![[2504.12522] Evaluating the Diversity and Quality of LLM Generated Content](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2504.12522] Evaluating the Diversity and Quality of LLM Generated Content
Summary
This article evaluates the diversity and quality of content generated by large language models (LLMs), highlighting the trade-offs between diversity and quality in outputs.
Why It Matters
Understanding the balance between diversity and quality in LLM outputs is crucial for applications requiring varied and high-quality content, such as creative writing and data generation. This research provides a framework for measuring effective semantic diversity, which can guide future improvements in LLM design and deployment.
Key Takeaways
- Preference-tuning techniques can reduce output diversity in LLMs.
- Effective semantic diversity is a better measure of utility than simple diversity metrics.
- Larger models may show greater effective semantic diversity but smaller models are more efficient in generating unique content.
- Quality considerations are essential when evaluating LLM outputs.
- The findings have implications for applications needing both diversity and quality.
Computer Science > Computation and Language arXiv:2504.12522 (cs) [Submitted on 16 Apr 2025 (v1), last revised 26 Feb 2026 (this version, v2)] Title:Evaluating the Diversity and Quality of LLM Generated Content Authors:Alexander Shypula, Shuo Li, Botong Zhang, Vishakh Padmakumar, Kayo Yin, Osbert Bastani View a PDF of the paper titled Evaluating the Diversity and Quality of LLM Generated Content, by Alexander Shypula and 5 other authors View PDF HTML (experimental) Abstract:Recent work suggests that preference-tuning techniques -- such as Reinforcement Learning from Human Feedback (RLHF) methods like PPO and GRPO, as well as alternatives like DPO -- reduce diversity, creating a dilemma given that these models are widely deployed in applications requiring varied outputs. We argue that diversity without consideration of quality has limited practical value. To address this issue, we introduce a framework for measuring effective semantic diversity -- diversity among outputs that meet quality thresholds -- which better reflects the practical utility of large language models (LLMs). Using open-ended tasks that require no human intervention, we find counterintuitive results: when using diversity metrics that do not explicitly consider quality, preference-tuned models -- particularly those trained via RL -- often produce outputs with lower diversity; however, these same preference-tuned models generate greater effective semantic diversity than supervised fine-tuned (SFT) or base m...
