Good Papers: read the good ones

LucidNFT improves flow-based real-world super-resolution via LR-anchored multi-reward preference optimization that reduces hallucinations while preserving perceptual quality.

Song Fei, Tian Ye, Sixiang Chen, Zhaohu Xing and 2 more

Sydney Poster Session 2, Tue, Dec 8, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

11/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

83%Must read

?Must readVote to see the score

NeurIPS 2026The Hong Kong University of Scie The Hong Kong University of Scie Hong Kong University of Science Zhejiang Alignment & preference optimization

Strengthening LLMs for Tabular Prediction with Structural Priors

PRPO incorporates column-permutation invariance into LLM post-training via label-preserving permutations and two-level advantage estimation, enabling an 8B model to match specialized tabular baselines and outperform 685B reasoning LLMs by up to 53%.

Pengxiang Cai, Zihao Gao, Wanchen Lian, Guocong Li and 1 more

Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

13/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

72%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026Harbin Institute of Technology Tsinghua Harvard Alibaba Group Hong Kong University of Science RL for LLMs

LoopRPT: Reinforcement Pre-Training for Looped Language Models

LoopRPT applies reinforcement pre-training to looped language models by assigning rewards to latent reasoning steps, improving per-step quality and accuracy-computation trade-offs.

Guo Tang, Shixin Jiang, Heng Chang, Zihan Zhang and 6 more

Paris Poster Session 4, Thu, Dec 10, 5:30 PM–7:30 PM, Paris Poster Hall · Published 2026 · ▲ 16 on Hugging Face

– ReadersNo votes yet

8/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 8 of 20 reviewers recommend it

lenient 4/5

medium 4/10

strict 0/5

78%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026The Hong Kong University of Scie The Hong Kong University of Scie LLM reasoning & chain of thought

Ask, Answer, and Detect: Role-Playing LLMs for Personality Detection with Question-Conditioned Mixture-of-Experts

ROME uses role-playing LLMs to generate questionnaire answers from posts, then routes them via mixture-of-experts to improve personality detection and mitigate label scarcity.

Yifan Lyu, Liang Zhang

Sydney Poster Session 2, Tue, Dec 8, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

11/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 11 of 20 reviewers recommend it

lenient 5/5

medium 6/10

strict 0/5

78%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026The Hong Kong University of Scie Southwest Jiaotong The Hong Kong University of Scie Physics-informed ML & PDEs

Reformulating Neural Operators in $d+1$ Dimensions for Embedding Evolution

Reformulating neural operators in d+1 dimensions via auxiliary embedding evolution achieves lowest relative L2 error across benchmarks without brute-force scaling.

Haoze Song, Zhihao Li, Xiaobo Zhang, Zecheng Gan and 2 more

Sydney Poster Session 3, Wed, Dec 9, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

11/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

80%Must read

?Must readVote to see the score

NeurIPS 2026The Hong Kong University of Scie The Hong Kong University of Scie Guangzhou Medical critical care medicine, Guangzho HKUST-GZ Uncertainty & calibration

Robust Conditional Conformal Prediction via Branched Normalizing Flow

Branched Normalizing Flow bounds conditional invalidity via Wasserstein distance and improves robust conditional coverage under distribution shift.

Rui Xu, Xingyuan Chen, Wenxing Huang, Minxuan Huang and 3 more

Sydney Poster Session 4, Wed, Dec 9, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

12/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

80%Must read

?Must readVote to see the score

NeurIPS 2026Harbin Institute of Technology The Hong Kong University of Scie Department Of Computer Science a Tsinghua Kuaishou- 快手科技 RL for LLMs

SpatialFlow-GRPO: Where Spatial Credit Drives Image Editing

SpatialFlow-GRPO introduces region-aware rewards to replace whole-image feedback, improving fine-grained image editing via spatially aligned policy updates.

