ET‑Prune is a training‑free framework that dynamically allocates visual token budgets in multimodal large language models based on question‑conditioned evidence. It protects text‑like spatial regions, converts evidence uncertainty into a token floor, and progressively prunes concentrated evidence while retaining diffuse or text‑dense tokens. In six backbone‑benchmark comparisons, ET‑Prune matches or outperforms other pruned methods while keeping roughly half the visual tokens, achieving notable gains on OCRBench‑v2 and MMBench v1.1.
By Zizhong Ding, Junxian Li, Kai Liu, Shaoqiu Zhang, Xiao Xiao, Linghe Kong, Yulun Zhang
The paper investigates test‑time adaptation (TTA) techniques for 3D point‑cloud registration in laparoscopic surgery, where synthetic training data must be adapted to noisy, sparse, and occluded real intraoperative reconstructions. It adapts three families of TTA methods—model, normalization, and input adaptation—to handle asymmetric shifts between preoperative meshes and intraoperative clouds, replacing classification‑based entropy objectives with correspondence‑based ones. Experiments on synthetic and real targets show that input adaptation consistently reduces registration error with low inference latency, making it the most promising approach for surgical applications.
By Nina Bodelot, Soufiane Belharbi, Eric Granger
Video-OPSD introduces a post‑training framework for Video Large Language Models that leverages privileged visual evidence to enhance on‑policy self‑distillation. The method constructs a self‑teacher conditioned only on annotated evidence frames, while the student processes the full video, allowing the teacher to provide more focused supervision. Additionally, an evidence‑guided token optimization scheme weights distillation based on each token’s reliance on privileged evidence, improving perceptually grounded reasoning. Experiments demonstrate consistent gains over standard OPSD and comparable performance to GRPO with less training time.
By Ziyue Wang, Shiqi Huang, Weiwen Xu, Bihan Wen, Xudong Jiang
RECAP-Forcing is a new method for long autoregressive video generation that addresses the memory challenge by organizing memory based on appearance novelty rather than recency. The approach retains key-value caches for newly appearing content—such as entering subjects, disoccluded regions, and new scenes—at the moment they first appear, ensuring consistent identities over time. It combines an attention sink for the initial scene with an optical-flow-based novelty bank for later frames, improving visual quality and semantic fidelity without adding learnable parameters.
By Haiyang Xu, Zheng Ding, Zhuowen Tu
Refine-POI introduces a reinforcement fine-tuned framework for next point-of-interest recommendation that tackles two key issues: topology-blind indexing of semantic IDs and the limitation of supervised fine-tuning to top‑1 predictions. It uses a hierarchical self‑organizing map to generate topology‑aware semantic IDs and a policy‑gradient approach to produce top‑k recommendation lists. Experiments on three real‑world datasets show that Refine‑POI outperforms state‑of‑the‑art baselines, combining LLM reasoning with accurate, explainable recommendations.
By Peibo Li, Shuang Ao, Hao Xue, Yang Song, Maarten de Rijke, Johan Barth\'elemy, Tomasz Bednarz, Flora D. Salim
The paper investigates how scaling the amount of model-generated, off‑task distillation data affects the recoverability of teacher‑induced traits in student models. In a controlled subliminal‑learning setup, teachers are prompted to express a target trait, producing restricted data such as number‑only completions. Students trained on larger independent datasets show a clearer manifestation of the teacher’s trait in a separate evaluation domain, with the effect being strongest when the trait is already favored or when alternative traits are present. The authors find this trend holds across model families, trait types, multi‑trait settings, and cross‑model transfer, and suggest that scaling should be coupled with trait‑aware curation and evaluation.
By Zhichen Dong, Zhixuan Liu, Yuyu Fan, Xiangtian Li, Shuyang Zhang, Chao Yang
The paper investigates low‑bit quantization for Multimodal Large Language Models (MLLMs), showing that MXFP8 retains near‑lossless performance while 4‑bit formats like MXFP4 and HiF4 cause significant degradation. It identifies activation quantization as the main source of this loss and introduces Residual Fallback Quantization (RFQ), a lightweight framework that adds a quantized residual pathway to improve activation fidelity without architectural changes. Experiments on Wan2.2 and Qwen3‑VL demonstrate that RFQ recovers much of the performance gap to BF16 baselines across generation and reasoning tasks.
