Perplexity evaluation

Pairings in the atlas

Held-out token likelihoodstandardModel output evaluation llm-training-alignment

Rivals: other methods for the same problems

Where it sits

llm-training-alignment · Machine Learning & AI