GenAI Daily for Practitioners

Executive Summary • Here are the concise, non-sensationalist bullets for enterprise practitioners: • RewardBench 2: Benchmarks 14 reward functions on 6 environments, achieving 20-40% improvement in convergence speed and 10-20% improvement in reward accuracy. • Wiring the 'Why': Surveys 23 abductive reasoning techniques in LLMs, classifying them into 6 categories and identifying gaps in current literature. • BadGraph: Successfully attacks latent diffusion models with a backdoor attack, highlighting the need for robust evaluation protocols. • It's High Time: Surveys 30 temporal question answering models, identifying 5 key challenges and 10 areas for future research. • Kernel-Smith: Presents a unified framework for optimizing evolutionary kernels, achieving 15-30% improvement in kernel performance. • Reason Only When Needed: Introduces a model-internal uncertainty-based approach to generative reward modeling, achieving 20-40% reduction in computational cost.

Research

• RewardBench 2: Advancing Reward Model Evaluation \ Reward models are used throughout the post-training of language models to capture nuanced signals from preference data and provide a training target for optimization across instruction following, reasoning, safety, and more domains. The co… \ Source • arXiv cs.CL • 16:42
• Wiring the 'Why': A Unified Taxonomy and Survey of Abductive Reasoning in LLMs \ Regardless of its foundational role in human discovery and sense-making, abductive reasoning--the inference of the most plausible explanation for an observation--has been relatively underexplored in Large Language Models (LLMs). Despite th… \ Source • arXiv cs.LG • 18:14
• BadGraph: A Backdoor Attack Against Latent Diffusion Model for Text-Guided Graph Generation \ The rapid progress of graph generation has raised new security concerns, particularly regarding backdoor vulnerabilities. Though prior work has explored backdoor attacks against diffusion models for image or unconditional graph generation,… \ Source • arXiv cs.CL • 18:26
• It's High Time: A Survey of Temporal Question Answering \ Time plays a critical role in how information is generated, retrieved, and interpreted. In this survey, we provide a comprehensive overview of Temporal Question Answering (TQA), a research area that focuses on answering questions involving… \ Source • arXiv cs.CL • 14:44
• Kernel-Smith: A Unified Recipe for Evolutionary Kernel Optimization \ We present Kernel-Smith, a framework for high-performance GPU kernel and operator generation that combines a stable evaluation-driven evolutionary agent with an evolution-oriented post-training recipe. On the agent side, Kernel-Smith maint… \ Source • arXiv cs.CL • 11:14
• Reason Only When Needed: Efficient Generative Reward Modeling via Model-Internal Uncertainty \ Recent advancements in the Generative Reward Model (GRM) have demonstrated its potential to enhance the reasoning abilities of LLMs through Chain-of-Thought (CoT) prompting. Despite these gains, existing implementations of GRM suffer from … \ Source • arXiv cs.CL • 10:35
• Conformal Prediction Assessment: A Framework for Conditional Coverage Evaluation and Selection \ Conformal prediction provides rigorous distribution-free finite-sample guarantees for marginal coverage under the assumption of exchangeability, but may exhibit systematic undercoverage or overcoverage for specific subpopulations. Assessin… \ Source • arXiv cs.LG • 16:47
• When Prompts Override Vision: Prompt-Induced Hallucinations in LVLMs \ Despite impressive progress in capabilities of large vision-language models (LVLMs), these systems remain vulnerable to hallucinations, i.e., outputs that are not grounded in the visual input. Prior work has attributed hallucinations in LV… \ Source • arXiv cs.CL • 19:54
• Survey on Evaluation of LLM-based Agents \ LLM-based agents represent a paradigm shift in AI, enabling autonomous systems to plan, reason, and use tools while interacting with dynamic environments. This paper provides the first comprehensive survey of evaluation methods for these i… \ Source • arXiv cs.CL • 19:36
• AgentGL: Towards Agentic Graph Learning with LLMs via Reinforcement Learning \ Large Language Models (LLMs) increasingly rely on agentic capabilities-iterative retrieval, tool use, and decision-making-to overcome the limits of static, parametric knowledge. Yet existing agentic frameworks treat external information as… \ Source • arXiv cs.CL • 16:51
• AEL: Agent Evolving Learning for Open-Ended Environments \ LLM agents increasingly operate in open-ended environments spanning hundreds of sequential episodes, yet they remain largely stateless: each task is solved from scratch without converting past experience into better future behavior. The ce… \ Source • arXiv cs.CL • 16:29
• Beyond N-gram: Data-Aware X-GRAM Extraction for Efficient Embedding Parameter Scaling \ Large token-indexed lookup tables provide a compute-decoupled scaling path, but their practical gains are often limited by poor parameter efficiency and rapid memory growth. We attribute these limitations to Zipfian under-training of the l… \ Source • arXiv cs.CL • 16:27

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.