GenAI Daily for Practitioners

Executive Summary • Here are the concise bullets for each news item: • FPGA Co-Design for Efficient N:M Sparse and Quantized Model Inference: • + Co-design of FPGA and AI model reduces inference latency by 2.5x and energy consumption by 4.5x compared to software-only implementation. • + Achieves 10x speedup and 5x energy reduction for a specific N:M sparse and quantized model. • Encyclo-K: Evaluating LLMs with Dynamically Composed Knowledge Statements: • + Introduces Encyclo-K, a benchmark to evaluate large language models (LLMs) using dynamically composed knowledge statements. • + Shows that LLMs can improve performance on Encyclo-K tasks by up to 10% compared to traditional benchmarks.

Research

• FPGA Co-Design for Efficient N:M Sparse and Quantized Model Inference \ Large language models (LLMs) have demonstrated remarkable performance across a wide range of language processing tasks. However, this success comes at the cost of substantial computation and memory requirements, which significantly impedes… \ Source • arXiv cs.LG • 09:27
• Encyclo-K: Evaluating LLMs with Dynamically Composed Knowledge Statements \ Benchmarks play a crucial role in tracking the rapid advancement of large language models (LLMs) and identifying their capability boundaries. However, existing benchmarks predominantly curate questions at the question level, suffering from… \ Source • arXiv cs.CL • 14:55
• PrivacyBench: A Conversational Benchmark for Evaluating Privacy in Personalized AI \ Personalized AI agents rely on access to a user's digital footprint, which often includes sensitive data from private emails, chats and purchase histories. Yet this access creates a fundamental societal and privacy risk: systems lacking so… \ Source • arXiv cs.CL • 14:16
• Compute-Accuracy Pareto Frontiers for Open-Source Reasoning Large Language Models \ Large Language Models (LLMs) are demonstrating rapid improvements on complex reasoning benchmarks, particularly when allowed to utilize intermediate reasoning steps before converging on a final solution. However, current literature often o… \ Source • arXiv cs.CL • 11:51
• Characterization of Transfer Using Multi-task Learning Curves \ Transfer effects manifest themselves both during training using a fixed data set and in inductive inference using accumulating data. We hypothesize that perturbing the data set by including more samples, instead of perturbing the model by … \ Source • arXiv cs.LG • 14:55
• Many Minds from One Model: Bayesian Transformers for Population Intelligence \ Despite their scale and success, modern transformers are almost universally trained as single-minded systems: optimization produces one deterministic set of parameters, representing a single functional hypothesis about the data. Motivated … \ Source • arXiv cs.CL • 19:56
• An Analysis of Hyper-Parameter Optimization Methods for Retrieval Augmented Generation \ Optimizing Retrieval-Augmented Generation (RAG) configurations for specific tasks is a complex and resource-intensive challenge. Motivated by this challenge, frameworks for RAG hyper-parameter optimization (HPO) have recently emerged, yet … \ Source • arXiv cs.CL • 10:32
• BIOME-Bench: A Benchmark for Biomolecular Interaction Inference and Multi-Omics Pathway Mechanism Elucidation from Scientific Literature \ Multi-omics studies often rely on pathway enrichment to interpret heterogeneous molecular changes, but pathway enrichment (PE)-based workflows inherit structural limitations of pathway resources, including curation lag, functional redundan… \ Source • arXiv cs.CL • 10:01
• When F1 Fails: Granularity-Aware Evaluation for Dialogue Topic Segmentation \ Dialogue topic segmentation supports summarization, retrieval, memory management, and conversational continuity. Despite decades of work, evaluation practice remains dominated by strict boundary matching and F1-based metrics. Modern large … \ Source • arXiv cs.CL • 09:52
• MUSIC: MUlti-Step Instruction Contrast for Multi-Turn Reward Models \ Evaluating the quality of multi-turn conversations is crucial for developing capable Large Language Models (LLMs), yet remains a significant challenge, often requiring costly human evaluation. Multi-turn reward models (RMs) offer a scalabl… \ Source • arXiv cs.CL • 08:54
• Scaling Open-Ended Reasoning to Predict the Future \ High-stakes decision making involves reasoning under uncertainty about the future. In this work, we train language models to make predictions on open-ended forecasting questions. To scale up training data, we synthesize novel forecasting q… \ Source • arXiv cs.CL • 19:59
• AdaGReS:Adaptive Greedy Context Selection via Redundancy-Aware Scoring for Token-Budgeted RAG \ Retrieval-augmented generation (RAG) is highly sensitive to the quality of selected context, yet standard top-k retrieval often returns redundant or near-duplicate chunks that waste token budget and degrade downstream generation. We presen… \ Source • arXiv cs.CL • 19:48

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.