Even (very) noisy LLM evaluators are useful for improving AI agents(tensorzero.com)26 points by GabrielBianconi 5 months ago | 7 comments
Scheming a mise-en-abîme in BQN(panadestein.github.io)38 points by todsacerdoti 1 year ago | 0 comments
LLMs as the new high level language(federicopereiro.com)168 points by swah 8 months ago | 324 comments