Yankai Yang, Yancheng Long, Wei Chen, Xingyu Lu and 6 more

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

12/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

83%Must read

?Must readVote to see the score

NeurIPS 2026HKUST Department of Computer Science a The Hong Kong University of Scie Hong Kong University of Science Test-time compute

Attention-Based Sampler for Diffusion Language Models

Attention-based sampling orders diffusion language model tokens by attention-matrix column sums to maximize likelihood and improve generation quality with greater parallelism.

Yuyan Zhou, Kai Syun Hou, Weiyu Chen, James Kwok

Sydney Poster Session 3, Wed, Dec 9, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

13/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 13 of 20 reviewers recommend it

lenient 4/5

medium 8/10

strict 1/5

83%Must read

?Must readVote to see the score

NeurIPS 2026SUN YAT-SEN UNIVERSITY South China University of Techno Harbin Institute of Technology Institute of automation, Chinese Shanghai Jiaotong Vision-language-action models

RePO-VLA: Recovery-Driven Policy Optimization for Vision-Language-Action Models

RePO-VLA improves vision-language-action robustness by assigning roles to success, recovery, and failure trajectories, raising adversarial success from 20% to 75%.

Weijia Liufu, Xiaoyu Guo, Ruiyi Chen, Jingzhi Liu and 15 more

Sydney Poster Session 2, Tue, Dec 8, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

13/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

88%Must read

?Must readVote to see the score

NeurIPS 2026Shanghai Jiaotong Tencent Huawei Technologies Ltd.The Hong Kong University of Scie U Edinburgh, University of Edinb Hallucination & factuality

BALTO: Balanced Token-Level Policy Optimization for Hallucination Mitigation

BALTO applies balanced token-level credit assignment to mitigate LLM hallucinations by redistributing probability from unsupported to faithful content, outperforming response-level methods on faithfulness benchmarks.

Ning Li, Zixuan Guo, Yan Xu, Wenbo Fei and 6 more

Sydney Poster Session 6, Thu, Dec 10, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

15/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 15 of 20 reviewers recommend it

lenient 5/5

medium 9/10

strict 1/5

74%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026OralThe Hong Kong University of Scie Eindhoven University of Technolo Hong Kong University of Science 3D vision & reconstruction

TransmissiveGS: Residual-Guided Disentangled Gaussian Splatting for Transmissive Scene Reconstruction and Rendering

TransmissiveGS disentangles reflective and transmissive components via dual Gaussian representations and residual-guided multi-view separation for transmissive scene reconstruction.

Zhenyu Liang, Xiao Zhang, Tianchao Li, Jack C Cheng and 1 more

Atlanta Poster Session 1, Wed, Dec 9, 10:00 AM–1:00 PM, Hall C1 · Published 2026

– ReadersNo votes yet

9/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 9 of 20 reviewers recommend it

lenient 4/5

medium 5/10

strict 0/5

78%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026The Hong Kong University of Scie U Science and Technology of Chin Hong Kong University of Science The Chinese University of Hong K Vision-language models

Direct Product Flow Matching: Decoupling Radial and Angular Dynamics for Few-Shot Adaptation

Direct product flow matching decouples radial and angular dynamics via constant-speed geodesic transport and hidden-state conditioning for state-of-the-art few-shot vision-language adaptation.

Hongxu Chen, Yanghao Wang, Bowei Zhu, Hongxiang Li and 5 more

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

11/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

83%Must read

?Must readVote to see the score

NeurIPS 2026The Hong Kong University of Scie The Hong Kong University of Scie Bournemouth Shanghai Jiaotong Beijing University of Posts and Vision-language-action models

Beyond Appearance Shifts: Task-Semantic Action Calibration for VLA Models

BAS-VLA calibrates frozen VLA actions via breaking-centered calibration and selective preservation gating to suppress stale-task drift and separate semantics. It achieves 98% clean success, 0% under target swaps, and 70% under style shifts versus 42%.

