metadata
datasets:
- mideind/icelandic-common-crawl-corpus-IC3
language:
- is
library_name: transformers
Ice-gpt common crawl
A light-wheight gpt model trained on the icelandic-common-crawl-corpus from Mideind. It uses ice-tokenizer a bpe tokenizer for both training and inference.
Author
Sigurdur Haukur Birgisson