Meme joke text (template names + captions + labels), structured for
template-conditioned generation of byte-level language models. Built for
training BDH (Baby Dragon Hatchling, vocab=256 raw bytes) — contains text
only, no images.
Files
File
Contents
meme_text_corpus.txt
Byte-level training rendering (TEMPLATE: / TEXT: / --- records)
meme_text_corpus.jsonl
One JSON record per line with metadata (schema below)
Schema… See the full description on the dataset page: https://huggingface.co/datasets/Alogotron/meme-text-corpus.
