Quantization, distillation, pruning and serving work aimed at the same accuracy for less memory, latency and money.
arXiv:2608. 05069v1 Announce Type: cross Abstract: Video Anomaly Detection (VAD) is inherently challenging due to the scarcity of anomalies and the large visual variability in surveillance footage, including changes in lighting, viewpoint, and human appearance.
arXiv:2608. 04458v1 Announce Type: new Abstract: Agentic AI is emerging in datacenters, but its architectural implications remain unexplored.
arXiv:2608. 04942v1 Announce Type: cross Abstract: CheMLFlow is an open-source platform for building and executing end-to-end, high-throughput, and agentic workflows for scientific and technological applications.
arXiv:2608. 05127v1 Announce Type: cross Abstract: Achieving local differential privacy in distributed optimization while maintaining low communication cost remains challenging.
arXiv:2608. 04788v1 Announce Type: cross Abstract: Large language model agents are commonly trained through reinforcement learning with sparse trajectory-level rewards, which offer limited guidance on how strongly individual tokens should be updated.
arXiv:2502. 15952v3 Announce Type: replace Abstract: Recent works exploring the training dynamics of homogeneous neural network weights under gradient flow with small initialization have established that in the early stages of training, the weights remain small and near the origin, but converge in direction.
arXiv:2608. 04593v1 Announce Type: cross Abstract: Echo State Networks (ESNs) offer an efficient framework for temporal prediction, but their randomly initialized reservoirs are often over-parameterized and dynamically redundant.
arXiv:2608. 04057v1 Announce Type: cross Abstract: Top-$k$ selection determines which components of a sparse model remain active.
arXiv:2608. 04472v1 Announce Type: cross Abstract: The development of foundation models (FMs) is crucial for advancing endoscopic image analysis.
arXiv:2608. 04048v1 Announce Type: cross Abstract: Serving large language models (LLMs) under diverse deployment constraints requires flexible trade-offs between accuracy, memory footprint, and throughput.
arXiv:2608. 04405v1 Announce Type: cross Abstract: Long-context large language models (LLMs) are increasingly deployed in real-world applications, yet self-attention remains a major efficiency bottleneck -- especially during decoding -- due to the necessity of repeatedly processing ever-growing key-value (KV) caches.
arXiv:2608. 04496v1 Announce Type: cross Abstract: Visual inputs in vision-language models (VLMs) are often encoded into substantially longer token sequences than text, making visual tokens a major bottleneck for efficient inference.
arXiv:2608. 04655v1 Announce Type: cross Abstract: Curvilinear structure analysis is an important and fundamental task in multimedia.
arXiv:2608. 04460v1 Announce Type: cross Abstract: The quantitative analysis of 3D neuronal morphologies requires capturing both graph topology and spatial geometry.
arXiv:2608. 04408v1 Announce Type: cross Abstract: On-policy distillation (OPD) supervises student-visited trajectories, yet divergence-based rules cannot determine whether an erroneous prefix remains correctable.
arXiv:2608. 04419v1 Announce Type: cross Abstract: On-policy distillation (OPD) provides dense teacher supervision on student-generated trajectories, but standard reverse-KL training can assign insufficient probability to other plausible continuations.
arXiv:2608. 05104v1 Announce Type: new Abstract: Deep neural networks have shown impressive success in NLP tasks owing to their complex structure and huge number of edges.
arXiv:2608. 05111v1 Announce Type: new Abstract: In partially observable reinforcement learning, agents face a dual bottleneck: they must explore to encounter rewarding states and retain that experience in memory to optimize their policies.
arXiv:2608. 04562v1 Announce Type: new Abstract: Agent skills are increasingly optimized by automated feedback loops, producing long structured artifacts whose internal value remains unclear.
arXiv:2608. 04502v1 Announce Type: cross Abstract: Attention--Feed-Forward Network (FFN) Disaggregation (AFD) is emerging as a promising architecture for serving Mixture-of-Experts (MoE) language models.