Shuaijun Liu, Feiyang You, Chengyu Wu, Shuyang Hao and 5 more

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

13/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 13 of 20 reviewers recommend it

lenient 4/5

medium 9/10

strict 0/5

70%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026The Hong Kong University of Scie Institute of AI for Industries, Vision-language models

VVTRec: Radio Interferometric Reconstruction through Visual and Textual Modality Enrichment

VVTRec uses visual and textual visibility enrichment with vision-language models to reduce artifacts and improve radio interferometric image reconstruction.

Kai Cheng, Ruoqi Wang, Qiong Luo

Sydney Poster Session 6, Thu, Dec 10, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

4/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

83%Must read

?Must readVote to see the score

NeurIPS 2026HKUST Tencent IEG The Hong Kong University of Scie Digital Content Technology Cente Zhejiang 3D vision & reconstruction

Real2Sim in HOI: Toward Physically Plausible HOI Reconstruction from Monocular Videos

HA-HOI reconstructs physically plausible 4D human-object interactions from monocular video by anchoring object motion to human action and refining via physics simulation. It improves alignment, contact consistency, and simulation readiness over prior monocular reconstruction methods.

Yubo ZHAO, Yujin Chai, Yunao Dong, Chengfeng Zhao and 3 more

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

13/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 13 of 20 reviewers recommend it

lenient 5/5

medium 8/10

strict 0/5

72%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026Jilin Immunocan Biotech Tsinghua Hong Kong University of Science Nanyang Technological Proteins

ProteinOPD: Towards Effective and Efficient Preference Alignment for Protein Design

ProteinOPD balances multi-objective protein preferences via on-policy distillation from preference-specific teachers, preserving designability with 8x training speedup over RL methods.

Yulin Zhang, He CAO, Zihao Jiang, Chenyi Zi and 5 more

Sydney Poster Session 4, Wed, Dec 9, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

8/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 8 of 20 reviewers recommend it

lenient 4/5

medium 4/10

strict 0/5

71%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026The Hong Kong University of Scie The Hong Kong University of Scie U Macau Hong Kong University of Science Locomotion

PDF-HR: Pose Distance Fields for Humanoid Robots

PDF-HR learns continuous, differentiable pose distance fields from retargeted humanoid motions to score pose plausibility, substantially improving tracking, mimicry, and retargeting baselines.

Yi Gu, Yukang Gao, Yangchen Zhou, Xingyu Chen and 6 more

Sydney Poster Session 2, Tue, Dec 8, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

7/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

76%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026Southwest The Hong Kong University of Scie Tsinghua ByteDance Inc.Guangdong OPPO Mobile Telecommun Web & GUI agents

Search More, Think Less: Rethinking Long-Horizon Agentic Search for Efficiency and Generalization

SMTL replaces sequential reasoning with parallel evidence acquisition for efficient long-horizon agentic search, achieving state-of-the-art results on multiple benchmarks with far fewer reasoning steps.

Chengjun Yu, Shu XU, Jiaqi Wu, Qianben Chen and 20 more

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026 · ▲ 23 on Hugging Face · Code ★ 3

– ReadersNo votes yet

10/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 10 of 20 reviewers recommend it

lenient 4/5

medium 6/10

strict 0/5

83%Must read

?Must readVote to see the score

NeurIPS 2026Hong Kong University of Science The Hong Kong University of Scie Beijing Institute of Mathematica Molecules & drug discovery

Probe Before You Edit: Probing-Guided Molecular Optimization for LLM Agents in Structure-Based Drug Design

PROBE uses edit-response probing to build pocket-specific site maps and EditManuals that guide multi-agent optimization of both affinity and druggability in structure-based drug design, achieving state-of-the-art on CrossDocked2020.

