Hallucination

Evaluation methodology: perplexity, pass@k, BLEU/ROUGE, embeddings, AI-as-a-judge, comparative evaluation (Elo/Bradley-Terry, LMSYS Arena); factual consistency, safety, instruction-following; model selection workflow; API vs self-hosting; evaluation pipeline design; benchmark contamination

Sampling strategies (temperature, top-k, top-p); test-time compute; structured outputs; hallucination (self-delusion + knowledge-mismatch hypotheses); inconsistency mitigation