Topic
architecture
Transformer Feed-Forward Block Linearity: Learned, Not Architectural, According to New Research
A new study introduces R^2_lin, a measure of linearity for transformer feed-forward blocks. Across models like GPT-2 and Pythia-160m, R^2_lin varies widely and is not determined by activation function. The findings offer targeted compression signals and reveal pitfalls in training linear baselines.
Agent Memory Forgetting Study Reveals Control-Plane Trade-offs for Enterprise AI Systems
A new architectural study of AI agent memory, based on 13 system configurations and a 385-case adversarial benchmark, reveals that forgetting failures—not recall failures—are the dominant cause of production errors. The research introduces ForgetEval, a benchmark for evaluating forgetting, and an Adapter Protocol for integrating heterogeneous memory stores. Three placement regimes for LLM intervention are compared, with a mutation-time hook achieving 91.7-93.2% overall accuracy at $0.17 per run.
Parallel Hybrid Architecture Combines GSS and Attention for Efficient Long-Context Language Modeling
Researchers propose the Parallel Hybrid Architecture (PHA), combining Gated State Spaces, Grouped Query Attention, and Feed-Forward Networks in parallel branches fused by a learnable mixing mechanism. On WikiText-103, PHA achieves 16.51 PPL at 125M parameters, outperforming comparable models, and scales to 180M parameters with 16.42 PPL while delivering 24% higher throughput and up to 40% lower memory usage.
A Framework for Governing Optimization in AI Systems: Architectural Wisdom
The paper 'Architectural Wisdom' argues that modern AI failures stem from optimizing underspecified objectives, not lack of intelligence. It proposes a corrigible objective-governance layer above the optimization substrate, made of four components and a six-coordinate wisdom tuple. The framework is motivated by eight cases of contemporary AI failures and aims to prevent harmful outcomes.
TrustedARI: A New Trust-Native Infrastructure Secures Agentic AI Routing for Enterprise Deployments
TrustedARI, presented by a research team on arXiv, is the first trust-native agentic routing infrastructure for agentic AI. It addresses fundamental trust risks in agent routing, offering a 39.34% reduction in handshake overhead and verifiable billing with 28.20x faster proof generation, all without modifying service providers.
Technology How Linear Achieves Millisecond Response Times: A Technical Breakdown for Enterprise Decision-Makers
Linear's web app updates issues in milliseconds by treating IndexedDB as the primary database, applying mutations locally before syncing via WebSocket. Co-founder Tuomas built the sync engine from day one. For CTOs evaluating performance, this approach eliminates network bottlenecks and loading states.