Zaifei Yang, Weiyu Chen, Yaqing Wang, James Kwok

Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

13/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

83%Must read

?Must readVote to see the score

NeurIPS 2026GRADIENT DATA Soochow Independent Researcher Amazon Microsoft Agent benchmarks & environments

TwinRouterBench: Fast Static and Live Dynamic Evaluation for Realistic Agentic LLM Routing

TwinRouterBench introduces static and live dynamic tracks to benchmark LLM routing at agent step-level using deterministic scoring and live execution on SWE-bench.

Pei Yang, Wanyi Chen, Tongyun Yang, Pengbin Feng and 11 more

Sydney Poster Session 4, Wed, Dec 9, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026 · ▲ 5 on Hugging Face

– ReadersNo votes yet

13/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

91%Must read

?Must readVote to see the score

NeurIPS 2026The Hong Kong University of Scie Alibaba Group Renmin University of China The Chinese University of Hong K Peking LLM agents & planning

Towards On-Policy Data Evolution for Visual-Native Multimodal Deep Search Agents

A visual-native harness with an image bank and on-policy data evolution improves multimodal deep search agents, raising Qwen3-VL-8B to 39.0% average and surpassing Gemini-2.5 Pro.

Shijue Huang, Hangyu Guo, Guanting Dong, Chenxin Li and 7 more

Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026 · ▲ 21 on Hugging Face · Code ★ 30

– ReadersNo votes yet

17/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

83%Must read

?Must readVote to see the score

NeurIPS 2026The Hong Kong University of Scie HKUST Huawei Technologies Ltd.Long-context modeling

Efficient Scaling of LLM Training with Flexible Context Parallelism

Flexible Context Parallelism adaptively reconfigures communication groups to eliminate load imbalance and redundant communication, achieving up to 1.46x training speedup over Megatron-LM and DeepSpeed.

Yifan Niu, Han Xiao, Dongyi Liu, Wei zhou and 1 more

Sydney Poster Session 6, Thu, Dec 10, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

13/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

74%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026HKUST Hong Kong University of Science VNET Group The Hong Kong University of Scie Zhejiang Diffusion models

AHPA: Adaptive Hierarchical Prior Alignment for Diffusion Transformers

AHPA adaptively selects hierarchical VAE feature priors via a timestep-conditioned router to match diffusion transformer alignment granularity to denoising needs, improving convergence without inference overhead.

Ruibin Min, Yexin Liu, Aimin PAN, Changsheng Lu and 4 more

Sydney Poster Session 3, Wed, Dec 9, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

9/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

89%Must read

?Must readVote to see the score

NeurIPS 2026The Hong Kong University of Scie The Hong Kong University of Scie Hong Kong University of Science Xidian Baidu Video-language models

MedHorizon: Towards Long-context Medical Video Understanding in the Wild

MedHorizon benchmarks long medical video understanding via sparse evidence retrieval and multi-hop reasoning, with top models reaching only 41.1% accuracy.

Bodong Du, Bowen Liu, Yang YU, Xinpeng Ding and 7 more

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

16/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

80%Must read

?Must readVote to see the score

NeurIPS 2026Hohai Nanjing University of Science an The University of Manchester Guangdong OPPO Mobile Telecommun The Hong Kong University of Scie LLM agents & planning

A Self-Evolving Framework for Efficient Terminal Agents via Observational Context Compression

TACO is a training-free framework that learns adaptive compression rules from terminal agent trajectories to filter noisy observations, improving accuracy by 1-4% and reducing token usage across benchmarks.

JinCheng Ren, Siwei Wu, Yizhi Li, Zhu and 9 more

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026 · ▲ 24 on Hugging Face · Code ★ 47

– ReadersNo votes yet

12/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

88%Must read

?Must readVote to see the score

NeurIPS 2026The Hong Kong University of Scie Hong Kong University of Science Bandits & online learning

Direction-Aware Offline-to-Online Learning in Linear Contextual Bandits

A directional bias certificate enables Ellipsoidal-MINUCB to safely exploit offline data in linear contextual bandits, reducing regret when low-bias directions align with historical coverage.

