Reproduce “LLM Flow Processes for Text-Conditioned Regression” 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
Felix Biggs, Samuel Willis
LLM-Flow Processes combine a neural diffusion process for coherent function trajectories with language-model predictions at selected locations, using text to guide sampling. On Mauna Loa carbon dioxide forecasting, it reports a CRPS of 0.56 ± 0.07 ppm, versus 0.62 ± 0.16 for independent language-model predictions. The intervals overlap, and guidance remains limited by the process model’s learned support.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.