Search: “models”

When Contextual Inference Fails: Cancelability in Interactive Instruction Following

arXiv:2603.19997v2 Announce Type: replace Abstract: We investigate the separation of literal interpretation from contextual inference in a collaborative block-building tasks, where an agent must resol

21 Aug 2026

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Local Ai

Write Once, Run Everywhere: The Axon DSL for Shape-Safe and Framework-Agnostic LLM Architectures

DGX agent

arXiv:2608.19889v1 Announce Type: new Abstract: The entire ecosystem of open-source language models effectively relies on a single platform. What if this platform was forced to shut down tomorrow? Imp

local-aiarxiv-cs-ai

21 Aug 2026

AMD setup is fast with 200k ctx with Qwen 3.8, i didn't understand how/why?

Hi all, it is my very first post, because I am so confused. First my setup: I have 2 X 7900XTX with rocm7.2.1 and I use it to fine tune small models and local llm etc. I run Qwen 3.8 27b 8 bit version

model-releasesr-ollama

Beyond the Transcript: Detecting Covert Co ordination in Latent Multi-Agent Communication

arXiv:2608.19161v1 Announce Type: new Abstract: Language-model agents can communicate through continuous hidden states that are invisible in public transcripts, creating opportunities for covert harmf

Bidirectional representational alignment between biological and artificial neural networks

arXiv:2608.18244v1 Announce Type: cross Abstract: Recent work has shown that representational alignment between biological and artificial neural networks is asymmetric: model representations predict n

safetyarxiv-cs-ai

ClosureBench: A Constructive Benchmark for Compositional Graph Reasoning

model-releasesarxiv-cs-lg

arXiv:2608.18242v1 Announce Type: new Abstract: We introduce ClosureBench, a constructive benchmark for compositional graph-relational reasoning with programmatically verified ground truth. Unlike fix

Comment-level Topic Drift Analysis in the Reddit Corpus

arXiv:2608.19133v1 Announce Type: new Abstract: We present a novel application of embedding-based dynamic topic modeling techniques to detect and quantify topic drift at the comment level in a massive

researcharxiv-cs-cl

ERASE: EaRly bAckpropagation SchEdule for Faster Training of Modern Recommendation Systems

arXiv:2608.18469v1 Announce Type: cross Abstract: Lightweight proxy models enable rapid experimentation without repeatedly training frontier-scale systems, but their small kernels often leave modern a

Figurative and Cultural Knowledge in LLMs: Investigating Cross-Domain Transfer through Fine-Tuning

arXiv:2608.18361v1 Announce Type: new Abstract: Figurative language is deeply culturally embedded; fluent use requires not just linguistic competence but cultural immersion. We ask whether LLMs can le

FiLoRA: Focus-and-Ignore LoRA for Controllable Feature Reliance

arXiv:2602.02060v2 Announce Type: replace-cross Abstract: Multimodal foundation models integrate heterogeneous signals across modalities, yet it remains unclear whether their predictions can be contro

Grading the Graders: Verification Autonomy Levels (L0-L5) for LLM Reasoning

arXiv:2608.19009v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly paired with verifiers (step checkers, self-consistency filters, tool-based fact checkers, formal proof ass

Applications

GraphK: Variable-Size Graph Generation with Efficient Edge Construction

arXiv:2608.18777v1 Announce Type: new Abstract: Graph generation models have advanced significantly with deep learning, yet they remain limited in scalability, flexibility, and ability to model underl

applicationsarxiv-cs-lg

I built a visual multi-agent workflow editor that exports runnable CrewAI code — here’s a 5-agent workflow running locally with Ollama

I've been building AgentGraph Studio, a visual editor for designing multi-agent workflows and exporting them as runnable Python code. One thing I wanted to verify was whether the generated code actual

model-releasesr-ollama

I just built a mini Kimi-K3 from Scratch under 250$. Already beats GPT-2 (124M)!

I pre-trained a 1.02-billion-parameter on Kimi K3 replica trained on 5.00 billion decontaminated tokens for $250. This model has 1.02 billion parameters, of which 145 million are active per token. It

Jailbreaking in the Haystack

arXiv:2511.04707v2 Announce Type: replace-cross Abstract: Recent advances in long-context language models (LMs) have enabled million-token inputs, expanding their capabilities across complex tasks lik

Latent Space Refusal Anchoring for Low-Resource African Languages: Mechanistic Safety Recovery Without Retraining

arXiv:2608.18089v1 Announce Type: cross Abstract: Instruction-tuned models often refuse harmful requests in English but comply with the same requests in Yoruba, Igbo, Igala, and Hausa. This suggests t

