All entries

Optimal Rates for Pure {arepsilon}-Differentially Private Stochastic Convex Optimization with Heavy Tails

arXiv:2604.06492v1 Announce Type: new Abstract: We study stochastic convex optimization (SCO) with heavy-tailed gradients under pure epsilon-differential privacy (DP). Instead of assuming a bound on t

Optimizing Vercel Sandbox snapshots

DGX agent

Vercel Sandbox snapshots capture the complete filesystem state of a running sandbox — including installed packages and configured environments — allowing new sandboxes to be launched from that save...

toolsvercel-blog

10 Apr 2026

Tools

Opus 4.6 Fast Mode available on AI Gateway

Vercel AI Gateway now supports Fast Mode for Claude Opus 4.6, an early experimental feature that delivers 2.5x faster output token speeds while maintaining the same model intelligence. It is partic...

toolsvercel-blog

Oracle is down more than 50% since two men (jointly) took over Safra Catz’s CEO job. Sexism is up 500%? 5000%?

Oracle is down more than 50% since two men (jointly) took over Safra Catz’s CEO job. Sexism is up 500%? 5000%? Nick Fuentes says women can only do three things “Women can be three things: they can be

safetygary-marcus--x

Agents

ORACLE-SWE: Quantifying the Contribution of Oracle Information Signals on SWE Agents

arXiv:2604.07789v1 Announce Type: cross Abstract: Recent advances in language model (LM) agents have significantly improved automated software engineering (SWE). Prior work has proposed various agenti

agentsarxiv-cs-cl

Agents

OrgForge: A Multi-Agent Simulation Framework for Verifiable Synthetic Corporate Corpora

arXiv:2603.14997v2 Announce Type: replace Abstract: Building and evaluating enterprise AI systems requires synthetic organizational corpora that are internally consistent, temporally structured, and c

agentsarxiv-cs-cl

Industry

Orion helium leak no threat to Artemis II reentry but will require redesign

During the Artemis II mission, a helium leak was identified within the oxidizer pressurization system of the European Service Module's propulsion system. The in-flight leak rate was 'an order of...

industryars-technica

Orion-Lite: Distilling LLM Reasoning into Efficient Vision-Only Driving Models

arXiv:2604.08266v1 Announce Type: new Abstract: Leveraging the general world knowledge of Large Language Models (LLMs) holds significant promise for improving the ability of autonomous driving systems

oslash Source Models Leak What They Shouldn't nrightarrow: Unlearning Zero-Shot Transfer in Domain Adaptation Through Adversarial Optimization

arXiv:2604.08238v1 Announce Type: new Abstract: The increasing adaptation of vision models across domains, such as satellite imagery and medical scans, has raised an emerging privacy risk: models may

Our first successful Gemma 4 Runtime in London with @swyx @patloeber @nick_kango @cormacb and others! 💎Great to go out for a run and talk a…

Our first successful Gemma 4 Runtime in London with @swyx @patloeber @nick_kango @cormacb and others! 💎Great to go out for a run and talk about Gemma, agents, evals and more @osanseviero @swyx and oth

model-releasesswyx--x

Our Lab just posted a new research report from Zimran Ahmed about how the game industry is adapting to AI. He spoke to people at 20 differen…

applicationsethan-mollick--x

Our Lab just posted a new research report from Zimran Ahmed about how the game industry is adapting to AI. He spoke to people at 20 different studios and found a wide range of approaches to adapt (or

Our response to the Axios developer tool compromise

Axios, a widely used third-party JavaScript developer library with approximately 100 million weekly downloads, was compromised on March 31, 2026, as part of a broader software supply chain attack a...

model-releasesopenai

OV-Stitcher: A Global Context-Aware Framework for Training-Free Open-Vocabulary Semantic Segmentation

arXiv:2604.08110v1 Announce Type: new Abstract: Training-free open-vocabulary semantic segmentation(TF-OVSS) has recently attracted attention for its ability to perform dense prediction by leveraging

OVS-DINO: Open-Vocabulary Segmentation via Structure-Aligned SAM-DINO with Language Guidance

arXiv:2604.08461v1 Announce Type: new Abstract: Open-Vocabulary Segmentation (OVS) aims to segment image regions beyond predefined category sets by leveraging semantic descriptions. While CLIP based a

OxEnsemble: Fair Ensembles for Low-Data Classification

arXiv:2512.09665v2 Announce Type: replace Abstract: We address the problem of fair classification in settings where data is scarce and unbalanced across demographic groups. Such low-data regimes are c

[P] ibu-boost: a GBDT library where splits are absolutely rejected, not just relatively ranked[P]

researchr-machinelearning

The search results did not return the specific Reddit post about ibu-boost. Let me try fetching it directly. I was unable to retrieve the specific Reddit post or any direct information about the **...

