Search: “models”

SpecQuant: Speculative Decoding with Multi-Parent Quantization for Adaptive LLM Inference

arXiv:2609.21704v1 Announce Type: new Abstract: Running large language models (LLMs) locally continues to be limited by restrictions of compute and memory on consumer hardware. The popular acceleratio

local-aiarxiv-cs-lg

21 Sept 2026

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Model Releases

Understanding LLM Quantization through Activation-Guided Compensation and Orthogonal Residuals

DGX agent

arXiv:2609.21450v1 Announce Type: new Abstract: Post-training weight-activation quantization reduces the memory and inference costs of large language models, but aggressive W4A4 quantization remains d

model-releasesarxiv-cs-lg

21 Sept 2026

DIY Jev

model-releasesr-localllama

So this jev thingy is getting kind of big... tbh it seems overhyped by a large margin, but here we are. Not that its bad, just feels like we usually ignored larger things... Anyway to the point of thi

GPT-6 Astra pushed a simulated person off a ledge in multiple trials. Grok, Gemini, and Claude did not.

model-releaseselon-musk--x

Research posted by Alex Wormuth indicates that in multiple simulation trials, the AI model GPT‑6 Astra interacted with a virtual person by pushing them off a ledge. In contrast, other contemporary mod

I built an open-source local agent experiment with Ollama to explore what sits between 'chat' and actual everyday work

Hey everyone, Like a lot of people here, I’ve been experimenting with running models locally through Ollama. Athena started as a small experiment, but it quickly grew into something more, a modular Py

model-releasesr-ollama

I gave Jev, Laya, finetuned ModernCE and Qwen3.5 the controls to Doom

I gave Jev, Laya, a finetuned ModernCE-base-nli and a finetuned Qwen3.5-4B the controls to Doom. Thanks to TypeSafe AI for Jev access. The video lines up the starts of four separate games with the sam

local-air-localllama

Qwen-Image-2.1 also improves visual quality over its predecessor, particularly in the aesthetics of text and portraits. Below are several ex…

Qwen‑Image‑2.1 improves upon its predecessor by delivering higher visual fidelity, notably in the rendering of textual content and portrait imagery. The model produces cleaner, more realistic portrait

model-releasesqwen--x

One of the craziest use cases I’ve found for Jev: verifiers. I am so excited about this that I at least wanted to share the high-level idea.…

One of the craziest use cases I’ve found for Jev: verifiers. I am so excited about this that I at least wanted to share the high-level idea. I used Jev to build a custom verifier for the /goal feature

model-releasesdair-ai--x

19 Sept 2026

ALIBI: Adversarial Legitimacy Injection in Binary Input against LLM Malware Analyzers

arXiv:2609.19722v1 Announce Type: cross Abstract: Large language models are being integrated into malware triage workflows as reasoning components that summarize static evidence and produce analyst-fa

Chunk Twice, Embed Once: A Systematic Study of Segmentation and Representation Trade-offs in Chemistry-Aware Retrieval-Augmented Generation

arXiv:2506.17277v2 Announce Type: replace-cross Abstract: The retrieval stage of retrieval-augmented generation (RAG) for scientific question answering depends on how documents are segmented and how c

EmbodiedMind: Adaptive Data Curation and Prefix-Tree Reinforcement Learning for Efficient Embodied Intelligence

arXiv:2609.19659v1 Announce Type: cross Abstract: Training embodied foundation models typically requires massive-scale datasets and extensive computational resources, yet often suffers from three crit

Evaluating Explanation Methods by the Predictors They Induce

arXiv:2609.20058v1 Announce Type: new Abstract: Explanations of machine learning models are usually judged by criteria that are hard to compare. We propose a simpler test: if an explanation really des

local-aiarxiv-cs-lg

F^{2}DR: A Fine-Grained Full-Pipeline Reward Framework for DeepSearch Workflows

arXiv:2609.19827v1 Announce Type: new Abstract: With the widespread industrial deployment of Large Language Models (LLMs), DeepSearch has emerged as the dominant paradigm for resolving complex user qu

Form Over Content In Gradient-Based Data Attribution Methods

arXiv:2609.19589v1 Announce Type: cross Abstract: Data attribution methods using gradient similarity are widely used to analyze and select training data for large language models, but what gradient si

Generative Verification: Rethinking the Uncertainty Signal for Active Learning of Object Detection

arXiv:2609.20262v1 Announce Type: new Abstract: Nearly every acquisition function for active object detection shares one arrangement, in that the model being improved is also the model being interroga

local-aiarxiv-cs-cv

Grok Bot summary of SpaceX CFO Bret Johnsen’s new interview with Tony Kim at SpaceX Mission Control. Bret’s path - Came from semiconductors …

model-releaseselon-musk--x

Grok Bot summary of SpaceX CFO Bret Johnsen’s new interview with Tony Kim at SpaceX Mission Control. Bret’s path - Came from semiconductors (Qualcomm, Broadcom; was CFO of MindSpeed) - Barely knew Spa

