Skip to content
archive

huggingface-daily-papers-2026-05-15

AI · · 3 min read

HuggingFace Daily Papers — Top Picks · May 15, 2026

Ingested: 2026-05-15T09:08:00Z Source: https://huggingface.co/papers Sorted by community upvotes (top-upvoted first)

Executive Summary

  • Total papers ingested: 42 scraped, 24 top picks written
  • Top category: Agent memory & evaluation (4 papers in top 12)
  • Notable orgs: NVIDIA, Tsinghua, HKUST, Meta, Microsoft Research, InternLM
  • Key trend 1: Agentic RL with self-distillation (Self-Distilled Agentic RL, 49 🌟)
  • Key trend 2: LLM self-evolution & reward models (Darwin Family, EvolveMem, RewardHarness)
  • Key trend 3: Multimodal agent memory & evaluation (MemEye, MemLens, STALE, PREPING)

Top Papers by Upvotes

1. Achieving Gold-Medal-Level Olympiad Reasoning via Simple and Unified Scaling

  • Upvotes: 104
  • Authors: Yizhuo Li, Shunkai Zhang, Haoran Zhang, Runzhe Zhan, Yafu Li + 23 others
  • Submitted by: yaful
  • Topic: Reasoning
  • URL: https://huggingface.co/papers/2605.13301

2. Self-Distilled Agentic Reinforcement Learning

  • Upvotes: 49
  • Authors: Jinyang Wu, Zi-Han Wang, Zhuowen Han, Zhiyuan Yao, Zhengxi Lu + 7 others
  • Submitted by: taesiri
  • Topic: Agentic RL
  • URL: https://huggingface.co/papers/2605.15155

3. MemEye: A Visual-Centric Evaluation Framework for Multimodal Agent Memory

  • Upvotes: 46
  • Authors: Boxuan Zhang, Yihao Quan, Zeru Shi, Qingyue Jiao, Minghao Guo + 12 others
  • Submitted by: DarkBluee
  • Topic: Multimodal / Agent memory
  • URL: https://huggingface.co/papers/2605.15128

4. MemLens: Benchmarking Multimodal Long-Term Memory in Large Vision-Language Models

  • Upvotes: 43
  • Authors: NVIDIA
  • Submitted by: ZhaoweiWang (NVIDIA)
  • Topic: Vision-language / Memory
  • URL: https://huggingface.co/papers/2605.14906

5. SANA-WM: Efficient Minute-Scale World Modeling with Hybrid Linear Diffusion Transformer

  • Upvotes: 42
  • Authors: NVIDIA
  • Submitted by: HaoyiZhu (NVIDIA)
  • Topic: World models / Video
  • URL: https://huggingface.co/papers/2605.15178

6. Causal Forcing++: Scalable Few-Step Autoregressive Diffusion Distillation for Real-Time Interactive Video Generation

  • Upvotes: 35
  • Authors: Tsinghua Machine Learning Group
  • Submitted by: zhuhz22 (Tsinghua)
  • Topic: Video generation / Diffusion
  • URL: https://huggingface.co/papers/2605.15141

7. WildClawBench: A Benchmark for Real-World, Long-Horizon Agent Evaluation

  • Upvotes: 35
  • Authors: Intern Large Models
  • Topic: Agent benchmark / Long-horizon
  • URL: https://huggingface.co/papers/2605.10912

8. STALE: Can LLM Agents Know When Their Memories Are No Longer Valid?

  • Upvotes: 34
  • Authors: HKUST NLP Group
  • Topic: Agent memory
  • URL: https://huggingface.co/papers/2605.06527

9. Darwin Family: MRI-Trust-Weighted Evolutionary Merging for Training-Free Scaling of Language-Model Reasoning

  • Upvotes: 33
  • Authors: FINAL_Bench
  • Submitted by: FINAL_Bench
  • Topic: Model merging / Reasoning
  • URL: https://huggingface.co/papers/2605.14386

10. Warp-as-History: Generalizable Camera-Controlled Video Generation from One Training Video

  • Upvotes: 29
  • Authors: n/a
  • Topic: Video generation / Camera control
  • URL: https://huggingface.co/papers/2605.15182

11. Beyond Individual Intelligence: Surveying Collaboration, Failure Attribution, and Self-Evolution in LLM-based Multi-Agent Systems

  • Upvotes: 24
  • Authors: Xi'an Jiaotong University
  • Topic: Multi-agent systems / Survey
  • URL: https://huggingface.co/papers/2605.14892

12. PREPING: Building Agent Memory without Tasks

  • Upvotes: 22
  • Authors: KAIST AI
  • Topic: Agent memory
  • URL: https://huggingface.co/papers/2605.13880

13. EvolveMem:Self-Evolving Memory Architecture via AutoResearch for LLM Agents

  • Upvotes: 19
  • Authors: n/a
  • Topic: Agent memory / Self-improving
  • URL: https://huggingface.co/papers/2605.13941

14. ATLAS: Agentic or Latent Visual Reasoning? One Word is Enough for Both

  • Upvotes: 15
  • Authors: n/a
  • Topic: Visual reasoning / Agents
  • URL: https://huggingface.co/papers/2605.15198

15. RouteProfile: Elucidating the Design Space of LLM Profiles for Routing

  • Upvotes: 14
  • Authors: University of Illinois at Urbana-Champaign
  • Topic: LLM routing
  • URL: https://huggingface.co/papers/2605.00180

16. IntentVLA: Short-Horizon Intent Modeling for Aliased Robot Manipulation

  • Upvotes: 13
  • Authors: DeepCybo
  • Submitted by: DeepCybo
  • Topic: Robotics / VLA
  • URL: https://huggingface.co/papers/2605.14712

17. DiffusionOPD: A Unified Perspective of On-Policy Distillation in Diffusion Models

  • Upvotes: 12
  • Authors: n/a
  • Topic: Diffusion / Distillation
  • URL: https://huggingface.co/papers/2605.15055

18. VGGT-Edit: Feed-forward Native 3D Scene Editing with Residual Field Prediction

  • Upvotes: 12
  • Authors: Peking University
  • Topic: 3D generation
  • URL: https://huggingface.co/papers/2605.15186

19. FrontierSmith: Synthesizing Open-Ended Coding Problems at Scale

  • Upvotes: 12
  • Authors: n/a
  • Topic: CodeGen / Synthetic training data
  • URL: https://huggingface.co/papers/2605.14445

20. Realiz3D: 3D Generation Made Photorealistic via Domain-Aware Learning

  • Upvotes: 10
  • Authors: AI at Meta
  • Topic: 3D generation
  • URL: https://huggingface.co/papers/2605.13852

21. Orchard: An Open-Source Agentic Modeling Framework

  • Upvotes: 9
  • Authors: Microsoft Research
  • Topic: Agent framework / Open source
  • URL: https://huggingface.co/papers/2605.15040

22. RewardHarness: Self-Evolving Agentic Post-Training

  • Upvotes: 2
  • Authors: Natural and Artificial Intelligence Lab
  • Topic: RLHF / Agent post-training
  • URL: https://huggingface.co/papers/2605.08703

23. Nexus: An Agentic Framework for Time Series Forecasting

  • Upvotes: 1
  • Authors: n/a
  • Topic: Agents / Time series
  • URL: https://huggingface.co/papers/2605.14389

24. FutureSim: Replaying World Events to Evaluate Adaptive Agents

  • Upvotes: 1
  • Authors: Max Planck Institute for Intelligent Systems
  • Topic: Agent evaluation
  • URL: https://huggingface.co/papers/2605.15188