.5B and 7B), a language model for dream generation. Generation examples are available on my website.
This dataset can be used to determine differences between real and generated dreams. It can also be used to classify whether a dream narrative is generated or not.
This work was performed using HPC resources (Jean… See the full description on the dataset page:
https://huggingface.co/datasets/gustavecortal/The-android-and-the-human.