PAC-Bayesian Bounds on Constrained f-Entropic Risk Measures

arXiv:2510.11169v2 Announce Type: replace-cross Abstract: PAC generalization bounds on the risk, when expressed in terms of the expected loss, are often insufficient to capture imbalances between subg

researcharxiv-cs-lg

PANC: Prior-Aware Normalized Cut via Anchor-Augmented Token Graphs

arXiv:2602.06912v2 Announce Type: replace Abstract: Unsupervised segmentation from self-supervised ViT patches holds promise but lacks robustness: multi-object scenes confound saliency cues, and low-s

PanoSAM2: Lightweight Distortion- and Memory-aware Adaptions of SAM2 for 360 Video Object Segmentation

arXiv:2604.07901v1 Announce Type: new Abstract: 360 video object segmentation (360VOS) aims to predict temporally-consistent masks in 360 videos, offering full-scene coverage, benefiting applications,

Paragraph Segmentation Revisited: Towards a Standard Task for Structuring Speech

arXiv:2512.24517v2 Announce Type: replace Abstract: Automatic speech transcripts are often delivered as unstructured word streams that impede readability and repurposing. We recast paragraph segmentat

researcharxiv-cs-cl

Agents

ParkSense: Where Should a Delivery Driver Park? Leveraging Idle AV Compute and Vision-Language Models

arXiv:2604.07912v1 Announce Type: new Abstract: Finding parking consumes a disproportionate share of food delivery time, yet no system addresses precise parking-spot selection relative to merchant ent

agentsarxiv-cs-cv

ParseBench: A Document Parsing Benchmark for AI Agents

arXiv:2604.08538v1 Announce Type: new Abstract: AI agents are changing the requirements for document parsing. What matters is semantic correctness: parsed output must preserve the structure and

Part^{2}GS: Part-aware Modeling of Articulated Objects using 3D Gaussian Splatting

arXiv:2506.17212v2 Announce Type: replace Abstract: Articulated objects are common in the real world, yet modeling their structure and motion remains a challenging task for 3D reconstruction methods.

PASK: Toward Intent-Aware Proactive Agents with Long-Term Memory

model-releasesarxiv-cs-cl

arXiv:2604.08000v1 Announce Type: cross Abstract: Proactivity is a core expectation for AGI. Prior work remains largely confined to laboratory settings, leaving a clear gap in real-world proactive age

Path Regularization: A Near-Complete and Optimal Nonasymptotic Generalization Theory for Multilayer Neural Networks and Double Descent Phenomenon

arXiv:2503.02129v2 Announce Type: replace-cross Abstract: Path regularization has shown to be a very effective regularization to train neural networks, leading to a better generalization property than

safetyarxiv-cs-ai

PD-SOVNet: A Physics-Driven Second-Order Vibration Operator Network for Estimating Wheel Polygonal Roughness from Axle-Box Vibrations

arXiv:2604.06620v1 Announce Type: new Abstract: Quantitative estimation of wheel polygonal roughness from axle-box vibration signals is a challenging yet practically relevant problem for rail-vehicle

applicationsarxiv-cs-lg

Industry

Peak AI response to a normal gardening question

I was unable to retrieve the specific Reddit post at the provided URL, and my search did not return results directly related to it. The content of that particular Reddit thread — its screenshot, AI...

industryr-chatgpt

PEER: Unified Process-Outcome Reinforcement Learning for Structured Empathetic Reasoning

arXiv:2508.09521v2 Announce Type: replace Abstract: Emotional support conversations require more than fluent responses. Supporters need to understand the seeker's situation and emotions, adopt an appr

safetyarxiv-cs-cl

People ask me how I choose what model to route queries to. It's simple. Claude gets knowledge work anything below that would be degrading Cl…

model-releasesdylan-patel--x

I was unable to retrieve the specific tweet content from that URL, as the post requires a logged-in X (Twitter) session to access, and search results did not surface the full text of that specific ...

People in Washington get played, yet again We really should worry about cybersecurity - a lot – but Mythos is not the model these guys think…

People in Washington get played, yet again We really should worry about cybersecurity - a lot – but Mythos is not the model these guys think it is. (See my newsletter today for three reasons why it is

safetygary-marcus--x

Tools

People of pi. You can view my 18 minutes @aiDotEngineer Europe talk here. May it spark joy. And laughter. https://www.youtube.com/live/_zdro…

Mario Zechner ([@badlogicgames](https://x.com/badlogicgames)), creator of the open-source **pi coding agent** (`@mariozechner/pi-coding-agent`), shared a recording of his 18-minute talk at the **ai...

toolsswyx--x

PeReGrINE: Evaluating Personalized Review Fidelity with User Item Graph Context

model-releasesarxiv-cs-cl

arXiv:2604.07788v1 Announce Type: cross Abstract: We introduce PeReGrINE, a benchmark and evaluation framework for personalized review generation grounded in graph-structured user--item evidence. PeRe