Zean Han, Ruihan Lin, Zezhen Ding, Jiheng Zhang

Paris Poster Session 6, Fri, Dec 11, 2:30 PM–4:30 PM, Paris Poster Hall · Published 2026

– ReadersNo votes yet

15/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

70%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026Tsinghua The Chinese University of Hong K The Hong Kong University of Scie Applied Research Center, Tencent Wuhan Controllable generation

AnchorWorld: Embodied Egocentric World Simulation with View-based Evolution Customization

AnchorWorld improves egocentric world simulation via full-body interaction supervision and anchor-view customization with consistent spatio-temporal dynamics.

Yu Li, Menghan Xia, Gongye Liu, Xintao Wang and 7 more

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026 · ▲ 30 on Hugging Face

– ReadersNo votes yet

5/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 5 of 20 reviewers recommend it

lenient 3/5

medium 2/10

strict 0/5

83%Must read

?Must readVote to see the score

NeurIPS 2026The Hong Kong University of Scie The Hong Kong University of Scie University College London, Unive Microsoft Research Nanyang Technological RL for LLMs

CodeScaler: Scaling Code LLM Training and Test-Time Inference via Reward Models

CodeScaler uses a reward model to scale code LLM training and inference without test cases, improving benchmarks by up to 14.64 points and cutting latency tenfold.

Xiao Zhu, Xinyu Zhou, Boyu Zhu, Hanxu Hu and 4 more

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026 · ▲ 24 on Hugging Face · Code ★ 50

– ReadersNo votes yet

13/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

88%Must read

?Must readVote to see the score

NeurIPS 2026SpotlightThe Hong Kong Polytechnic The Hong Kong University of Scie U California, Berkeley Hong Kong Polytechnic ByteDance Inc.LLM agents & planning

VisInteract: Towards Dynamic Interactive Text-to-Visualization under Imperfect Queries

VisInteract introduces interactive text-to-visualization with imperfect queries via VisInteract-Bench and Vis-MCTS, boosting success by over 13% versus interactive baselines.

Wenxin XU, Jinwei Lu, Hwanhee Kim, Chen J Zhang and 3 more

Sydney Poster Session 2, Tue, Dec 8, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

15/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

74%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026The Hong Kong University of Scie ByteDance Inc.Applied Research Center, Tencent Kuaishou Technology Alibaba group Unified multimodal models

Edit-R2: Context-Aware Reinforcement Learning for Multi-Turn Image Editing

Edit-R2 uses reinforcement learning to reconstruct session intent and jointly optimize reasoning and generation for multi-turn image editing. It improves instruction following and consistency over accumulated constraints on the MICE-Bench benchmark.

Yuxiao YE, Haoran He, Fangyuan Kong, Xintao Wang and 3 more

Sydney Poster Session 6, Thu, Dec 10, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

9/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 9 of 20 reviewers recommend it

lenient 5/5

medium 4/10

strict 0/5

88%Must read

?Must readVote to see the score

NeurIPS 2026Beijing Academy of Artificial In Shanghai Jiaotong U Electronic Science and Technol Hainan The Hong Kong University of Scie 3D vision & reconstruction

Focusable Monocular Depth Estimation

FocusDepth uses spatially-aligned multi-scale prompt fusion to boost target-region depth accuracy and sharp boundaries while preserving global geometry, outperforming global baselines on FDE-Bench.

Yuxin Du, Tao Lin, Zile Zhong, Runting Li and 6 more

Sydney Poster Session 3, Wed, Dec 9, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

15/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

70%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026The Hong Kong University of Scie ByteDance Inc.Bytedance Hong Kong University of Science Shanghai Jiaotong Video generation

HOMIE: Human-object Centric Video Personalization via Multimodal Intelligent Enhancement

HOMIE unifies inter- and intra-subject video personalization via multimodal guidance and reference embeddings, achieving state-of-the-art human-object interaction fidelity.

