pith. sign in

Pplqa: An unsupervised information-theoretic quality metric for comparing generative large language models

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it

fields

cs.LG 1

years

2025 1

verdicts

UNVERDICTED 1

representative citing papers

Self-Aligned Reward: Towards Effective and Efficient Reasoners

cs.LG · 2025-09-05 · unverdicted · novelty 5.0

Self-aligned reward uses relative perplexity differences to encourage concise, query-specific reasoning in LLMs, yielding 4% accuracy gains and 30% lower inference cost when added to PPO or GRPO.

citing papers explorer

Showing 1 of 1 citing paper.

  • Self-Aligned Reward: Towards Effective and Efficient Reasoners cs.LG · 2025-09-05 · unverdicted · none · ref 11

    Self-aligned reward uses relative perplexity differences to encourage concise, query-specific reasoning in LLMs, yielding 4% accuracy gains and 30% lower inference cost when added to PPO or GRPO.