July 2026
Calibrated confidence for zero-shot LLMs
How llmex turns a causal LM's logits into probabilities you can actually filter and trust — with visual explainers of logit scoring, collision fallback, recognition prompts, and temperature scaling.
Read note →