This study uses CRP and CRAFT to reveal concept-level biases and confusion trends in MS-COCO multi-label classification, aiding interpretability.

Haadia Amjad, Ronald Tetzlaff

ConceptFormer enhances visual document retrieval by learning adaptive latent concepts, achieving a 16.7% NDCG@10 improvement.

Chunyi Peng, Zhipeng Xu, Yukun Yan et al.

SubZero+ enhances zeroth-order optimization stability with large learning rates, outperforming baselines on SuperGLUE.

Ziming Yu, Shuyao Xiao, Xingyu Zhao et al.

ADMITOR enhances optimization modeling precision with label-free certification, achieving candidate precision of 0.927.

Junbo Jacob Lian, Huiling Chen, Hanzhang Qin et al.

Proposes synchrony-aware sparse attention for audio-visual generation, achieving near 2× speedup while maintaining quality.

Shengchuan Gao, Teng Hu, Bohao Feng et al.

Proposes associative and declarative tasks to probe model year encoding; finds mechanisms differ, with limited updateability.

Suze van Adrichem, Aditi Bhaskar, Diyi Yang et al.

Q-First protocol enables parallel attention decoding using only the query, achieving a relative error of 3.2×10^-3 without new operators or hardware.

WenJie Fan

Hierarchical byte-level model with multi-byte prediction (MBP) using variable-length windows and LCA masks achieves optimal speed-performance trade-off.

Abraham Toluwase Owodunni, Chibuzor Okocha, Christan Grant et al.

StructFlow introduces spatially structured noise in flow models, enhancing local control and structure preservation in image generation.

Arman Zarei, Mahdi M. Kalayeh

This paper uncovers the 'curse of ambiguity' in large language models, analyzing its roots in model capacity, embedding size, and learning difficulty, highlighting practical limitations.

Nicolas Zucchet, Hyun Dong Lee, Scott Linderman

Using bee swarm optimization (BSO) for feature selection, achieving 0.997 detection accuracy with only 8 features out of 16 in large-scale WSNs.

Seungwoo Han, Sawako Kitagata, Ingon Chanpornpakdi et al.

Proposes an explicit circuit construction method achieving O(d log d) time for block encoding matrix polynomials.

Taehee Ko

Proposed a unified backbone-expert framework achieving 67.28% accuracy on RML2016.10b.

Zhixiang Deng, Houbiao Li, Zongyong Cui

Proposes low-rank (Rank4) latent carriers enabling autonomous counterfactual rollouts in 192D recurrent world models, validated on a two-object collision environment.

Yang Liu, Yuming Chen

This study compares 192 artificial and natural languages, revealing that language models develop left-branching bias artificially but favor right-branching SVO in natural languages, driven by data resources.

Varvara Arzt, Allan Hanbury, Terra Blevins

FoT uses early voting and rollout pruning based on hesitation signals to cut 28.8% attention FLOPs while maintaining SC@32 accuracy.

Chanhee Park, Sungbin Han, Jeongho Yoon et al.

MOSS-VL is an open-source vision-language model supporting real-time interaction via gated cross-attention, with 11.3B parameters, excelling in proactive response tasks.

Pengyu Wang, Chenkun Tan, Shaojun Zhou et al.

S2-MoE employs routing-aware adaptive speculative expansion and expert reuse gating to accelerate MoE inference on edge devices, achieving up to 5.3× speedup.

Haochen Huang, Shengxuan Qiu, Meng Li

Holistic evaluation of 11 memory substrates across models and tasks reveals task-dependent performance and scalability trade-offs, advocating multi-substrate routing.

Wei-Chieh Huang, Weizhi Zhang, Yuchen Wu et al.

Proposes Influence Calibration (ICSD) to address trust-utility mismatch, improving task success rates by 3-5% on benchmarks.

Qizhen Lan, Xi Xiao, Xiangchen Guan et al.