Diversity Collapse

Follow this topic Ask about this topic

Momentum

1 paper in the last four weeks, against 1 the four weeks before. 0.0% of all new papers.

Jul 6Week of Sep 21

Latest papers 21

All topics

DiVid: Diagnosing Dimension-Specific Diversity Collapse in Video Generation Models
Oct 1, 2026Huanran Hu, Zihui Ren, Dingyi Yang +4Diversity Collapse Video Generation
Renmin University of China · Alibaba Group · Tianjin University
Low-Confidence Remasking Traps Flexibility: Realizing Arbitrary-Order Potential for Diverse Rollouts in Diffusion LLMs
Sep 28, 2026Moongyu Jeon, Dongjae Jeon, Bumjun Kim +2Masked Diffusion Language Models Diffusion Language Models
Yonsei University · KRAFTON AI · Kookmin University
Aligned but Flattened: Analyzing the Trade-off between Cultural Alignment and Diversity in LLMs
Sep 1, 2026Jingshen Zhang, Shaoyang Xu, Wenxuan ZhangLarge Language Model Alignment Cultural Awareness
School of Computer Science and Technology, Tianjin University · 2iNLP Lab, Singapore University of Technology and Design
Influence-Directed Distillation: Solving the Diversity Bottleneck in Sampled-Token On-Policy Distillation
Aug 30, 2026Run Yang, Runpeng Dai, Jie Sun +5Teacher Diversity Collapse
University of North Carolina at Chapel Hill · University of Science and Technology of China · Shanghai University of Finance and Economics
Breaking Diversity Collapse in Spiking Pseudo-Ensembles for Efficient OOD Detection in Remote Sensing
Aug 2, 2026Srinivas Anumasa, Rushi Shah, Qiran Zou +1Spiking Neural Networks Out-Of-Distribution Detection
National University of Singapore
MAPLE: Efficient and Diverse Multi-Alpha Generation for Portfolio Construction
Jul 27, 2026Yu-Chen Den, Kuan-Yu Chen, Kendro Vincent +1Portfolio Construction Diversity Collapse
SinoPac Holdings · Taipei, Taiwan · National Chengchi University
Controllable Diversity in Normalization-Based Implicit Ensembles via Softmax-Temperature Modulation
Jul 26, 2026Mihai Suteu, Ovidiu SerbanEnsemble Gumbel-Softmax Relaxation
Data Science Institute, Imperial College London
When Reasoning Narrows the Move: Diversity Collapse in LLM Game Play
Jul 21, 2026Junyi Sha, Renfei Tan, David Simchi-LeviDiversity Collapse Diversity
Institute for Data, Systems, and Society MIT Cambridge, MA 02139 · Department of Civil and Environmental Engineering MIT Cambridge, MA 02139
Improving Sample Diversity in Autoregressive Text-to-Image Generation via Cluster Truncation
Jul 12, 2026Trang Nguyen, Shuang Wu, Runyan Tan +1Autoregressive Image Generation Text-To-Image
University of Massachusetts Amherst · Thoughtworks
Fidelity-Diversity Metrics for Text
Jul 6, 2026Amanda Wang, Tudor Manole, Florentina Bunea +1Diversity Data Quality
Department of Computer Science, Cornell University · Statistics and Data Science Center, Massachusetts Institute of Technology · Department of Statistics, Cornell University
Don't Settle at the Mode! Mitigating Diversity Collapse in Pretrained Flow Models via Feature Self-Guidance
Jun 25, 2026Pradhaan S Bhat, Rishubh Parihar, Abhijnya Bhat +1Generative Flow Networks Diversity Collapse
Indian Institute of Science · Stanford University
DivRL: Disentangled Self-Similarity Rewards for Diverse Subject-Driven Generation
Jun 22, 2026Qian Wang, Zhenyu Li, Abdelrahman Eldesokey +1Multi-Reference Image Generation Diversity Collapse
KAUST, Saudi Arabia
Understanding Diversity Collapse in RLVR via the Lens of Overtraining
Jun 13, 2026Suqin Yuan, Jinkun Chen, Jiyang Zheng +6Reinforcement Learning With Verifiable Reward Verifiable Rewards
Sydney AI Centre, The University of Sydney · Southeast University · Microsoft +3
Rapid co-design of Buoyancy-assisted robots for Challenging Locomotion using Gaussian Evolutionary Specialists
Jun 5, 2026Ankit Sinha, Nitish Sontakke, Dennis Hong +2Agile Locomotion Co-Design
School of Interactive Computing, Georgia Institute of Technology, TSRB 85 5th St NW, Atlanta, GA - 30332, USA · Dept. of Mechanical and Aerospace Engineering, University of California, Los Angeles, 420 Westwood Plaza, CA 90095, USA · Robotic Systems Lab, ETH Zurich, Leonhardstrasse 21, 8092 Zurich, Switzerland
Sampling More, Getting Less: Calibration is the Diversity Bottleneck in LLMs
May 11, 2026Amin Banayeeanzade, Qingchuan Yang, Dhruv Tarsadiya +6Diversity Collapse Diversity
♠University of Southern California · ♣Capital One
Escaping the Diversity Trap in Robotic Manipulation via Anchor-Centric Adaptation
May 8, 2026Yanzhe Chen, Kevin Yuchen Ma, Qi Lv +4Robotic Manipulation Diversity Collapse
National University of Singapore · Institude for Information Research, A*STAR · Harbin Institute of Technology (Shenzhen).
Ex Ante Evaluation of AI-Induced Idea Diversity Collapse
May 7, 2026Nafis Saami Azad, Raiyan Abdul BatenCreativity Diversity Collapse
Bellini College of Artificial Intelligence, Cybersecurity, and Computing University of South Florida Tampa, FL–33620, USA
Diversity Collapse in Multi-Agent LLM Systems: Structural Coupling and Collective Failure in Open-Ended Idea Generation
Apr 20, 2026Nuo Chen, Yicheng Tong, Yuzhe Yang +5Diversity Collapse Multi-Agent Large Language Model Systems
National University of Singapore · The Chinese University of Hong Kong, Shenzhen
Where does output diversity collapse in post-training?
Apr 17, 2026Constantinos Karouzos, Xingwei Tan, Nikolaos AletrasDiversity Collapse Diversity
School of Computer Science University of Sheffield, UK
Zoom-IQA: Image Quality Assessment with Reliable Region-Aware Reasoning
Jan 6, 2026Guoqiang Liang, Jianyi Wang, Zhonghua Wu +2Image Quality Assessment Recent Vision-Language Models
S-Lab, Nanyang Technological University, Singapore · SenseTime Research
AI's Blind Spots: Geographic Knowledge and Diversity Deficit in Generated Urban Scenario
Jun 20, 2025Ciro Beneduce, Massimiliano Luca, Bruno LepriText-To-Image Diffusion Models Diffusion Models
* Bruno Kessler Foundation · **University of Trento