By Tanzila Rahman, Mehran Taghian Jazi, Yunke Peng, Zhuang Ma, Anandharaju Durai Raju, Yao Wang, Xing Huang, Hei Yi Mak, Shadan Golestan, Hoang Le, Yonghan Dong, Wei Guo, Yaoyuan Wang
The paper introduces DMFL-SQ, a decentralized multi-task learning algorithm that integrates graph-based personalization, agnostic fairness, and compressed event-triggered communication. It provides convergence guarantees for non-convex objectives, achieving an ≠O(T^{-1/2}) stationarity rate despite sparse, quantized, and event-triggered communication, and offers PAC-Bayes generalization bounds for the fairness objective. Experiments on CIFAR-10 and the MUSMET EEG dataset show that DMFL-SQ reduces communication while preserving predictive performance and improving fairness across clients.
By Krishnendu S. Tharakan, Carlo Fischione
SPEAR (Symbolic Process Evaluation and Alignment Reward) is a training‑free, plug‑and‑play reward method for on‑policy distillation in reinforcement learning. It converts natural‑language reasoning traces into domain‑adaptive symbolic milestones and uses the longest common subsequence to align student exploration with teacher milestones, producing a dense, order‑aware reward that enforces logical consistency without an external neural verifier. Experiments on math, science, and commonsense reasoning tasks show that SPEAR effectively bridges the reasoning gap between student and teacher models through sequence‑level distillation with efficient dense process rewards.
By Zhuochun Li, Yuelyu Ji, Yiming Zeng, Daqing He
The paper presents a PyTorch-based framework for designing and optimizing binarized neural networks, incorporating freezing and pruning mechanisms. It introduces a novel pruning method that uses a global weighting scheme to assess parameter importance across abstraction levels, achieving a 70% pruning rate on VGG11 without sacrificing accuracy—outperforming existing binarized pruning results of 41%. The framework facilitates rapid, reproducible evaluation and prototyping of state‑of‑the‑art binarized network techniques.
By Roan Rubiales, Jean Pierre David
The paper introduces ZipCal, a model‑agnostic data curation method that selects calibration data for post‑training compression of large language models by maximizing lexical diversity using Zipfian power laws. ZipCal outperforms uniform random sampling on pruning benchmarks and matches a state‑of‑the‑art perplexity‑based approach while being roughly 240× faster due to its linear complexity. The authors provide code and experiments at their GitHub repository.
By Francesco Pio Monaco, Elia Cunegatti, Flavio Vella, Giovanni Iacca
The paper reports the development of 2‑bit microkernels for CPUs and mixed‑precision 2‑bit kernels for Intel Xe2 GPUs, achieving near‑roofline performance. Integrated into LLM inference pipelines, these kernels deliver up to 7× speedup over 16‑bit inference on CPUs and 6.7× on GPUs, surpassing the current state‑of‑the‑art bitnet.cpp runtime by 2.2×. The work demonstrates that ultra‑low‑bit LLM models can be deployed efficiently, offering significant gains in latency, memory, throughput, and energy consumption.
By Evangelos Georganas, Dhiraj Kalamkar, Alexander Heinecke, Pradeep Dubey
The paper proposes a multi‑modal AI framework for real‑time queue prediction, management, and resource optimisation in border control systems. It integrates heterogeneous data sources using LSTM networks for forecasting and applies Model Predictive Control and scheduling optimisation to generate actionable policies for officers. Evaluation on synthetic traffic data shows up to 35% reduction in prediction error, 30% lower average waiting time, and nearly 20% higher throughput compared to ARIMA and rule‑based methods.
By Varvara Mama, Eleni Veroni, Nikolaos Kapsalis, Christos D. Nikolopoulos, Anargyros T. Baklezos
MIMO: Multilingual Information Retrieval via Monolingual Objectives proposes a two‑stage framework that first aligns a student model to a stable English semantic space using knowledge distillation, then jointly optimizes distillation and cross‑lingual contrastive learning to improve retrieval discrimination while preserving alignment. The approach addresses language clustering and the trade‑off between cross‑lingual alignment and embedding uniformity, outperforming existing cross‑lingual training baselines on both multilingual and multi‑monolingual benchmarks. MIMO also remains competitive with larger off‑the‑shelf models and its alignment‑uniformity analysis clarifies the distinct roles of the two loss components.
whyItMatters":"The study demonstrates a practical method to enhance multilingual information retrieval performance by balancing alignment and uniformity, which is crucial for real‑world search environments where queries and documents span multiple languages."