Personalized RewardBench: Evaluating Reward Models with Human Aligned Personalization

arXiv:2604.07343v1 Announce Type: cross Abstract: Pluralistic alignment has emerged as a critical frontier in the development of Large Language Models (LLMs), with reward models (RMs) serving as a cen

Tutorials

Personalizing ChatGPT

The OpenAI Academy 'Personalizing ChatGPT' tutorial covers several methods users can employ to tailor ChatGPT's behavior to their needs, including custom instructions, memory, and prompt engineerin...

tutorialsopenai

Personalizing Text-to-Image Generation to Individual Taste

arXiv:2604.07427v1 Announce Type: new Abstract: Modern text-to-image (T2I) models generate high-fidelity visuals but remain indifferent to individual user preferences. While existing reward models opt

Phantasia: Context-Adaptive Backdoors in Vision Language Models

arXiv:2604.08395v1 Announce Type: new Abstract: Recent advances in Vision-Language Models (VLMs) have greatly enhanced the integration of visual perception and linguistic reasoning, driving rapid prog

Phantom: Physics-Infused Video Generation via Joint Modeling of Visual and Latent Physical Dynamics

arXiv:2604.08503v1 Announce Type: new Abstract: Recent advances in generative video modeling, driven by large-scale datasets and powerful architectures, have yielded remarkable visual realism. However

applicationsarxiv-cs-cv

PhyAVBench: A Challenging Audio Physics-Sensitivity Benchmark for Physically Grounded Text-to-Audio-Video Generation

model-releasesarxiv-cs-ai

arXiv:2512.23994v3 Announce Type: replace-cross Abstract: Text-to-audio-video (T2AV) generation is central to applications such as filmmaking and world modeling. However, current models often fail to

PhyEdit: Towards Real-World Object Manipulation via Physically-Grounded Image Editing

arXiv:2604.07230v2 Announce Type: replace Abstract: Achieving physically accurate object manipulation in image editing is essential for its potential applications in interactive world models. However,

Physical Adversarial Attacks on AI Surveillance Systems:Detection, Tracking, and Visible--Infrared Evasion

arXiv:2604.06865v1 Announce Type: cross Abstract: Physical adversarial attacks are increasingly studied in settings that resemble deployed surveillance systems rather than isolated image benchmarks. I

researcharxiv-cs-ai

Physical Knot Classification Beyond Accuracy: A Benchmark and Diagnostic Study

arXiv:2603.23286v3 Announce Type: replace Abstract: Physical knot classification is a fine-grained task in which the intended cue is rope crossing structure, but high accuracy may still come from appe

Tutorials

Physically Plausible Human-Object Rendering from Sparse Views via 3D Gaussian Splatting

arXiv:2503.09640v2 Announce Type: replace-cross Abstract: Rendering realistic human-object interactions (HOIs) from sparse-view inputs is a challenging yet crucial task for various real-world applicat

tutorialsarxiv-cs-cv

Physics-Informed Functional Link Constrained Framework with Domain Mapping for Solving Bending Analysis of an Exponentially Loaded Perforated Beam

arXiv:2604.07025v1 Announce Type: cross Abstract: This article presents a novel and comprehensive approach for analyzing bending behavior of the tapered perforated beam under an exponential load. The

Physics-Informed Neural Networks for Joint Source and Parameter Estimation in Advection-Diffusion Equations

arXiv:2512.07755v3 Announce Type: replace-cross Abstract: Recent studies have demonstrated the success of deep learning in solving forward and inverse problems in engineering and scientific computing

Physics-Informed Spectral Modeling for Hyperspectral Imaging

arXiv:2508.21618v2 Announce Type: replace-cross Abstract: We present PhISM, a physics-informed deep learning architecture that learns without supervision to explicitly disentangle hyperspectral observ

researcharxiv-cs-ai

PIArena: A Platform for Prompt Injection Evaluation

arXiv:2604.08499v1 Announce Type: cross Abstract: Prompt injection attacks pose serious security risks across a wide range of real-world applications. While receiving increasing attention, the communi

applicationsarxiv-cs-cl

PIKA: Expert-Level Synthetic Datasets for Post-Training Alignment from Scratch

model-releasesarxiv-cs-cl

arXiv:2510.06670v2 Announce Type: replace Abstract: High-quality instruction data is critical for LLM alignment, yet existing open-source datasets often lack efficiency, requiring hundreds of thousand

Tools

Pills for llm coding from an OpenAI talk at @aiDotEngineer : - have single shared utils to do stuff like bounded concurrency - treat errors …

Pills for llm coding from an OpenAI talk at @aiDotEngineer : - have single shared utils to do stuff like bounded concurrency - treat errors as values and part of the control flow - use telemetry for e

toolsswyx--x