Search: “models”

Running DeepSeek V4 Flash Q4_K_XL at ~100 tok/s prompt processing on 4× RTX 3060 12GB

I managed to run the 143–144 GiB DeepSeek-V4-Flash-0731 UD-Q4_K_XL GGUF on four RTX 3060 12GB cards while keeping a 360k–376k context window. Hardware: CPU: Intel Core i9-10920X, 12C/24T RAM: 128 GB D

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Model Releases

S2-MoE: Enabling Efficient Self-Speculative Decoding for Mixture-of-Experts on Edge Devices

DGX agent

arXiv:2608.15018v1 Announce Type: new Abstract: Deploying large language models (LLMs) for inference on edge devices is challenging due to severe memory and bandwidth constraints. While speculative de

model-releasesarxiv-cs-ai

18 Aug 2026

Score Attack: A Lower Bound Technique for Optimal Differentially Private Learning

model-releasesarxiv-cs-lg

arXiv:2303.07152v3 Announce Type: replace-cross Abstract: Achieving optimal statistical performance while ensuring the privacy of personal data is a challenging yet crucial objective in modern data an

Securing AI-Generated Code: A Just-in-Time Vulnerability Detection and Remediation Pipeline

arXiv:2608.16187v1 Announce Type: cross Abstract: AI-assisted development tools generate vulnerable code at significant rates, yet few automated mechanisms exist to detect, enrich, fix, and verify sec

Sharing a new way to work with Stable Audio

model-releasesstability-ai

Stable Audio 3.0 now offers a DAW‑plugin that brings real‑time audio generation directly into your favorite digital audio workstation, letting you arrange, edit, and mix generated tracks like any othe

StateM: Reaching 95.3% Raw Accuracy, or a $15 Frontier Run, on Terminal-Bench 2.1 via Harness Scaling

arXiv:2608.15089v1 Announce Type: new Abstract: Long-horizon agents can fail even when their underlying models can solve the constituent steps. They may lose track of mutable state, fail to reactivate

Sterilizable Scene Graph Generation for Operating Rooms

arXiv:2608.16469v1 Announce Type: new Abstract: Scene graph generation from surgical video enables a holistic and structured understanding of surgical scenes by modeling objects and their semantic rel

SubZero+: Efficient Zeroth-Order LLM Fine-Tuning via Large Learning Rates

model-releasesarxiv-cs-lg

arXiv:2608.15665v1 Announce Type: new Abstract: Zeroth-order (ZO) optimization enables backpropagation-free fine-tuning of large language models, but existing ZO methods suffer from high-variance grad

Tutorials

SUGFW+: An Uncertainty-guided Feature Weighting Framework for Cold Start Active Adaptation of SAM in Medical Image Segmentation

arXiv:2608.16110v1 Announce Type: new Abstract: Cold Start Active Learning (CSAL) is important in improving the performance of a medical image segmentation model with low annotation budget by querying

tutorialsarxiv-cs-cv

TDD-Agent: Test-Driven Reasoning for Code Generation

arXiv:2608.16742v1 Announce Type: cross Abstract: Large Language Models (LLMs) have achieved remarkable progress in code generation, yet ensuring correctness in complex, repository-level tasks remains

The Commercial Tax: Rent-vs-Own Blind Spots in Multi-Hop Retrieval Benchmarks

model-releasesarxiv-cs-cl

arXiv:2608.16096v1 Announce Type: cross Abstract: Enterprises connect language models to their own data through retrieval. The benchmarks that rank multi-hop retrieval systems leave out two facts a bu

Agents

Topological Attribution Distance (TAD): Revealing Segment-Level RAG Influence on LLM Output Geometry for Incident Log Analysis

arXiv:2608.16775v1 Announce Type: cross Abstract: Large Language Models (LLMs) are increasingly being deployed in cybersecurity operations to assist cybersecurity analysts with rapid decision-making a

agentsarxiv-cs-ai

TRACE-CASH: Trial-History-Conditioned Reinforcement Learning for Adaptive Configuration Exploration in Time-Series CASH

arXiv:2608.16410v1 Announce Type: new Abstract: Combined algorithm selection and hyperparameter optimization (CASH) searches a conditional space in which the selected model determines which hyperparam

local-aiarxiv-cs-lg

Unadapted Multilingual ASR on a Garrusi Kurdish Evaluation Set: A Common-Reference Staged Normalization Analysis

model-releasesarxiv-cs-cl

arXiv:2608.16379v1 Announce Type: new Abstract: Evaluating speech recognition for a Kurdish variety written in a Latin field orthography, using a model that outputs Arabic script, creates a measuremen

ViTaR: Visuo-Tactile Residual Adaptation for Foundation VLA Manipulation

model-releasesarxiv-cs-ro

arXiv:2608.15816v1 Announce Type: new Abstract: As Vision-Language-Action (VLA) models scale toward real-world deployment, contact-rich manipulation exposes a critical blind spot: these policies encod

