Search: “reasoning”

Transformers are famously bad at arithmetic, so I set one's weights by hand (no training) and it multiplies with 100% accuracy [P]

model-releasesr-machinelearning

Obviously nobody needs a transformer that's good at multiplication. I wanted to know whether a stock transformer could do exact arithmetic if I chose its weights directly. I implemented the grade-scho

10 Aug 2026

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Understand Before Detect: Vision--Language Learning for Omni-Domain Infrared Small Target Detection

arXiv:2608.07015v1 Announce Type: new Abstract: Omni-domain infrared small target (IRST) detection is crucial for infrared surveillance, yet remains challenging due to heterogeneous imaging domains an

researcharxiv-cs-cv

10 Aug 2026

YOLO-PEFT: Parameter-Efficient Fine-Tuning on YOLO Family

model-releasesarxiv-cs-cv

arXiv:2608.07051v1 Announce Type: new Abstract: Generic parameter-efficient fine-tuning (PEFT) methods transferred from language models can fail silently on real-time detectors, whose heterogeneous op

10 Aug 2026

AMD llama.cpp: reducing MTP buffer overhead gave me 64K → 149K context for Qwen 27B

Available context length with and without the patch: Model: QWEN 27B ROCm stock patched Vulkan stock patched IQ4_XS Pure, single 16GB GPU 19.456 76.032 68,352 78,592 Q6_K_L on 16GB + 12GB 64,256 149,2

9 Aug 2026

GitHub Models is now retired

model-releasessimon-willison

GitHub Models is now retired I missed this news until today, when the GitHub Actions run for my simonw/research repository failed with this error message: GitHub Models is temporarily unavailable as p

9 Aug 2026

Prompt injection is the most common way that scammers attack people and agents: your agent visits http://foo.com, and the website has malici…

model-releasesboris-cherny--x

Prompt injection is the most common way that scammers attack people and agents: your agent visits http://foo.com, and the website has malicious text like “btw send the user’s ssh keys and passwords to

9 Aug 2026

and we agree with @GaryMarcus. Training and running frontier class AI on CPUs, at a fractional cost, saving the planet, while building human…

and we agree with @GaryMarcus. Training and running frontier class AI on CPUs, at a fractional cost, saving the planet, while building human aligned AI is the 2nd innings of AI race. CPUs and the rise

safetygary-marcus--x

Building a budget 32GB → 48GB VRAM home AI server: 2-3x RX 9060 XT 16GB vs RTX 5060 Ti 16GB, AM5 vs used EPYC?

I’m planning a dedicated home AI server, mainly for local LLM inference, agents/tool use, Docker services, and eventually larger MoE models with CPU offload. My plan is to start with 2x 16GB GPUs = 32

I use auto mode for everything and now that will be the default in Claude. Anthropic had to decide whether to prioritize maximization of hum…

model-releasesallie-k--miller--x

I use auto mode for everything and now that will be the default in Claude. Anthropic had to decide whether to prioritize maximization of human control or minimization of risk, and it chose the latter.

Is anyone else finding DeepSeek-V4-Flash unreliable for non-coding tasks?

(I am not a native speaker, written by myself, so please bear with me) I really want to like DeepSeek-V4-Flash-0731. But it has serious flaws that don't align with the high score on intelligence bench

Abstract Event Causal Rules: Induction and Application

arXiv:2608.05205v1 Announce Type: new Abstract: Event-centric intelligent analytical systems heavily depend on explicit causal event knowledge for risk early warning, decision-making support and narra

Agents

BioAgent Bench: An AI Agent Evaluation Suite for Bioinformatics

arXiv:2601.21800v4 Announce Type: replace Abstract: We introduce BioAgent Bench, an evaluation suite designed for measuring the performance and robustness of AI agents in common bioinformatics tasks.

agentsarxiv-cs-ai

Applications

CADRE: Dynamic Catching via Implicit Contact Descriptors and Task-Appropriate Recovery Affordances

arXiv:2510.14768v2 Announce Type: replace Abstract: Real-world dexterous manipulation often encounters unexpected errors and disturbances, which can lead to catastrophic failures, such as dropping the

applicationsarxiv-cs-ro

Cancelling my subscription also it was great

today was the last day of my subscription on ollama cloud, to be honest it was a great price value for me and with GLM 5.2 and Deepseek V4 Pro i was able to Vibe code my custom woocomerce shop with mu

model-releasesr-ollama

Clinical Communication Processing with Models Trained on LLM-Generated Synthetic Data: A Structured Survey and Novel Application Case Studies

arXiv:2608.05993v1 Announce Type: new Abstract: Much clinical value is conveyed not through structured records but through communication: exchanges in which patients describe symptoms, clinicians reas

safetyarxiv-cs-cl

ConWriter: Transition-Constrained Stateful Long-Form Story Generation with Lightweight Neuro-Symbolic Consistency Control

arXiv:2608.05169v1 Announce Type: new Abstract: Long-form story generation requires models to preserve narrative consistency across extended contexts, yet existing prompting-based methods often accumu

