Infrastructure & Hardware B
Showing 31–60 of 218
-
Zero-Mem: Zero-Token Memory Operations for LLM Agents
-
AWS looks to build 800,000 sq ft data center campus at GWU site in Ashburn, VirginiaAWS eyes 800,000 sq ft data center campus in Ashburn, VirginiaAWS is looking to build an 800,000 sq ft data center campus on a former GWU site in Ashburn, Virginia, DatacenterDynamics reported. A county supervisor vowed to use 'every tool available' to fight the development, underscoring growing local pushback against data center expansion.
-
DualDiT: A Conditional Dual-Output Diffusion Transformer for Joint OCT Image and Segmentation Mask Generation
-
CyrusOne files to develop three buildings for $1.5bn data center campus in Fairfield, TexasCyrusOne files for $1.5bn, three-building data center campus in TexasCyrusOne filed to develop three buildings for a $1.5bn data center campus in Fairfield, Texas. Construction has already begun on two of the buildings, with the third slated to start next month, reflecting the ongoing surge in large-scale data center investment across the US.
-
Veolia to operate 350MW gas-powered microgrid for Ohio data center campusVeolia to run 350MW gas-powered microgrid for Ohio data centerVeolia will operate a 350MW microgrid for an Ohio data center campus, according to DatacenterDynamics. The system will be anchored by natural gas generation and supplemented with a battery energy storage (BESS) unit to provide reliable power for the large facility.
-
Legacy Investing buys printing press in Minneapolis, Minnesota, for data center developmentLegacy Investing buys Minneapolis printing press for data centerLegacy Investing acquired a printing press site in Minneapolis, Minnesota, for data center development, DatacenterDynamics reported. The firm is planning a 20MW facility despite an ongoing moratorium on new data centers in the city.
-
OsteoCAD: A Human-in-the-Loop Cloud-Edge Framework for Bone Tumor Segmentation
-
CalibratedRubric: Task-Adaptive Rubric Banks for Open-Ended LLM Evaluation
-
Small Is Enough: Per-User Style Rewriting of AI-Edited Text via LoRA Adapters
-
MOSAIC: Masked Outsourcing of Secure AI Computations
-
Knowing When to Quit: Diagnosing and Training LLMs to Abort Futile Reasoning
-
キオクシアQ1決算、純利益は前年比4500％増 AIデータセンター向け需要がけん引Kioxia Q1 net profit up 4,506% YoY, driven by AI data-center demandSemiconductor maker Kioxia Holdings said net profit for its fiscal Q1 (April–June 2026, IFRS) reached ¥842.165bn, up 4,506% year on year. The company attributed the jump to surging memory demand from AI data centers.
-
Faster but Different: Diagnosing and Controlling Content Drift in Accelerated Multimodal Diffusion Language Models
-
Adjudicated Captioning: Multi-Agent Alignment Scoring and Consensus-Distilled Beam Arbitration for Strict Zero-Shot Image Captioning
-
PARALLEL: A Prefrontal-Aligned Reinforcement inspired Approach for Language-Model Learning under Explicit Limits
-
Token-Level Diagnosis of Sycophancy in LLMs with Attribution-Guided Steering
-
Cohere signs EU Code of Practice on Transparency of AI-Generated ContentCohere signs EU Code of Practice on AI content transparencyCohere said it signed the EU Code of Practice on Transparency of AI-Generated Content, joining other companies committing to clearer labeling and provenance for AI outputs. The move signals alignment with Europe's emerging AI governance framework.
-
Advancing the price-performance frontier with GPT‑5.6OpenAI slashes GPT-5.6 prices: Luna down 80%, Terra down 20%OpenAI announced steep price cuts for GPT-5.6, with Luna dropping 80% and Terra 20%. The company credits GPT-5.6 Sol for enabling the reduction by optimizing load balancing and even the model's forward pass, the computation that turns inputs into next-token predictions.
-
TORUS: A Test of Rendering-Understanding Self-Coherence for Unified Audio Models
-
llm 0.32rc2llm 0.32rc2 switches its default model to GPT-5.6 LunaSimon Willison released llm 0.32rc2, fixing a dependency issue and changing the default model for users who have not set one from GPT-4o mini to the newer, more capable GPT-5.6 Luna. Luna is slightly more expensive but a notable upgrade.
-
Run High-Performance Core Math at Scale with NVIDIA nvmath-pythonNVIDIA introduces nvmath-python for high-performance math at scaleNVIDIA presented nvmath-python, a library bridging the Python scientific community with CUDA-X math libraries. It lets developers run high-performance core math at scale from Python, making GPU acceleration easier to adopt in numerical workloads.
-
Four Ways to Deploy More Secure AI AgentsNVIDIA outlines four ways to deploy more secure AI agentsNVIDIA outlined four approaches to deploying AI agents more securely in production, covering access controls, guardrails, and monitoring. The guidance targets security risks that arise as autonomous agents take on real workloads.
-
ReToken: One Token to Improve Vision-Language Models for Visual Retrieval
-
MixFrag: Fragility-Guided Mixed-Precision Post-Training Quantization for Vision Transformers
-
Sample More, Reflect Less: Self-Refine and Reflexion Lose to Repeated Sampling at Equal Token Cost, from 1.5B to 7B
-
ORCA-bench: How Ready Are Language Model Agents for Oncall?
-
AI systems and the reproduction of (standard) language ideologies in World Englishes
-
Selective Credibility-Limited Belief Update
-
InfoOps Bench: A live information operations safety benchmark
-
Would You Walk to the Car Wash? Revealing the Salience Bias of Large Language Models in Commonsense Reasoning