ProgressCompass: Embodied Progress Reward Models Are Lost Without the Right Context

Embodied progress reward models fail at long tasks due to missing context, but ProgressCompass supplies needed context to cut progress estimation errors by up to 82%.

Jianshu Zhang, Keyi Wu, Chengxuan Qian, Xiyuan Yang and 5 more

Published Sep 29, 2026 · 0 citations · ▲ 11 on Hugging Face

NitroBox: Lightning-Fast Sandbox for Large-Scale RL Training

Yuzhou Nie, Ruilin Zhou, Zhaorun Chen, Jingyang Zhang and 5 more

Atlanta Poster Session 6, Fri, Dec 11, 4:30 PM–7:30 PM, Hall C1 · Published 2026

Synaptic Strength Controls Trainability and Structural Stability in Rank-Deficient RNNs

Fatih Dinc, Edouard Ponnat, Henrik Weyer, Yanin Guerra and 1 more

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

$\texttt{bispectrum}$: Selective $G$-Bispectra Made Practical

Johan Mathe, Adele Lantow, Simon Mataigne, Nina Miolane

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

What if Agents Could Imagine? Reinforcing Open-Vocabulary HOI Comprehension through Generation

Zhenlong Yuan, Yue Wang, Jing Tang, Rui Chen and 8 more

Atlanta Poster Session 6, Fri, Dec 11, 4:30 PM–7:30 PM, Hall C1 · Published 2026

Position: AI-Agent Pricing Should Become More Outcome-Dependent: An Economic Perspective

Yuheng Bu, Yueyuan Ma

Atlanta Poster Session 2, Wed, Dec 9, 4:30 PM–7:30 PM, Hall C1 · Published 2026

Denoising Time Matters: Diverse Generation in Diffusion Language Models

jingxuan wu, Zhenglin Wan, Yuzhe YANG, Yiqiao Huang and 4 more

Atlanta Poster Session 3, Thu, Dec 10, 10:00 AM–1:00 PM, Hall C1 · Published 2026

Hessian-Dependent Sample Complexity in Zeroth-Order Stochastic Optimization: Suboptimality of Convex-Support Sampling and Optimal Sample Complexity

Mengtian Hong, Jason Lee, Qian Yu

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

Sampling Is Not Curiosity: Why LLM Agents Should Investigate

Alfonso Amayuelas, Piotr Piękos, Xin Wang, William Yang Wang and 2 more

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

What Should a Streaming Video Model Remember?

Haonan Ge, Yiwei Wang, Hang Wu, Yujun Cai

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

Dual Dimensionality for Local and Global Attention

Distance-Adaptive Representation uses high-dimensional local and low-dimensional distant keys and values to cut KV cache size while matching full-dimensional baseline performance.

Zhiyuan Wang, Xuan Luo, Sirui Zeng, Xifeng Yan

Atlanta Poster Session 6, Fri, Dec 11, 4:30 PM–7:30 PM, Hall C1 · Published 2026

Selective Disk Bispectrum: A Complete and Rotation Invariant Image Descriptor

Selective disk bispectrum is a rotation-invariant descriptor that preserves all image information except orientation with reduced complexity, validated for noise-robust classification and multi-reference alignment reconstruction.

Adele Lantow, Nina Miolane

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

Powering Up Zeroth-Order Training via Subspace Gradient Orthogonalization

Subspace gradient orthogonalization unifies low-rank projection with spectral optimization into ZO-Muon, cutting zeroth-order queries by 75% versus MeZO while boosting accuracy on LLM and vision fine-tuning.

Yicheng Lang, Changsheng Wang, Yihua Zhang, Mingyi Hong and 3 more

Atlanta Poster Session 3, Thu, Dec 10, 10:00 AM–1:00 PM, Hall C1 · Published 2026

Learning to Follow In-Context Watermark Instructions via Self-Distillation

ICWBench reveals current LLMs fail at in-context watermarking, and self-distillation with reinforcement learning raises watermark detectability near perfect while preserving quality.

Yepeng Liu, Tianyi Chen, Xuandong Zhao, Dawn Song and 1 more

Atlanta Poster Session 2, Wed, Dec 9, 4:30 PM–7:30 PM, Hall C1 · Published 2026

WorldMemArena: Evaluating Multimodal Agent Memory Through Action–World Interaction

WorldMemArena evaluates multimodal agent memory through an action-world loop, showing writing and storage improvements do not guarantee performance and harness-based memory remains costly and unreliable.

Chengzhi Liu, Yuzhe YANG, Sophia Xiao Pu, Yepeng Liu and 15 more

Sydney Poster Session 3, Wed, Dec 9, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026 · ▲ 13 on Hugging Face · Code ★ 29

Spectral Graph Sparsification Preserves Representation Geometry in Graph Neural Networks

Spectral sparsification bounds polynomial GNN filter and embedding perturbations by O(epsilon), preserving Gram matrices, distances, and training dynamics.

Sanjukta Krishnagopal

Atlanta Poster Session 5, Fri, Dec 11, 10:00 AM–1:00 PM, Hall C1 · Published 2026

TOPPO: Rethinking PPO for Multi-Task Reinforcement Learning with Critic Balancing

TOPPO balances critic gradients to fix PPO's multi-task ill-conditioning, outperforming SAC baselines with fewer parameters and steps.

Yuanpeng Li, Rui Miao, Gefei Lin, Annie Qu

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

Ares: Adaptive Reasoning Effort Selection for Efficient LLM Agents

Ares uses a lightweight router to select per-step reasoning effort for LLM agents, cutting reasoning tokens by up to 52.7% with minimal accuracy loss.

Jingbo Yang, Bairu Hou, Jiayun (Peter) Wang, Wei Wei and 2 more

Atlanta Poster Session 3, Thu, Dec 10, 10:00 AM–1:00 PM, Hall C1 · Published 2026 · ▲ 1 on Hugging Face

Revealing the Gap in Human and VLM Scene Perception through Counterfactual Semantic Saliency

Counterfactual Semantic Saliency reveals VLMs diverge from human scene perception via size, center, and saliency biases while underweighting people.

Ziqi Wen, Parsa Madinei, Miguel Eckstein

Atlanta Poster Session 2, Wed, Dec 9, 4:30 PM–7:30 PM, Hall C1 · Published 2026

HOPSE: Scalable Higher-Order Positional and Structural Encoder for Combinatorial Representations

HOPSE replaces higher-order message passing with Hasse-graph encodings to scale linearly on combinatorial domains while matching or exceeding state-of-the-art performance.

Guillermo Bernárdez, Marco Montagna, Louis Van Langendonck, Martin Carrasco and 6 more

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

Survive or Collapse: The Asymmetric Roles of Data Gating and Reward Grounding in Self-Play RL

Self-play RL stability depends mainly on a strict data gate over proposer tasks, not reward design; ground-truth access accelerates collapse via a self-consistent attractor.

Sophia Xiao Pu, Zhaotian Weng, Chengzhi Liu, Jayanth Srinivasa and 3 more

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

MOVEBENCH: A Benchmark for Global-Scale Wildlife Movement Forecasting

MoveBench introduces a 2.6M-location wildlife movement forecasting benchmark across 110 species and finds existing methods generalize poorly to unseen individuals and deep learning does not consistently beat simpler baselines.

Justin Kay, Shir Bar, Ellen O Aikens, Martin Becker and 27 more

Atlanta Poster Session 4, Thu, Dec 10, 4:30 PM–7:30 PM, Hall C1 · Published 2026