When Do Explanations Help In-Context Learning? A Comparative Study of Natural Language Explanation Types and Faithfulness

arXiv:2608.16627v1 Announce Type: cross Abstract: Natural language explanations (NLEs) are increasingly used as inputs, for example, as few-shot rationales that influence model behavior in in-context

safetyarxiv-cs-ai

When Less Is Enough: Context Selection and Prompting Strategies for Bengali News Headline Generation

model-releasesarxiv-cs-cl

arXiv:2608.15879v1 Announce Type: new Abstract: Large language models (LLMs) have shown strong performance in text generation tasks, yet their effectiveness on headline generation remains sensitive to

Applications

Adaptive Stopping for Multi-Turn LLM Reasoning

arXiv:2604.01413v3 Announce Type: replace-cross Abstract: Large Language Models (LLMs) increasingly rely on multi-turn reasoning and interaction, such as adaptive retrieval-augmented generation (RAG)

applicationsarxiv-cs-ai

Attention Capture Is Not Detection: A Two-Stage Account of How Humans Miss Localized AI Image Edits

arXiv:2608.13865v1 Announce Type: new Abstract: As AI-generated image edits proliferate, the platforms meant to curb the resulting disinformation treat detectability as a single, undifferentiated prop

Buy the Rumor, Sell the News: When Is News Priced In?

arXiv:2608.14014v1 Announce Type: new Abstract: Two old market sayings hold that news is already priced in by the time it is published, and that the rumor is bought while the news is sold. Both place

CForce: Boosting Parallel Decoding for dLLMs via Consistency Forcing

arXiv:2608.13925v1 Announce Type: cross Abstract: Diffusion large language models (dLLMs) accelerate language generation by predicting multiple masks in a single forward pass. However, existing dLLMs

safetyarxiv-cs-ai

CVT-Bench: Probing Spatial-State Integrity through Counterfactual Viewpoint Transformations

arXiv:2603.21114v2 Announce Type: replace Abstract: Multimodal large language models (MLLMs) perform strongly on isolated spatial tasks, but whether their predictions remain persistent and mutually co

Evaluating Agentic Learning Harness Capabilities Without Labels via the Scaling Hypothesis

arXiv:2608.13608v1 Announce Type: new Abstract: Agentic 'Continual Learning Harnesses', systems that pair an LLM with retrieval or memory to improve from feedback without retraining, have shown growin

LightTeaNet: A Weakly Supervised Lightweight CNN for Multi-Label Tea Leaf Disease Detection and Localization

arXiv:2608.14178v1 Announce Type: new Abstract: Tea is known as an important crop in many parts of South and Southeast Asia, yet the production of tea is still hampered by the multiple diseases that d

Ling 3.0 support merged into llama.cpp

Support for the new ling 3.0 models has been merged into llama.cpp: https://github.com/ggml-org/llama.cpp/pull/26608#event-29549472828 Ling tiny 8b1b - https://huggingface.co/inclusionAI/Ling-3.0-tiny

MACS: A Hybrid Multi-Agent Framework for Reliable Conversational E-Commerce Recommendation

arXiv:2608.14068v1 Announce Type: cross Abstract: Conversational recommendation for e-commerce is increasingly mediated by large language models (LLMs), yet many real-world deployments operate under a

Not All Tokens Are Equal: Inflation-Aware Routing for Agentic LLM Systems

arXiv:2608.13571v1 Announce Type: cross Abstract: When a language model fails to answer a query on the first attempt, an agentic system retries, consuming additional tokens each time. This retry overh

local-aiarxiv-cs-ai

On the Brittleness of Maximum Likelihood Estimation for Gaussian Process Hyperparameter Optimization

arXiv:2608.13793v1 Announce Type: cross Abstract: Machine learning (ML) has become an indispensable part of modern engineering design workflows. A crucial step in training an ML model is the selection

researcharxiv-cs-lg

Responsiveness Verification: Will Predictions Change? How Much? How Often?

arXiv:2507.02169v2 Announce Type: replace Abstract: Machine learning models are often used in applications where their inputs change due to routine interactions, strategic manipulation, or noise. In s

safetyarxiv-cs-lg

Unpopular opinion : Qwen 3.8 27b is not an overthinker

Yes it uses a ton more reasoning tokens than 3.6 did But test in on the same tasks with the other chinese models, glm 5.3, deepseek v4 flash and pro, etc it's really similar, and they are needed The r

Wrong but Useful: Trajectory Value Beyond Answer Correctness in Multi-Agent Messages

arXiv:2608.14375v1 Announce Type: new Abstract: Multi-agent reasoning systems often use agreement, confidence, or automated scores to decide which messages should shape a final answer. Such filtering

Dear Dario, 1. If Claude can cure cancer to save people like your dad, why should we 'pace the progress'? Does that mean more people with He…

model-releasesyann-lecun--x

Dear Dario, 1. If Claude can cure cancer to save people like your dad, why should we 'pace the progress'? Does that mean more people with Hepatitis C will die? 2. If Fable is so cyber-capable that it

