Reference-Free Rating of LLM Responses via Latent Information

Leander Girrbach, Chi-Ping Su, Tankred Saanum, richard socher, Eric Schulz, Zeynep Akata

Keywords: large language model, llm, llm-as-a-judge

Abstract: How reliable are single-response LLM-as-a-judge ratings without references, and can we obtain fine-grained, deterministic scores in this setting? We study the common practice of asking a judge model to assign Likert-scale scores to free-text responses and show two systematic issues: scores are unstable under sampling and poorly calibrated, leading to compression near the top of the scale and frequent ties. We then evaluate Latent Judges, which derive scalar ratings from internal model signals: (i) probability-weighted scores over integer ratings, (ii) verifier-style probabilities of yes, and (iii) linear probes trained on hidden activations at the rating position. Across a broad suite of pairwise and single-rating benchmarks, latent methods match or surpass standard prompting, with consistent gains on pairwise accuracy and listwise ranking relevant to Best-of-N selection. Probability-weighted scores achieve the strongest single-rating correlations, while probes recover useful signals when output logits are miscalibrated. These results indicate that latent information provides deterministic and more discriminative signals for reference-free evaluation, and can improve selection and training approaches like Best-of-N, multi-teacher distillation, and routing.

Primary Area: foundation or frontier models, including LLMs

Submission Number: 2969