Reproduce “Language Modeling is Monotone Compression” 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
Noam Mazor, Andrew Morgan, Rafael Pass
The authors show that efficient next-token prediction and efficient monotone, prefix-free compression can be transformed into one another. A predictor with expected negative log-likelihood at most H(D)+δ yields a code with expected length at most H(D)+δ+2 bits; conversely, a code within δ of Shannon entropy yields a predictor within δ. A cryptographic separation shows why arbitrary codes are not covered.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.