TypewriterLM: 7.24B parameter language model trained on pre-1913 English text

Arxiv CS.CL (NLP/language models)

The 7.24B-parameter TypewriterLM is trained exclusively on English text predating 1913, addressing temporal leakage and data quality challenges. The model is designed for historical language analysis.

Arxiv CS.CL (NLP/language models)·Xiaoxi LuoZachary ShinnickNiclas GriesshaberYixuan WangJunchi YuFreda ShiPhilip TorrYao Lu

8 days ago