Multimodal Rapport Estimation in Real-World HRI

arXiv:2608.18401v1 Announce Type: cross Abstract: Evaluating interaction quality in real-world HRI is an important challenge. If interaction quality can be estimated reliably, the results can be used

Persona-Guided LLM Agents for Task-Oriented Dialogue

arXiv:2608.18085v1 Announce Type: new Abstract: Prior work has shown that large language models (LLMs) can express diverse personality traits in open-ended text generation. However, it remains unclear

Position: Profiling Game Worlds by Transition Complexity

arXiv:2608.18079v1 Announce Type: new Abstract: Game world modeling (GWM) and reinforcement learning (RL) are often confounded because research papers rarely quantify how difficult the underlying tran

Same Facts, Different Updates: Inference Setup Shapes LLM Behavior in Medical Allocation

arXiv:2608.18108v1 Announce Type: cross Abstract: Large language models are being incorporated into sensitive and important decision-making processes across nearly all fields. While prior work studies

safetyarxiv-cs-ai

Task-Conditioned Least-Privilege Learning for Executable Terminal and MCP Agents

arXiv:2608.18351v1 Announce Type: cross Abstract: Tool-using large language-model agents can complete a task while exercising authority that the user did not grant or the task does not need, causing e

VTONQA: A Multi-Dimensional Quality Assessment Dataset for Virtual Try-on

arXiv:2601.02945v2 Announce Type: replace Abstract: With the rapid development of e-commerce and digital fashion, image-based virtual try-on (VTON) has attracted increasing attention. However, existin

Certified but Private: Scalable Zero-Knowledge Proofs for Neural Network Guarantees

arXiv:2608.17070v1 Announce Type: new Abstract: With the growing deployment of machine learning models, formal guarantees of the robustness and fairness of these models have become increasingly import

safetyarxiv-cs-lg

CKAA: Cross-subspace Knowledge Alignment and Aggregation for Robust Continual Learning

arXiv:2507.09471v2 Announce Type: replace Abstract: Continual Learning (CL) empowers AI models to continuously learn from sequential task streams. Recently, parameter-efficient fine-tuning (PEFT)-base

Encoded but Not Actionable: Auditing the Decode-Generate-Steer Gap in Frozen LLMs for Geometric Constraints

arXiv:2608.17843v1 Announce Type: cross Abstract: Large language models (LLMs) have demonstrated strong performance on structured reasoning tasks, but what they encode and whether it informs model beh

researcharxiv-cs-ai

Foundation Agents Meet Agentic Deep Research: Evidence-Grounded Clinical Code Forecasting

arXiv:2608.17075v1 Announce Type: cross Abstract: Next-encounter ICD forecasting predicts which standardized diagnosis codes will be documented at a future visit from the longitudinal record available

Local Ai

Post-Train NVIDIA Cosmos 3 Edge for On-Device Robot Control

NVIDIA’s Cosmos 3 Edge is a 4‑billion‑parameter omni‑model (with a 2‑billion‑parameter Nemotron reasoner) that can run on an NVIDIA Jetson Thor, providing on‑device policy inference for robot manipula

local-ainvidia-developer

Local Ai

Predicting Male Domestic Violence Using Explainable Ensemble Learning and Exploratory Data Analysis

arXiv:2403.15594v4 Announce Type: replace-cross Abstract: Domestic violence is commonly viewed as a gendered issue that primarily affects women, which tends to leave male victims largely overlooked. T

local-aiarxiv-cs-lg

Replit Free Mode, powered by @OpenAI GPT-5.6 Luna. Let’s make intelligence accessible to everyone.

Replit has introduced a free mode powered by OpenAI’s GPT‑5.6 Luna model, announced on 19 August 2026. The service offers real‑time AI assistance for coding and collaboration at no cost, aiming to bro

model-releasesopenai--x

SE-MoLoRA: Shared-Expert LoRA Adapters for Domain-Specific Photographic Assessment

arXiv:2608.17514v1 Announce Type: new Abstract: Vision-language models can describe images fluently, but they often fail to provide actionable photographic critique because semantic content and aesthe

Local Ai

Seeing is Free, Speaking is Not: Uncovering the True Energy Bottleneck in Edge VLM Inference

arXiv:2607.09520v2 Announce Type: replace-cross Abstract: Vision-Language Models (VLMs) are the perceptual backbone of embodied AI, but their energy footprint on edge hardware remains poorly understoo

local-aiarxiv-cs-ai

StartupBench: Benchmarking General-Purpose Agents on Market-Validated End-to-End Workflows

arXiv:2608.17800v1 Announce Type: new Abstract: Recent advances in Large Language Models(LLMs) and agents have substantially improved the ability of AI systems to execute complex tasks. Yet existing b

Stop Anthropomorphisizing Intermediate Tokens: Qwen3.8 doesn't 'overthink'

