A LLM trained from scratch on bulgarian data. The model and the model's tokenizer are trained from scratch on bulgarian data from the chitanka dataset.
Perprelixty - 6.75