Presentation Granularity
No mentions found
This entity hasn't been tracked yet, or Iris is still building its knowledge base.
Related Articles from SNS
Skill Availability and Presentation Granularity in Large-Language-Model Agents: A Controlled SkillsBench Study
Announce Type: new Abstract: Skill documents provide procedural knowledge to large-language-model agents at inference time. This article studies whether the presentation granularity of controlled skill knowledge changes downstream task success. The experiment uses a pinned SkillsBench version, a 30-task domain-balanced subset validated by official oracle runs, two reasoning-enabled model configurations, six skill conditions, and five trials per task-condition-model cell.
Injection-rate effects on failure in a fluid-saturated granular fault gouge
arXiv:2606.08766v1 Announce Type: new Abstract: Fluid injection into the Earth's subsurface, performed for energy extraction, waste disposal, and resource development, is known to reactivate gouge-filled faults and induce seismicity, a key hazard in modern geotechnical operations. Nevertheless, the role of injection rate in controlling fault-gouge failure remains poorly understood. Here we present both an analytical theory and coupled fluid--granular (discrete element) numerical simulations...
I designed Microsoft's $5B EA channel architecture in 2001. The 2026 transition is missing what made it work
The Register's reporting on the Microsoft Enterprise Agreement (EA) commission collapse has been the most data-rigorous coverage in the trade press. And the trajectory it documented ($2.5 billion in LSP commissions in 2023, $1.67 billion in 2024, $583 million in 2025, zero in 2026) is the financial signature of a structural transition I have seen before, because I designed the original architecture that is now being retired. Between 1998 and 2001, I was the sole designer of the Enterprise...
GoodVibe: Security-by-Vibe for LLM-Based Code Generation
arXiv:2602.10778v2 Announce Type: replace Abstract: Large language models (LLMs) are increasingly used for code generation in fast, informal development workflows, often referred to as vibe coding, where speed and convenience are prioritized, and security requirements are rarely made explicit. In this setting, models frequently produce functionally correct but insecure code, creating a growing security risk. Existing approaches to improving code security rely on full-parameter fine-tuning or...
Activation Concentration: Characterizing Column-Level Output Sparsity Across Diffusion Model Architectures
Announce Type: replace Abstract: Recent diffusion accelerators exploit activation sparsity by skipping near-zero GELU outputs, reporting 52--85% element-level sparsity. However, systolic-array hardware processes activations at column granularity, where a single non-zero element forces the entire column to be computed. We present the first systematic column-level sparsity characterization across seven diffusion workloads spanning three workload groups and four modalities.
Generating and Refining Dynamic Evaluation Rubrics for LLM-as-a-Judge
Announce Type: new Abstract: LLM-as-a-Judge is a scalable alternative to human evaluation, yet existing rubric-based methods rely on human-annotated data such as reference answers or expert-crafted rubrics. We propose to automatically generate fine-grained evaluation rubrics without any human annotation. Our training-free method generates rubrics at dataset-specific and instance-specific granularities, achieving performance competitive with existing methods across four benchmarks.
Watch Duty Is Adding Flood Alerts to Its Wildfire App
Watch Duty, the wildfire alert app, is introducing flood alerts to its popular disaster-awareness service. This is the second disaster type to be broadly included, after wildfires; it’s available as a free update. If you have the app, allow it to track your location, and happen to be near a flood zone, Watch Duty will send you a push notification with more information about the flood.
AGENTSERVESIM: A Hardware-aware Simulator for Multi-Turn LLM Agent Serving
Announce Type: new Abstract: Multi-turn LLM agents interleave model calls with external tool invocations, shifting serving from stateless request processing to stateful program execution. Serving these workloads requires scheduling, KV-cache management, and routing policies that use program-level context, including turn dependencies, tool-induced gaps, and reusable KV state. Evaluating such policies directly on real systems is costly, since each design point may require dedicated accelerator...
Experiments in Agentic AI for Science
arXiv:2605.26305v2 Announce Type: replace Abstract: This paper details two novel frameworks for developing autonomous, agentic AI in scientific workflows. Both systems leverage a hybrid Local Body, Remote Brain architecture via Google Colab, utilizing Python-based local orchestrators to invoke large language model (LLM) cloud backends. The first agent, DeepTS/DeepCollector, automates the large-scale curation, extraction, and deduplication of time-series datasets.
Deep learning four decades of human migration
Abstract Human migration is a fundamental driver of global demographic change, shaping population structure, labour markets and social policy across countries1,2,3. Although long-term migration patterns are often linked to economic development4, they can shift rapidly in response to shocks such as conflict, environmental crises and political change5. Despite its importance, migration remains difficult to measure consistently: existing data are sparse, concentrated in high-income settings and...