Intermediate tokens, called 'thinking' or 'reasoning' actually are nothing like it. Humans do step-by-step reasoning leading to the conclusion. LLMs use intermediate traces to augment their prompt. Th

Applications

TabNSM: Neural Sparse Mixer for Tabular Regression

arXiv:2608.18026v1 Announce Type: new Abstract: Large-scale, high-dimensional tabular regression remains challenging: tree-based models are robust but lack end-to-end representation learning, while de

applicationsarxiv-cs-lg

The Emergence of Lab-Driven Alignment Signatures: A Psychometric Framework for Auditing Latent Bias and Compounding Risk in Generative AI

arXiv:2602.17127v2 Announce Type: replace Abstract: Large language models increasingly serve as reasoning layers in multi-agent systems, where one provider's models may generate, judge, and summarize

safetyarxiv-cs-cl

AA is the reason for Qwen3.8 27B shipped with xhigh

I know why Qwen3.8 27B shipped with xhigh reasoning as default, it's to do its best in benchmarks. Models from top labs often get benchmarked at multiple reasoning levels, but that same treatment does

ChainSpace: A Chained-Reasoning Paradigm for Spatial Intelligence

arXiv:2608.15788v1 Announce Type: new Abstract: Spatial intelligence requires foundation models to maintain coherent spatial state across interactions with the physical world. However, existing data-c

Do Uncertainty Signals Help? A Systematic Study of Uncertainty-Aware Decoding with Rollback Mechanisms

arXiv:2608.14653v1 Announce Type: cross Abstract: Prediction uncertainty is a widely adopted metric for quantifying model confidence, with downstream applications spanning model explanation, data sele

researcharxiv-cs-ai

Does a Tool Result Carry More Authority Than Plain Text? Three Prospective Studies of False-Claim Adoption in a Synthetic Assignment Task with Claude Opus 5

arXiv:2608.14992v1 Announce Type: new Abstract: Language-model systems increasingly read from stores they also write to, so a claim that was merely written earlier can return looking retrieved. We tes

Does the LM Head Create a Harmful Gradient Bottleneck? A Causal Test

arXiv:2608.16671v1 Announce Type: new Abstract: The language-model head maps a hidden state of width D to a vocabulary of size V, so its transpose can return at most D independent directions to the Tr

Don't ignore llama.cpp RPC with old hardware. Results of a 5070 Ti and 1080 Ti over gigabit ethernet: it's actually functional.

Results up front: I had to prioritize prefill or token generation - there was no happy medium. Using UD-Q4_K_XL, q8 kv cache, and 96k max context: focus on generation (MTP = 2): 350 pp and 36 tg @ 12k

Enhancing the Non-Functional Quality Compliance of LLM-Generated Code through Quality-Aware Preference Learning

arXiv:2503.09020v3 Announce Type: replace-cross Abstract: Large Language Models (LLMs) have been widely adopted in commercial code completion engines, significantly enhancing coding efficiency and pro

FirstDiff: One-Step Diffusion-Based Anomaly Detection for Multivariate Time Series via Initial Noise Prediction

arXiv:2608.15727v1 Announce Type: cross Abstract: Diffusion models have recently shown strong potential for multivariate time-series anomaly detection by learning the distribution of normal data throu

Forward Pass Domain Adaptation (Without Cross-Layer Backpropagation)

arXiv:2608.14563v1 Announce Type: cross Abstract: Forward-Pass-Only MLP training (FPO) adapts large language models without a backward pass through the model body, achieving 2.7--3.2x the throughput o

researcharxiv-cs-ai

Gathered, Not Admitted: How Attention Brings a Latent Variable into Verbalizable Form

arXiv:2608.15022v1 Announce Type: new Abstract: Language models hold latent quantities in a form they can report on, and more of a quantity is present in that form when the task requires reusing it fl

Tutorials

Human Pose Estimation in Trampoline Gymnastics: How to Improve Performance on Extreme Poses

arXiv:2604.01322v2 Announce Type: replace Abstract: Trampoline gymnastics involves extreme human poses and uncommon viewpoints, on which state-of-the art pose estimation models tend to under-perform.

tutorialsarxiv-cs-cv

HyMem: Hierarchical Context Management for Long-Horizon Agents via Information Isolation

arXiv:2608.15703v1 Announce Type: new Abstract: Large language model (LLM) agents often perform poorly on complex, long-horizon tasks because their context becomes increasingly cluttered over time. As

Improving Influence-based Instruction Tuning Data Selection for Balanced Learning of Diverse Capabilities

arXiv:2501.12147v2 Announce Type: replace-cross Abstract: Selecting appropriate training data is crucial for instruction fine-tuning of large language models (LLMs), which aims to (1) elicit strong ca