Intrinsic Sequence-Likelihood Confidence in Retrieval-Dominated Extractive QA: Two Pre-Specified Negatives, and What They Do and Do Not Attribute

arXiv:2609.19942v1 Announce Type: new Abstract: In extractive document question answering whose questions were generated from the passages that contain their answers -- so that retrieval recovers 92-9

KoNeoBench: A Curated Evaluation Dataset for LLM Understanding of Korean Neologisms

arXiv:2609.19916v1 Announce Type: cross Abstract: Large language models (LLMs) are typically evaluated on static benchmarks, even though natural language constantly evolves through newly emerging word

Learning-Based Reconstruction of Optical Properties in Bilayered Media from Single-distance Time-Resolved Reflectance Measurements

arXiv:2609.19786v1 Announce Type: new Abstract: The inverse problem of reconstructing optical properties, specifically absorption and scattering coefficients, in layered biological media from time-dom

OmniCalib: Target-Free, Task-Structured Self-Calibration for Humanoid Robots

model-releasesarxiv-cs-ro

arXiv:2609.19582v1 Announce Type: new Abstract: Assembly, wear, and component replacement perturb the sensor extrinsics and joint zeros encoded by a humanoid CAD model. Existing procedures calibrate o

PosteriorBench: From Point Estimates to Posterior Matching in Evaluating Generative Inverse Solvers

arXiv:2609.20794v1 Announce Type: new Abstract: Generative models are increasingly used to solve scientific inverse problems, but existing evaluations still focus primarily on whether a method can pro

QCPruner: Query-Conditioned Population Coverage for Visual Token Pruning

model-releasesarxiv-cs-cv

arXiv:2609.19990v1 Announce Type: new Abstract: The high visual-token load in multimodal large language models (MLLMs) motivates training-free pruning to reduce later-layer computation, but under a fi

SAFARI: An Industrial Benchmark for LLM-Assisted Hazard Analysis and Risk Assessment

arXiv:2609.20584v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly considered for safety-critical engineering, yet their reliability in regulated functional-safety workflows

Sample Count Is Not Enough: Candidate-Generation Strategy Shapes the Energy and Performance of LLM Test-Time Scaling

arXiv:2609.19499v1 Announce Type: new Abstract: Test-time scaling can improve large language model reasoning by generating and combining multiple candidate responses. In sampling-based methods, the in

SETTer: Sparse-Encoder Transformer for Long-term Multivariate Time Series Forecasting

arXiv:2609.20086v1 Announce Type: new Abstract: Long-term multivariate time series plays a significant role in many application areas such as power systems, trading, etc. However, their accurate predi

VakQA: A Benchmark and Evaluation Study for Telugu Spoken Factoid Question Answering

arXiv:2609.19879v1 Announce Type: new Abstract: Question answering has advanced rapidly with large language models, but predominantly for high-resource languages, in both text and spoken settings. Spo

Xeno-Interpretability: Investigating the Alien Minds of LLMs

arXiv:2609.20408v1 Announce Type: cross Abstract: Large language models are usually interpreted through concepts that humans already possess: truthfulness, refusal, deception, personality, harmfulness

safetyarxiv-cs-ai

A Zeroth-Order Paradigm for LLM Preference Alignment

arXiv:2609.19144v1 Announce Type: cross Abstract: Direct preference alignment methods are widely used to align large language models (LLMs) with human preferences because of their computational and me

AutoTuneBench: Trustworthy Measurement for Agent Auto-Tuning of LLM Serving Engines

arXiv:2609.18123v1 Announce Type: new Abstract: Large language model agents tune GPU kernels and serving engines through a closed loop of propose, measure, and keep, but the measurements behind this l

BadQubits: An LLM-Based Framework for Static Pre-Execution Detection of Structurally Harmful Quantum Circuits

arXiv:2609.18965v1 Announce Type: cross Abstract: This paper presents BadQubits, an LLM-based framework for static pre-execution detection of structurally harmful OpenQASM 2.0 circuits. The framework

Beyond Embedding Transfer: Component Roles in Grokking Transfer and Stability

arXiv:2609.18078v1 Announce Type: new Abstract: Warm-start transfer can make algorithmic tasks generalize rapidly, yet it is unclear which model components provide the gain and whether that gain remai

Beyond the Previous Layer: Residual Predictive Structure in Sparse MoE Routing

arXiv:2609.17940v1 Announce Type: new Abstract: Sparse mixture-of-experts models route each token through a sequence of expert selections. We ask whether the immediately preceding selection adequately

