Search: “benchmark”

Quantifying and Mitigating Korean Jamo-Level Typographical Vulnerabilities in Large Language Models

arXiv:2608.30229v1 Announce Type: new Abstract: Korean introduces an additional typographical perturbation level not captured by ordinary character-level edit models: because syllable blocks are inter

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Model Releases

Question: Why is prefill unbelievably faster in vLLM than other inference engines?

DGX agent

I only started using some vLLM forks recently in a 4 x 48GB 4090 system. DS4F - ~5000pp/180tg (DSpark) Qwen3.8 Flash next - ~7500pp/135tg (MTP) This is amazing, like having the API in my house. But it

model-releasesr-localllama

1 Sept 2026

Rad-R: A Raw-ADC Radar Dataset and Capture-Invariant SSM for Hardware-Fault Diagnosis

arXiv:2608.30896v1 Announce Type: new Abstract: Automotive mmWave radar can develop vibration, antenna misalignment, radome blockage, and receive-channel degradation that corrupt the signal before per

RENSA: Rich Environment Metadata to Navigate Shared and Distributed Endpoints for Automated Federated SPARQL Query Generation

arXiv:2608.28963v1 Announce Type: cross Abstract: The number of knowledge graph databases has increased significantly with the proliferation of knowledge graph technologies. Knowledge graphs enable th

Risk-Adjusted Harm Scoring for Automated Red Teaming for LLMs in Financial Services

arXiv:2603.10807v2 Announce Type: replace-cross Abstract: Existing LLM safety evaluations rely on binary attack-success rates and domain-agnostic taxonomies, leaving regulated Banking, Financial Servi

RSLM: Training-Free Vector Quantization for Approximate Nearest Neighbor Search

model-releasesarxiv-cs-lg

arXiv:2608.30384v1 Announce Type: new Abstract: By introducing RSLM (Rotated Scaled Lloyd-Max), a family of training-free vector quantization codecs compressing embeddings to 1--4 bits per dimension,

SafeAtlas-VL: Beyond Binary Multimodal Safety with Large-Scale Data and Guard Models

arXiv:2608.29098v1 Announce Type: new Abstract: Multimodal safety moderation requires distinguishing risks arising from visual content, user intent, and assistant behavior. Existing safeguards, howeve

Applications

SHAPE of Chain-of-Thought in Math Reasoning

arXiv:2608.28600v1 Announce Type: new Abstract: Large language models (LLMs) achieve strong performance on mathematical reasoning benchmarks, yet the mathematically meaningful skills underlying their

applicationsarxiv-cs-ai

SIR: Self-improving Red-teaming for Compute Use Agents

arXiv:2608.30207v1 Announce Type: cross Abstract: Computer use agents (CUAs) are vision-language models that perceive a screen and act on a real operating system through mouse, keyboard, and terminal,

SkillForge: Compositional Skill Synthesis with Verification-in-the-Loop for Generating Formally Verified Dafny Programs

arXiv:2608.29841v1 Announce Type: new Abstract: Generating formally verified programs from natural language remains challenging: existing approaches either produce code in a single pass without recour

There’s generally a massive difference in quality between specialized OCR providers, “simple” open-weight OCR models, and free/OSS solutions…

model-releasesjerry-liu--x

There’s generally a massive difference in quality between specialized OCR providers, “simple” open-weight OCR models, and free/OSS solutions. 1️⃣ Specialized OCR models (including LlamaParse) solve fo

Towards a Joint Khmer Text Recognition and Word Segmentation

arXiv:2608.30213v1 Announce Type: cross Abstract: Text recognition, or extracting electronic text from document images, has been indispensable for knowledge retrieval tasks, such as retrieval-augmente

Uncertainty-Aware Multimodal Anti-UAV Detection via Evidential Fusion and Conflict-Discounted Belief Aggregation

arXiv:2608.29235v1 Announce Type: new Abstract: Anti-UAV perception systems must remain reliable when sensor streams degrade under occlusion, fast motion, or modality-specific failure. Existing multim

Weaving Visual Narratives: Agentic Image Bundle Composition Beyond Atomic Visual Matching

arXiv:2608.28695v1 Announce Type: new Abstract: Image retrieval has traditionally been formulated as a point-wise matching problem, where each candidate image is scored in isolation. However, this ato

Research

What Does an Evaluation License? A Commit-Bound Census of Claim Replay in Inspect Evals

arXiv:2608.19269v4 Announce Type: replace-cross Abstract: Benchmarks can run without determining what their results license. We freeze a large evaluation collection and attempt to replay its historica

researcharxiv-cs-ai

Which current local models that can run within 128GB generate the best SVG pelicans?

