Where You Inject Diversity Matters: A Unified Framework for Diverse Generation
Quick Answer
This paper presents a unified framework for enhancing diversity in language model outputs by introducing varied sources during generation.
Quick Take
This paper presents a unified framework for enhancing diversity in language model outputs by introducing varied sources during generation. By employing specification-level injection across five open-ended tasks and four backbone models, the authors demonstrate improved output diversity while maintaining quality, emphasizing the importance of source design and transmission to the final output.
Key Points
- Introduces a framework for characterizing diversity in language model generation.
- Specification-level injection improves output diversity over traditional test-time methods.
- Analysis reveals that source diversity and transmission are critical for effective variation.
- Evaluated across five tasks and four backbone models, maintaining quality standards.
- Highlights the need for better source design to enhance generation systems.
Paper Resources
Article Content
From source RSS / original summaryarXiv:2606. 10302v1 Announce Type: new Abstract: Open-ended generation tasks often require a set of meaningfully different outputs, yet large language models often produce similar generations. Existing test-time diversity methods operate at different stages of generation with varying effectiveness, but it remains unclear what design choices lead to meaningful diversity in the output.
We introduce a framework that characterizes test-time diverse generation methods by the diversity source introduced during generation and provide a transmission score for measuring how effectively variation in the source reaches the final output. Guided by this framework, we propose fully automated specification-level generation methods that first generate diverse intermediate specifications and then condition on them to produce final responses.
Across five open-ended tasks and four backbone models, specification-level injection improves output diversity over test-time baselines while maintaining comparable quality. Our analysis shows that successful diversity injection depends on both the diversity of the sources and their transmission to the output, highlighting source design and source-to-output realization as two key levers for building more diverse generation systems.
Reader Mode unavailable (could not extract clean content).
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CL
See more →Time to REFLECT: Can We Trust LLM Judges for Evidence-based Research Agents?
The REFLECT benchmark reveals that current LLM judges are unreliable, achieving below 55% accuracy in evaluating reasoning and evidence use, highlighting the need for improved evaluation methods for deep research agents.