GenAI Daily for Practitioners

Executive Summary • Here are the concise, non-sensationalist bullets for enterprise practitioners: • CTkvr: KV Cache Retrieval for Long-Context LLMs via Centroid then Token Indexing: Improves retrieval efficiency by 2.4x and reduces memory usage by 1.5x for long-context LLMs. (arxiv.org/abs/2512.15550v1) • Attention in Motion: Secure Platooning via Transformer-based Misbehavior Detection: Achieves 93.2% accuracy in detecting misbehavior in platooning scenarios using transformer-based models. (arxiv.org/abs/2512.15503v1) • MCP-SafetyBench: A Benchmark for Safety Evaluation of Large Language Models with Real-World MCP Servers: Provides a benchmark for evaluating safety in large language models, with 14 tasks covering various domains. (arxiv.org/abs/2512.15163v1) • Image Complexity-Aware Adaptive Retrieval for Efficient Vision-Language Models: Improves retrieval efficiency by 30% and accuracy by 12% for complex images using adaptive retrieval. (arxiv.org/abs/2512.15372v1) • DiscoX: Benchmarking Discourse-Level Translation task in Expert Domains:

Research

• CTkvr: KV Cache Retrieval for Long-Context LLMs via Centroid then Token Indexing \ Large language models (LLMs) are increasingly applied in long-context scenarios such as multi-turn conversations. However, long contexts pose significant challenges for inference efficiency, including high memory overhead from Key-Value (K… \ Source • arXiv cs.CL • 16:56
• Attention in Motion: Secure Platooning via Transformer-based Misbehavior Detection \ Vehicular platooning promises transformative improvements in transportation efficiency and safety through the coordination of multi-vehicle formations enabled by Vehicle-to-Everything (V2X) communication. However, the distributed nature of… \ Source • arXiv cs.LG • 15:45
• MCP-SafetyBench: A Benchmark for Safety Evaluation of Large Language Models with Real-World MCP Servers \ Large language models (LLMs) are evolving into agentic systems that reason, plan, and operate external tools. The Model Context Protocol (MCP) is a key enabler of this transition, offering a standardized interface for connecting LLMs with … \ Source • arXiv cs.CL • 09:00
• Image Complexity-Aware Adaptive Retrieval for Efficient Vision-Language Models \ Vision transformers in vision-language models apply uniform computational effort across all images, expending 175.33 GFLOPs (ViT-L/14) whether analysing a straightforward product photograph or a complex street scene. We propose ICAR (Image… \ Source • arXiv cs.LG • 13:19
• DiscoX: Benchmarking Discourse-Level Translation task in Expert Domains \ The evaluation of discourse-level translation in expert domains remains inadequate, despite its centrality to knowledge dissemination and cross-lingual scholarly communication. While these translations demand discourse-level coherence and … \ Source • arXiv cs.CL • 17:24
• From Signal to Turn: Interactional Friction in Modular Speech-to-Speech Pipelines \ While voice-based AI systems have achieved remarkable generative capabilities, their interactions often feel conversationally broken. This paper examines the interactional friction that emerges in modular Speech-to-Speech Retrieval-Augment… \ Source • arXiv cs.CL • 13:31
• EvoLattice: Persistent Internal-Population Evolution through Multi-Alternative Quality-Diversity Graph Representations for LLM-Guided Program Discovery \ Large language models (LLMs) are increasingly used to evolve programs and multi-agent systems, yet most existing approaches rely on overwrite-based mutations that maintain only a single candidate at a time. Such methods discard useful vari… \ Source • arXiv cs.CL • 13:18
• Adversarial versification in portuguese as a jailbreak operator in LLMs \ Recent evidence shows that the versification of prompts constitutes a highly effective adversarial mechanism against aligned LLMs. The study 'Adversarial poetry as a universal single-turn jailbreak mechanism in large language models' demon… \ Source • arXiv cs.CL • 12:55
• mimic-video: Video-Action Models for Generalizable Robot Control Beyond VLAs \ Prevailing Vision-Language-Action Models (VLAs) for robotic manipulation are built upon vision-language backbones pretrained on large-scale, but disconnected static web data. As a result, despite improved semantic generalization, the polic… \ Source • arXiv cs.LG • 19:47
• From Trace to Line: LLM Agent for Real-World OSS Vulnerability Localization \ Large language models show promise for vulnerability discovery, yet prevailing methods inspect code in isolation, struggle with long contexts, and focus on coarse function or file level detections which offers limited actionable guidance t… \ Source • arXiv cs.LG • 19:10
• Photonics-Enhanced Graph Convolutional Networks \ Photonics can offer a hardware-native route for machine learning (ML). However, efficient deployment of photonics-enhanced ML requires hybrid workflows that integrate optical processing with conventional CPU/GPU based neural network archit… \ Source • arXiv cs.LG • 16:55
• HI-SQL: Optimizing Text-to-SQL Systems through Dynamic Hint Integration \ Text-to-SQL generation bridges the gap between natural language and databases, enabling users to query data without requiring SQL expertise. While large language models (LLMs) have significantly advanced the field, challenges remain in han… \ Source • arXiv cs.LG • 15:25

Big Tech

No items today.

Regulation & Standards

No items today.

Enterprise Practice

No items today.

Open-Source Tooling

No items today.

— Personal views, not IBM. No tracking. Curated automatically; links under 24h old.