I used a famous Simon Willison's pelican riding a bicycle prompt on the biggest local LLMs that can run on 128GB Apple Silicon. U used quantizations by Unsloth. Qwen3.8 Flash-Next gives a lot of detai

X-Coder: Advancing Competitive Programming with Synthetic Tasks, Solutions, and Tests

arXiv:2601.06953v3 Announce Type: replace Abstract: Competitive programming remains challenging for code LLMs. Despite recent progress, many training pipelines still depend on scarce real-world data,

3D MRI-Based Alzheimer's Disease Classification Using Multi-Modal 3D CNN with Leakage-Aware Subject-Level Evaluation

arXiv:2603.17304v2 Announce Type: replace Abstract: Deep learning has become an important tool for Alzheimer's disease (AD) classification from structural MRI. Many existing studies analyze individual

Local Ai

Acquire, Repair, Preserve: A Diagnosis-Guided Post-Training Recipe for Small-Model Dialogue Game Agents

arXiv:2608.28458v1 Announce Type: new Abstract: Interactive dialogue games test a capability that static benchmarks largely leave implicit: a model must carry state across turns, interpret feedback, a

local-aiarxiv-cs-cl

Research

Are These Modules Worth Their Cost? A Paradigm-Level Accuracy-Cost Analysis of In-context Learning Text-to-SQL

arXiv:2608.28432v1 Announce Type: new Abstract: Recent advances in in-context learning (ICL) text-to-SQL have substantially improved execution accuracy on public benchmarks by assembling increasingly

researcharxiv-cs-cl

BEACON: Behavior-Anchored Cross-Source Knowledge Graph Construction for Cyber Threat Intelligence

arXiv:2608.28394v1 Announce Type: cross Abstract: Cyber threat intelligence (CTI) is foundational to modern cyber defense, yet much of it resides in unstructured reports whose volume and heterogeneity

Blind Men and the Elephant: Probing the Epistemic Myopia of LLMs under Long-Tail Divergent Knowledge

arXiv:2608.28478v1 Announce Type: new Abstract: Factual question answering (QA) typically assumes a single canonical answer, obscuring whether large language models (LLMs) retain divergent accounts of

CommerceVibe: Learning to Design E-Commerce Creatives as Executable Visual Code via Dual-Feedback Reinforcement Learning

arXiv:2608.27893v1 Announce Type: new Abstract: High-quality e-commerce creatives are essential for presenting products and conveying marketing messages. Recent diffusion models enable scalable creati

CUDA: extend MOE fusion to specdec, earlier MOE glu fusion and topk-router fusion were restricted to 1 token by ynankani · Pull Request #27621 · ggml-org/llama.cpp

I haven't had a chance to test it yet, but it looks very promising. It seems to speed up MTP for MoE models across different draft widths (especially greater than 1). Check the benchmarks. submitted b

FUSED: Forensic-Semantic Mixture-of-Experts for AI Inpainting Detection and Localization

arXiv:2608.28302v1 Announce Type: new Abstract: Diffusion-based inpainting models modify only a localized part of an image, while many AI-image detectors rely on global artifacts and do not localize.

LWiAI Podcast #255 - Gemini 3.7, Jalapeño, Qwen 3.8, Drones

model-releaseslast-week-in-ai

The LWiAI Podcast #255 (August 26 2026) reviewed recent AI alerts, including Google’s Gemini 3.7 Flash, Jalapeño’s record‑speed benchmark results, Qwen 3.8, and a case of an entirely AI‑guided drone t

New beellama fork 76% faster tg with kvarn KV quants

When using kvarn quants at low context depth tg speed is similar to llama.cpp on and equivalent qx_x quant. However, as context depth grows kvarn tanks your tg speed. This fork optimises kvarn to have

Non-Uniform Quantisation for 3DGS Compression

arXiv:2608.28272v1 Announce Type: new Abstract: 3D Gaussian Splatting (3DGS) has emerged as a powerful technique for novel view synthesis, yet its high bitrate requirements pose significant challenges

One Model for All: Universal Pre-training for EEG based Emotion Recognition across Heterogeneous Datasets and Paradigms

model-releasesarxiv-cs-lg

arXiv:2511.08444v2 Announce Type: replace Abstract: EEG-based emotion recognition is hampered by profound dataset heterogeneity (channel/subject variability), hindering generalizable models. Existing

Local Ai

Optimal 1.25 bit quantization of Qwen3.8-Flash-Next

Hello! I was looking into quantizing models and i saw how Hy4 was shrunk from 1.5 TB to 200GB with high retention in benchmarks (98% i think). I was wondering if: a) it would be worth it to attempt th

local-air-localllama

Sledgehammer or Scalpel? A Fine-grained Adaptive Framework for Implicit Hate Speech

arXiv:2608.27462v1 Announce Type: new Abstract: Unlike explicit attacks with obvious profanity, implicit hate speech hides malice within seemingly compliant expressions through metaphors and contextua

The Chrono Trigger plot challenge - Crono awakens in his modest bedroom of 2095...