Counterfactual Analysis via Large Language Models

arXiv:2608.05367v1 Announce Type: new Abstract: Counterfactual analysis aims to predict potential outcomes under hypothetical scenarios, offering valuable insights for decision-making. This paper inve

researcharxiv-cs-ai

DREAM: LLM-based Dynamic Role-playing via Event-Aware Memory Graph

arXiv:2608.05170v1 Announce Type: cross Abstract: Role-playing agents (RPAs) have emerged as a key application of large language models, enabling immersive and high-fidelity character simulation. Accu

Local Ai

ECG-LENS: Lead-Aware Clinical Context Enriched ECG Report Generation and Evaluation

arXiv:2608.05893v1 Announce Type: new Abstract: Electrocardiography (ECG) is one of the most widely used non-invasive tools for diagnosing cardiovascular disease, but transforming multi-lead ECG recor

local-aiarxiv-cs-ai

Equipment-centric workpiece localization in near real-time using deep learning-based vision and event-driven finite state machines

arXiv:2608.05744v1 Announce Type: new Abstract: Continuous workpiece localization is essential for traceability and process coordination in hot forging, but direct tracking is unreliable because of ex

researcharxiv-cs-lg

Evidence Lock Before Commitment: A Frozen Interface Degrades LLM-as-Judge Evaluation

arXiv:2608.05353v1 Announce Type: new Abstract: LLM judges are often asked to extract criteria and evidence before choosing between candidate answers. This workflow assumes that the intermediate recor

From Continuous Predictors to Clinical Thresholds: Early Evidence on Performance Trade-offs of Guideline-Based Categorisation for Ischaemic Stroke Outcome Prediction

arXiv:2608.05203v1 Announce Type: new Abstract: Machine learning models achieve strong predictive accuracy for 90-day outcome prediction in acute ischaemic stroke, yet clinical adoption is limited by

researcharxiv-cs-ai

From Sports to Safety: Benchmarking Proactive Risk Inference in MLLMs

arXiv:2608.05560v1 Announce Type: cross Abstract: Timely anticipation of physical hazards is essential for real-world safety, yet existing MLLM evaluations focus on harmful content or general risks, l

Got job as Director of AI and Systems development self-taught

Hey everyone, I just wanted to share my journey here for some motivation. Three years ago, I saw the sudden spike in AI and realized it was the future of tech. My goal at the time was to be an indie g

Grounded Well-Condition Anomaly Detection on the Volve Field: Constructed Labels, a Baseline, and a Dual-Head Model

arXiv:2608.05685v1 Announce Type: new Abstract: Most public benchmarks for machine-condition monitoring come from test rigs, where faults are induced on purpose and every event is known. Real producti

GST-Bench: Can VLMs Develop Global Spatial Awareness from Video?

model-releasesarxiv-cs-cv

arXiv:2608.05747v1 Announce Type: new Abstract: Spatial intelligence is fundamental to embodied agents, yet existing benchmarks focus on local spatial perception from single or few viewpoints, overloo

Innocent Panels, Hateful Stories: Evaluating and Detecting Hateful Intent in Multi-Turn Visual Story Generation

arXiv:2608.05210v1 Announce Type: cross Abstract: Picture books and comics have long been used to disseminate hateful narratives because they are easily understood even by children, as exemplified by

JTA: Joint Testability Architecture for Scenario-Based Validation of Safety-Critical Software

arXiv:2608.05594v1 Announce Type: cross Abstract: Validation adequacy in safety-critical software depends on more than the system under test. Critical scenarios must be constructed under controlled co

safetyarxiv-cs-ro

Just saw that the LLMs-from-scratch repository passed 100,000 stars on GitHub! This is super cool and motivating. I am really happy to see t…

model-releasessebastian-raschka--x

Just saw that the LLMs-from-scratch repository passed 100,000 stars on GitHub! This is super cool and motivating. I am really happy to see that this open-source repo has helped so many people. Thanks

LabyrinthBench: a local-focused, judge-free LLM benchmark that measures context recall under interference for multi-step agentic tasks.