By Youngjoon Jang, Seongtae Hong, Heuiseok Lim
The paper introduces Adaptive Visual Token Pruning (AVTP), a training‑free framework that dynamically selects pruning layers and ratios for large vision‑language models (LVLMs) when processing multiple image sequences. By analyzing visual attention distributions across different LVLM architectures, AVTP adapts token retention to image importance, enabling efficient inference without relying on attention‑based computations incompatible with FlashAttention. Experiments show significant speedups—up to 2× for Qwen3VL‑8B—while preserving or even improving accuracy on multi‑image benchmarks.
By Rongyang Zhang, Chengqiang Lu, Cong Li, Hongchao Gu, Tingjia Shen, Xuyang Zhi, Qimeng Wang, Yan Gao, Yi Wu, Yao Hu, Hao Wang, Enhong Chen
KISS-GS is a modular compression pipeline for 3D Gaussian Splatting (3DGS) scenes that separates compression from training. It first compacts a vanilla 3DGS scene by 15.7× using state‑of‑the‑art pruning, then encodes the result into the SOG‑XT image‑based format, achieving an additional 6.6× reduction. Optional encoding‑aware fine‑tuning can further cut the size by 2.2×, yielding total reductions of 85× to 319× on standard benchmarks while enabling web‑native decoding.
By Wieland Morgenstern, Friedrich Elias Branschke, Florian Fleischmann, Adrian Szatmari, Paul Schlack, Florian Barthel, Peter Eisert, Anna Hilsmann
Self-OPD introduces a teacher‑free on‑policy distillation framework for flow matching models, using the student’s own exploration to generate step‑wise supervision. At each timestep the deterministic next‑state prediction is branched into multiple stochastic SDE candidates, rolled out, and compared against a deterministic baseline to compute normalized advantages. The velocity field is then optimized with a pull‑push objective that attracts high‑advantage branches and repels low‑advantage ones, while multi‑objective alignment is achieved by fusing normalized scores at the reward level.
By Shiyi Zhang, Mushui Liu, Yunze Tong, Wanggui He, Siyu Zou, Jinlong Liu, Yunlong Yu, Jian Song, Hao Jiang, Pipei Huang, Bo Zheng
NestedKV is a training‑free key‑only KV cache compression technique for long‑context language models that uses global, block‑level, and sliding‑window key anchors to score tokens via multi‑time‑scale cosine anomaly. It combines these rankings with a head‑adaptive outer learner and surprise‑gated token routing, requiring no model modification or additional training. Experiments on Qwen3 and Llama‑3.2 across benchmarks such as RULER, LongBench, and MMLU‑Pro show that NestedKV outperforms existing methods when the retained cache is small, achieving up to 19‑point gains on RULER and LongBench at a 75% retention rate.
By Hong Chen, Xiang Liu, Yubo Gao, Yuxuan Fan, Bo Wang, Yuanlin Chu, Yuanguo Lin, Xuming Hu
TwinKV is a training‑free, attention‑free repair pass that identifies and swaps orphaned and redundant tokens in a KV cache, improving long‑context inference for small models. It works by detecting near‑duplicate keys and can be composed with existing eviction policies without altering their scoring rules. Experiments on Qwen3‑4B and Llama‑3.2‑1B across LongBench, LooGLE, RULER, and MMLU‑Pro show that TwinKV consistently improves performance for most configurations, especially at tighter compression ratios.
By Hong Chen, Yudong Zeng, Yongwei Huang, Zuhao Ouyang, Junyan Zhang, Xuming Hu
The paper introduces a decoupled, I/O-aware pipeline for extracting embeddings from large whole-slide images (WSIs). It splits the workflow into patch generation, parallel embedding inference, and sharded vector database ingestion, isolating data movement from compute to reduce I/O overhead. The resulting distributed vector database stores embeddings with rich metadata, enabling efficient filtering, retrieval, and downstream tasks such as classification and few-shot learning, especially in low-resource settings.
By Mayanka Chandrashekar, Xi Zhang, Ethan Seefried, Tirthankar Ghosal, John Gounley, Heidi Hanson