Qwen 3.8 2.4T at 288k tokens/s on Nvidia GB300 NVL72

https://developer.nvidia.com/blog/serve-qwen3-8-2-4t-a95b-a-2-4t-parameter-model-with-configurable-reasoning-on-nvidia-gb300-nvl72/ 4k tokens per second per GPU of which there are 72. 350 tokens per s

Qwen3.8 27B reasoning effort low/medium/xhigh comparison

I did a short test of the different reasoning efforts, since on default xhigh the model thinks a lot. Not very scientific, just a quick 'generate an SVG of a pelican on a bicycle' prompt with 3 differ

There are two separate questions regarding Dario Amodei’s post about AI and biology: motivation and reality. Motivation. Dario started in bi…

model-releasesyann-lecun--x

There are two separate questions regarding Dario Amodei’s post about AI and biology: motivation and reality. Motivation. Dario started in biology (was a PhD student of the great Bill Bialek). It might

Qwen3.8-27B is now part of your everyday life — from smartphones to vehicles. ⚡️🚀Appreciate your work! @MediaTek

Qwen3.8-27B is now part of your everyday life — from smartphones to vehicles. ⚡️🚀Appreciate your work! @MediaTek Congratulations to the @Alibaba_Qwen on the launch of Qwen 3.8! MediaTek continues our

model-releasesqwen--x

15 Aug 2026

The new DeepSeek-V4-Pro (0813) is now fully rolled out on Ollama's cloud and included in Pro and Max subscriptions. ollama run deepseek-v4-p…

The new DeepSeek-V4-Pro (0813) is now fully rolled out on Ollama's cloud and included in Pro and Max subscriptions. ollama run deepseek-v4-pro:cloud Hosted in the US with Zero Data Retention (ZDR) and

model-releasesollama--x

15 Aug 2026

A Generative Approach for Improving Multi-Label Defect Classification in Photovoltaic Modules

arXiv:2608.12725v1 Announce Type: new Abstract: This paper addresses the challenge of multi-label defect classification in electroluminescence (EL) images of photovoltaic (PV) cells. Training models o

Are You Sure You're Sure? On the Impact of Instruction Tuning on Confidence and Lexical Diversity

arXiv:2608.13430v1 Announce Type: cross Abstract: Instruction-tuned language models achieve strong performance across a range of generation tasks, but have also recently been shown to exhibit verbaliz

researcharxiv-cs-ai

AutoDesign: Meta-Harness Optimization for Long-Horizon Agentic Design

arXiv:2608.13560v1 Announce Type: cross Abstract: Transforming multimodal sources into condensed and structured media outputs can be fundamentally conceptualized as a long-horizon agentic process cent

CoMedBench: A Multi-Source Benchmark of Synthetic Medical Data Fidelity and Downstream Utility

model-releasesarxiv-cs-lg

arXiv:2608.12805v1 Announce Type: new Abstract: Access to clinical data is essential for developing reliable healthcare machine learning systems, but direct use of electronic health records is constra

Dead text or binding clause? Measuring and restoring constraint influence in black-box LLM dialogues

arXiv:2608.12599v1 Announce Type: new Abstract: Multi-turn dialogues let users revoke constraints as easily as impose them, but revocation does not reliably take effect: models keep enacting withdrawn

researcharxiv-cs-ai

EEG-PRIME: Prototype-Aligned Representation Learning with Multi-Level Conditioning for EEG Decoding

arXiv:2608.13072v1 Announce Type: new Abstract: Electroencephalography (EEG) decoding models often generalize poorly across datasets and subjects due to domain shifts in acquisition protocols and indi

researcharxiv-cs-ai

From Local Mismatch to Global Impact: Optimizing Cache Reuse Policy for Efficient Diffusion

arXiv:2608.13043v1 Announce Type: new Abstract: Diffusion models have achieved dominant performance in visual generation but suffer from substantial inference overhead. While cache-based acceleration

local-aiarxiv-cs-ai

Geometric and Behavioral Stratification in Transformer Residual Streams

arXiv:2608.12447v1 Announce Type: cross Abstract: Trained transformer models develop privileged bases: coordinate axes whose statistics differ from the rest of the residual stream. But what kind of di

researcharxiv-cs-cl

Applications

Incremental Evaluation and Training in Relational Deep Learning

arXiv:2608.13023v1 Announce Type: new Abstract: Relational Deep Learning (RDL) models multi-tabular databases as temporal heterogeneous graphs to enable end-to-end representation learning. However, pr

applicationsarxiv-cs-lg

It's been a pleasure to partner with the @Alibaba_Qwen team.

On LinkedIn, the user @ollama announced that it has partnered with Alibaba's Qwen team, making the Qwen 3.8‑27B language model available on the Ollama platform. The post highlights that the model cons

local-aiollama--x

Jagged Judges: Epistemic Stability Under Silence, Pressure, and Persistence

arXiv:2608.12645v1 Announce Type: new Abstract: LLM judges have become central infrastructure for model evaluations, online grading, and reward modeling. Judges are typically validated by accuracy on

safetyarxiv-cs-ai