Reproduce “SumTablets: A Transliteration Dataset of Sumerian Tablets” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Cole Simmons, Richard Diehl Martinez, Dan Jurafsky

SumTablets pairs digital Unicode Sumerian glyphs with scholarly Latin-script transliterations in a dataset of 91,606 tablets. The authors train a supervised encoder–decoder model to convert glyph sequences into readings, comparing it with a frequency-weighted dictionary sampler. On held-out tablets, it scores 97.54 chrF versus 61.22. This measures character-level agreement with existing transliterations, not expert-verified correctness.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.