Oct 1, 2026, cs.CVJ/K move · Enter open · S save

DiVid: Diagnosing Dimension-Specific Diversity Collapse in Video Generation Models

Huanran Hu, Zihui Ren, Dingyi Yang, Zhinan Song+3

Renmin University of China · Alibaba Group · Tianjin University

Diversity Collapse Video Generation Diversity Creativity

Open paper AskPDF arXiv

Despite remarkable progress, video generation models often produce highly similar outputs when repeatedly sampled from the same prompt, limiting their usefulness for creative exploration. Existing diversity evaluations primarily rely on global scalar metrics, which obscure where diversity collapses in the spatiotemporal space of videos. We introduce DiVid, a dimension-level diagnostic framework that decomposes video generation diversity into six interpretable dimensions: Semantic, Style, Subject, Scene, Motion, and Camera. Each dimension is measured through a reproducible computer-vision pipeline and analyzed alongside quality and instruction faithfulness to examine potential trade-offs. Systematic evaluation of representative video generation models reveals that diversity is highly dimension-specific: models with strong global diversity scores still collapse on specific factors, particularly Motion and Camera. These rankings persist after filtering unfaithful generations, indicating genuine capability differences rather than off-prompt outputs. Beyond measurement, controlled prompt interventions identify two fundamental bottlenecks: default mode convergence, where models fall back to dominant patterns under open-ended prompts; and realization gaps, where models fail to faithfully realize diverse, explicitly requested alternatives, particularly for temporal factors. The larger faithfulness losses for temporal factors highlight the difficulty of controlling motion and camera variation through text alone. DiVid thus shifts the study of diversity from measuring whether it exists to diagnosing where and why it collapses, and provides actionable directions for dimension-aware training objectives and control signals. The framework will be released to facilitate future research on diverse and controllable video generation.

Momentum

Related topics

Latest papers 21