Yiyang Cai, Nan Chen, Rongchang Xie, Junwen Pan and 7 more

Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026 · ▲ 54 on Hugging Face · Code ★ 177

– ReadersNo votes yet

4/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 4 of 20 reviewers recommend it

lenient 1/5

medium 3/10

strict 0/5

91%Must read

?Must readVote to see the score

NeurIPS 2026Hong Kong University of Science The Hong Kong University of Scie WeBank AI Lab Graphs & LLMs

Beyond the Golden Teacher: Enhancing Graph Learning through LLM-GNN Co-teaching

LLM-GNN Co-Teaching replaces golden-teacher design with bidirectional pseudo-label exchange and trajectory-based preference optimization, boosting few-shot graph accuracy by up to 7.86%.

Zhuoyi Peng, Hanlin Gu, Lixin Fan, Yi Yang

Paris Poster Session 3, Thu, Dec 10, 12:30 PM–2:30 PM, Paris Poster Hall · Published 2026

– ReadersNo votes yet

18/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 18 of 20 reviewers recommend it

lenient 5/5

medium 10/10

strict 3/5

80%Must read

?Must readVote to see the score

NeurIPS 2026The Hong Kong University of Scie The Hong Kong University of Scie Hong Kong University of Science Beijing University of Technology Shenzhen MSU-BIT Vision-language-action models

HumanoidArena: Benchmarking Egocentric Hierarchical Whole-body Learning

HumanoidArena benchmarks egocentric hierarchical whole-body learning via seven leg-critical tasks, finding policies solve diverse interactions but cross-tracker transfer remains fragile.

Taowen Wang, Zikang Xie, Bin Yang, Yunheng Wang and 12 more

Sydney Poster Session 4, Wed, Dec 9, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026 · ▲ 18 on Hugging Face

– ReadersNo votes yet

12/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

76%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026Sun Yat-sen university Shanghai AI Lab The Hong Kong University of Scie SUN YAT-SEN UNIVERSITY City University of Hong Kong Tool use & function calling

Train the Agent, Not the Expert: Learning to Harness Heterogeneous Experts for Multi-Turn Visual Reasoning

VisHarness trains a visual agent to orchestrate heterogeneous experts for multi-turn reasoning, achieving strong results on segmentation, detection, and counting tasks.

Yaowu Fan, Tao Han, Dazhao Du, Jinhua Ma and 1 more

Sydney Poster Session 6, Thu, Dec 10, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

10/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

74%Highly rated

?Highly ratedVote to see the score

NeurIPS 2026U Washington Microsoft Research UT Austin Duke The Hong Kong University of Scie Multi-agent LLM systems

Population-Aligned Persona Generation for LLM-based Social Simulation

A framework generates population-aligned personas from social media via quality filtering, importance sampling, and task-specific adaptation, reducing bias in LLM social simulations.

Zhengyu Hu, Jianxun Lian, Zheyuan Xiao, Max Xiong and 9 more

Sydney Poster Session 3, Wed, Dec 9, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

– ReadersNo votes yet

9/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

80%Must read

?Must readVote to see the score

NeurIPS 2026Hong Kong University of Science metax MetaX Alibaba Group MetaX Integrated Circuits Co., L Efficient inference & serving

SpecBlock: Block-Iterative Speculative Decoding with Dynamic Tree Drafting

SpecBlock drafts block-iterative trees with hidden-state path dependence and adaptive branching, cutting EAGLE-3 drafting cost by about half while boosting speedup 8, 19%.

Weijie Shi, Qiang Xu, fan deng, Yaguang Wu and 8 more

Sydney Poster Session 6, Thu, Dec 10, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026 · ▲ 3 on Hugging Face

– ReadersNo votes yet

12/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 12 of 20 reviewers recommend it

lenient 4/5

medium 7/10

strict 1/5

Show 20 more papers