Hardware/Chips Hacker News

The Return of Rigorous Full-System Timing Simulation

A discussion on the revival and importance of rigorous full-system timing simulation in computer architecture.

Software Engineering Hacker News

Bringing down my ZSH load times from ~3.1s to ~230ms

A guide on optimizing ZSH shell load times, reducing them from over 3 seconds to 230ms.

Other The Verge

VSCO launches Studio Pro mobile photo editing app and plans $500 per year subscription

VSCO launches Studio Pro, a mobile photo editing app for high-volume projects with a high annual subscription fee.

AI/ML arXiv cs.AI

Implicit vs. Explicit Prompting Strategies for LVLMs in Referential Communication

Research comparing implicit and explicit prompting in Vision-Language Models (LVLMs), finding that explicit prompting is necessary for communicative efficiency.

AI/ML arXiv cs.AI

MeiBRD: Meta-Learning Intraoperative Biomechanical Residual Deformation

Introduction of MeiBRD, a hybrid registration framework using meta-learning to improve intraoperative liver registration accuracy.

AI/ML arXiv cs.AI

Model Validation of Agentic AI Systems: A POMDP-Based Framework for Belief-State, Forecast, and Policy Validation

A proposed POMDP-based framework for the validation and governance of autonomous agentic AI systems, focusing on belief-state and policy validation.

AI/ML arXiv cs.AI

TerraTransfer: Learning End-to-End Driving Policies Without Expert Demonstrations

TerraTransfer proposes a method for end-to-end autonomous driving policies without needing expert demonstrations by leveraging self-play in simulators.

AI/ML arXiv cs.AI

Visuals Lie, Consistency Speaks: Disentangling Spatial Attention from Reliability in Vision-Language Models

A study challenging the Attention-Confidence Assumption in Vision-Language Models, finding that reliability is driven by generation dynamics rather than spatial attention.

AI/ML arXiv cs.AI

NarrativeWorldBench: A Frontier-Saturated Benchmark and a Latent World Model for Long-Horizon Co-Creative Audio Drama

Introduction of NarrativeWorldBench for long-horizon narrative evaluation and N-VSSM, a state-space model for consistent long-form audio drama creation.

Cybersecurity arXiv cs.AI

SoK: AI-Augmented Binary Reversing

A comprehensive systematization of knowledge (SoK) on AI-augmented binary reversing, analyzing 144 papers to create a unified taxonomy.

Other Hacker News

Krugman: Musk, a Human Ponzi Scheme

An opinion piece by Paul Krugman criticizing Elon Musk's business practices, characterizing him as a 'human Ponzi scheme'.

Software Engineering Hacker News

Made a free macOS menu bar app that fixes typing in the wrong keyboard layout

A developer has released a free macOS menu bar app to solve the common issue of typing in the wrong keyboard layout.

Tech Business/VC TechCrunch

Roelof Botha joins SpaceX’s board of directors

Roelof Botha of Sequoia Capital has joined the board of directors at SpaceX following the company's IPO.

Tech Business/VC TechCrunch

After unveiling ridiculously expensive AR glasses, Snap’s stock takes a dive

Snap's stock price declined following the unveiling of their new, expensive AR glasses.

Tech Business/VC TechCrunch

NEA’s Tiffany Luck says enterprises are still figuring out their AI ROI

Enterprises are struggling to find a clear return on investment (ROI) for AI, with some companies exceeding their budgets rapidly.

AI/ML arXiv cs.AI

Geometry-Consistent Endoscopic Representations for Image-Guided Navigation via Structured Foundation Model Adaptation

Researchers propose a framework for learning geometry-consistent image representations in monocular endoscopy for better navigation.

AI/ML arXiv cs.AI

Counterfactual Optimization of Baseball Pitch Sequences and Estimation of Its Impact on Season-Level Statistics

A study uses a Transformer-based model to optimize baseball pitch sequences and estimate their impact on seasonal statistics.

AI/ML arXiv cs.AI

Do Large Language Models Always Tell The Same Stories?

An investigation into LLM-generated stories reveals that frontier models tend to converge on generic narratives, lacking human-like diversity.

AI/ML arXiv cs.AI

Translating the Untranslatable: An Operationalizable Ontology for Untranslatability

The paper introduces an ontology and dataset for 'untranslatability' to improve machine translation of sentences where meaning cannot be directly preserved.

AI/ML arXiv cs.AI

DriveJudge: Rethinking Autonomous Driving Evaluation with Vision-Language Models

DriveJudge is introduced as a driving evaluation agent that combines rule-grounded evaluation with VLM reasoning for autonomous driving.