Call for compute - help optimize Qwen inference speed on local hardware

model-releasesr-localllama

Yoyo I renamed https://github.com/syv-ai/qwen38-27b-rtx3090 to https://github.com/syv-ai/HyperQwen/ to focus more on Qwen models and 'local' hardware in general, than just a single card and a single m

Compiled Agency: Frontier General-Purpose Coding Agents Build Winning Game Players from Bare Interaction - from Flappy Bird to StarCraft II and Civilization

arXiv:2609.18996v1 Announce Type: new Abstract: LLM agents have repeatedly struggled to convert knowledge of a game into competent play, even when researchers build the agent around the model - supply

safetyarxiv-cs-ai

CzechTopic: A Benchmark for Zero-Shot Topic Localization in Historical Czech Documents

arXiv:2603.03884v2 Announce Type: replace-cross Abstract: Topic localization aims to identify spans of text that express a given topic defined by a name and description. To study this task, we introdu

Decodability is Not Causality: Dissociating Probe Readouts from Behavioral Drivers via SAE Decomposition

arXiv:2609.18080v1 Announce Type: new Abstract: Linear probes can decode safety-relevant concepts such as truthfulness from language-model activations, but probe accuracy may show only decodability, n

safetyarxiv-cs-ai

Decoupling Vision, Language, and Action for Efficient Multi-Task Robot Policies

arXiv:2609.18374v1 Announce Type: new Abstract: Vision-Language-Action (VLA) models attach an action module to a Vision-Language Model (VLM) with billions of parameters and pay for that backbone at ev

safetyarxiv-cs-ro

deepseek v4.1 flash on together ai is leading on ttft for pre-warmed queries for agent workflows, that startup time shows up again on every …

model-releasestogether-ai--x

deepseek v4.1 flash on together ai is leading on ttft for pre-warmed queries for agent workflows, that startup time shows up again on every model turn, so shaving it down matters more as the loop gets

Tutorials

DR.WILSS: Diffusion-Based Replay for Weakly Supervised Continual Semantic Segmentation

arXiv:2609.18444v1 Announce Type: new Abstract: Weakly supervised class-incremental semantic segmentation (WILSS) aims to train a segmentation model over multiple steps, each introducing new concepts

tutorialsarxiv-cs-cv

Research

GeoCond: A Conditioning-Aware Reliability Adapter for Feed-Forward 3D Reconstruction

arXiv:2609.18465v1 Announce Type: new Abstract: Feed-forward 3D foundation models such as VGGT predict cameras, depth, and point maps in a single pass, but can fail silently under low overlap, low par

researcharxiv-cs-cv

GraphEcho: Structural Redundancy and Evidence Provenance in LLM Graph Agents

arXiv:2609.17695v1 Announce Type: new Abstract: A large language model (LLM) agent can follow more graph paths without acquiring more independent evidence. GraphEcho tests whether agents mistake these

How a solo founder runs a five-continent tender platform on AlloyDB and MCP

model-releasesgoogle-cloud-ai

Editor's note: Lucius AI, a tender-intelligence startup covering markets across five continents, runs its entire data platform on AlloyDB for PostgreSQL with a single operator. By migrating semantic s

Humor Arena: Which LLM is the funniest?

We compared 20 model versions on 360 frozen joke prompts, with four jokes requested per prompt and model names hidden from our humor-trained judge. Fable 5 had the highest estimated score: 66.8 points

local-air-localllama

Imitation Learning for Autonomous Driving in CARLA

arXiv:2609.17757v1 Announce Type: new Abstract: Behavioral cloning trains a policy offline on expert demonstrations, but deployment is closed loop: each action affects the observations the policy rece

Learning Where to Focus: Self-Supervised Multi-Scale ViTs for Histopathology

model-releasesarxiv-cs-cv

arXiv:2609.18578v1 Announce Type: new Abstract: Pathologists diagnose diseases by first locating suspicious tissue and then examining it at higher magnification, whereas self-supervised vision transfo

Agents

Market Signal Injection: Adversarial Context Manipulation of LLM Pricing Agents

arXiv:2609.18357v1 Announce Type: new Abstract: Large language model (LLM) pricing agents may respond to how market data is presented, even when its numerical values remain unchanged. We introduce mar

agentsarxiv-cs-ai

Multi-Appliance Non-Intrusive Load Monitoring via Label-Preserving Aggregate Recomposition and Prediction Consistency

arXiv:2609.18315v1 Announce Type: new Abstract: Non-intrusive load monitoring (NILM) estimates appliance power sequences from aggregate power, but models trained on source households commonly lose acc

Research

Objective vs. Search: Decomposing What Makes a Good Tokeniser

arXiv:2609.19145v1 Announce Type: cross Abstract: Two dominant tokenisation algorithms are used by modern language models: byte-pair encoding (BPE) and UnigramLM. These differ along two orthogonal axe

researcharxiv-cs-ai