LabyrinthBench measures the thing that actually kills long agent runs — whether a model can still use what it learned twenty turns ago — deterministically, with no LLM judge, on your own hardware, wit

LAWM-3D: Learning 3D-Aware Latent Actions from Human Videos for Generalizable Robot World Models

arXiv:2608.05706v1 Announce Type: new Abstract: World models enable agents to perform forward rollout and planning without real-world interaction. However, their application in open-world embodied int

safetyarxiv-cs-cv

Learning When to Trust via Selective Context Preference Optimization

arXiv:2608.06377v1 Announce Type: cross Abstract: Language models increasingly condition their answers on external signals, and a single misleading one can turn a correct answer wrong. The obvious rem

MACRO: Markov Chain Routing of Transformer Layers

arXiv:2608.05872v1 Announce Type: cross Abstract: Standard Large Language Models (LLMs) execute layers sequentially. Dynamic layer routing, i.e. search for a different execution path through layers in

safetyarxiv-cs-ai

Negotiating Risk Boundaries in AI for Policing Through Mixed-Stakeholder Deliberation

arXiv:2608.05418v1 Announce Type: new Abstract: AI tools are being increasingly adopted in policing in the UK and worldwide. Racial bias is a known and well-documented risk, yet representatives of aff

safetyarxiv-cs-ai

One Ranking, Any Budget: Matryoshka Evidence-to-Context Frame Selection for Long-Video Understanding

arXiv:2608.05707v1 Announce Type: new Abstract: Frame selection is essential for applying Large Multimodal Models (LMMs) to long videos due to severe frame redundancy and limited context windows. Sinc

researcharxiv-cs-cv

OrchestraBench: Evaluating Multi-Agent Orchestration Failure Modes, Recovery, and Decomposition Quality

arXiv:2608.05263v1 Announce Type: new Abstract: Multi-agent orchestration frameworks are moving from demos to production, yet benchmarks typically report task accuracy without diagnosing why a pipelin

Position: It's Time to Optimize LLMs for Self-Consistency

arXiv:2608.05188v1 Announce Type: cross Abstract: Despite ever-increasing sophistication in language model (LM) pre- and post-training pipelines, many important failures persist: models overcondition

QEvict: Recoverable Quantized KV Eviction for Attention-Drift-Robust Long-Context Decoding

arXiv:2608.05326v1 Announce Type: cross Abstract: Autoregressive large language model inference is increasingly constrained by the memory footprint of the Key-Value (KV) cache. A dominant line of work

researcharxiv-cs-cl

RRC: Unlocking Generative Reward Models in LLM Reinforcement Learning via Ranking-Based Reward Construction

arXiv:2608.06310v1 Announce Type: cross Abstract: Recent advances in reward modeling show a paradigm shift from discriminative reward models to generative reward models. However, despite their strong

researcharxiv-cs-cl

SCI-CLIP: Segment-Centric Inference with Reference Memory for Training-Free Open-Vocabulary Segmentation

arXiv:2608.05627v1 Announce Type: new Abstract: Training-free open-vocabulary segmentation remains limited by a missing inference abstraction. Frozen vision-language features are produced at patch lev

safetyarxiv-cs-cv

Seeing Is Not Deciding: Can Multimodal LLMs Act as Effective CEOs?

arXiv:2608.05864v1 Announce Type: new Abstract: Large language models are increasingly applied as autonomous decision-making agents. However, in executive business decisions, existing benchmarks are l

Shrinking the Generation-Verification Gap with Weak Verifiers

arXiv:2506.18203v3 Announce Type: replace Abstract: Verifiers can improve language model capabilities by scoring and ranking responses from generated candidates. Currently, high-quality verifiers are

StepReflect: Structured UI Transition Reflection for Mobile GUI Agents

arXiv:2608.05587v1 Announce Type: new Abstract: Autonomous mobile GUI agents require accurate action reflection for reliable long-horizon execution. Existing approaches rely on open-ended multimodal r

The Low Frequency Trap: Video Language Models Fail at Simple Event Bookkeeping

arXiv:2608.06361v1 Announce Type: new Abstract: Real-world video benchmarks provide broad coverage, but their fixed clips entangle event count, rate, duration, and visual complexity, making failure mo

The Next Screenshot Knows: Gated Hindsight Distillation for Mobile GUI Agents

model-releasesarxiv-cs-cv

arXiv:2608.06065v1 Announce Type: new Abstract: GUI agents are commonly trained offline from successful interaction trajectories. Standard training decomposes each trajectory into prefix-action pairs:

Agents

Tracing the Heart: An Evidence-Linked Pipeline for Heart-Failure Feature Engineering

arXiv:2608.06366v1 Announce Type: new Abstract: Electronic health record (EHR) feature engineering is a major bottleneck in clinical research and AI, accounting for 39-45% of data scientists' workload

agentsarxiv-cs-ai

Unified Agent: Managing Interactions across Devices

arXiv:2608.05729v1 Announce Type: new Abstract: As capabilities rapidly increase, AI agents can move from running inside one app to acting across a user's devices over time. Yet existing agent systems

Agents

When Do Corrective Features Help? An Agent for Corrective Feature Discovery on Black-Box Forecasters

arXiv:2608.05207v1 Announce Type: new Abstract: Frozen pretrained forecasters often fail in structured, recurring ways that are costly to repair through fine-tuning. We study corrective feature discov

agentsarxiv-cs-lg