Adam ZlamalGitHub
← Projects
Classification
Live STAGES 5METRICS 8

SurveyMaster

Classifying papers by multi-LLM consensus voting.

Open the appView on GitHub

What it does

Several LLMs vote, weighted by confidence, to place papers in hierarchical taxonomies more reliably than any single model. A refinement loop fixes dead and overloaded leaves. The whole pipeline also runs in the browser: open a finished run, or bring an OpenRouter key and start your own.

Highlights