Reproduce “Tracing Morphology, Semantics, and Divergence in Language Acquisition” 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
Forough Shirin Abkenar, Haoshuo Fu, Liang Feng, Ning Yan, Masood S. Mortazavi
The authors track checkpoints in Pythia and OLMo3, comparing grammatical preferences with changes in word embeddings. They measure probabilities of preferred and dispreferred forms and track embedding similarity to final checkpoints. In tested English forms, morphology emerges by or before 100 million tokens, before semantic acquisition well after 1 billion tokens. Tests cover selected categories, not comprehensive language competence.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.