A hallucinated event, Crono awakens in his modest bedroom of 2095... I am using local models since 2025 January. My daily driver is qwen 3.6 35B A3B, which works pretty well for coding tasks, but I al

FlashMLA sm_120 kernel build with 2-3x performance increase from SPDA

Hey guys I have been working on a open source LLM development repo. Anyone can train any amount of parameter against any arch (GQA, MLA, Dense...) When I tried to implement MLA I read the technical re

30 Aug 2026

Here my pretty good qwen3.8 27B setup, hope it helps

Since I spent the time to figure it out and it is not like it will make me any money ever. I think I'd share with you all what I managed to cobble together. First and foremost here the relevant bits a

30 Aug 2026

Qwen3.8-Flash-Next optimised for Macs

Running on a M1 Max 64 GB: - SSD streaming for tensors - SSD streaming for engrams - SSD streaming for MTP How is it possible? * Custom Q4 quant: benchmarked all metal tensors then picked and spliced

30 Aug 2026

AtomicChat/Qwen3.8-Flash-Next-GGUF is Really Good

specs hardware: M4 Max 128GB Studio inference engine: llama.cpp (qwen4exp branch) judge: claude-opus-4-6 AtomicChat/Qwen3.8-Flash-Next-GGUF Qwen3.8-Flash-Next is a great model I benched in my previous

Is it possible to run it with a combined memory setup: 16 GB VRAM + 64 GB RAM + SSD for offloading n-grams?

Hardware: rtx 5080 16 gb vram; 64 gb ram ddr5 6000hz; ssd with unlimited memory; ryzen 7 9800 x3d. OS: Windows 11 Software: I’d prefer llama.cpp, but it’s not a strict requirement; I’ll use whatever y

Qwen3.8 Flash Quants

~20–30GB smaller than Unsloth/AesSedai Q4 at similar PPL. After several days of testing I released a set of mainline-compatible imatrix quants for Qwen3.8-Flash-Next. Goal: same quality band as the po

Terminal Bench 4.0 just dropped, GLM-5.3 is at the same level as Fable 5, accounting for margin of error

Announcement: https://www.tbench.ai/news/terminal-bench-4-0 Leaderboard: https://www.tbench.ai/ Imo the best aspect in their announcement is their focus on rapidly iterating on TerminalBench to keep t

This paper does a good job of showing the promise and gaps of autonomous AI scientists. Big question is how much more advanced models close …

model-releasesethan-mollick--x

This paper does a good job of showing the promise and gaps of autonomous AI scientists. Big question is how much more advanced models close those gaps. Today, we're excited to share early progress on

A Very Big Video Reasoning Suite

arXiv:2602.20159v3 Announce Type: replace-cross Abstract: Rapid progress in video models has largely focused on visual quality, leaving their reasoning capabilities underexplored. Video reasoning grou

Agent Seer: Synthesizing Scenarios from Specification Understanding

arXiv:2608.26133v1 Announce Type: new Abstract: Evaluating AI agents that use external tools requires realistic test scenarios that capture how practitioners compose tools and iterate across conversat

Agents Don't Paginate: First-Chunk Selection for LLM Tool Responses

arXiv:2608.26130v1 Announce Type: new Abstract: Coding agents built on large language models (LLMs), such as Claude Code, Cursor, OpenAI Codex, GitHub Copilot, and Aider, receive tool responses that r

Anatomy-Guided Foundation Model Adaptation with Within-Case Prototype Supervision for Standard Plane Detection in Fetal Ultrasound Blind Sweeps

arXiv:2608.27051v1 Announce Type: new Abstract: Detecting the fetal abdominal circumference standard plane in low-cost obstetric blind sweeps is a highly imbalanced frame-classification problem: posit

Assessing mentalization in humans and large language models

arXiv:2608.26291v1 Announce Type: new Abstract: Mentalization - the ability to infer others' beliefs and intentions to guide one's own choices - is a key cognitive function underlying human social int

DRL: A Deterministic Relational Middleware Layer for Transaction-Safe Enterprise NL2SQL Under Schema-Graph Scaling

arXiv:2608.26172v1 Announce Type: cross Abstract: Deploying natural-language interfaces over enterprise OLTP catalogs fails at scale because semantic parsers collapse under schema-graph scaling, infla

Agents

GLM-5.3 has arrived on Together AI on Day 0 @Zai_org pushed the GLM-5.2 base further with scaled post-training, delivering stronger coding a…

GLM-5.3 has arrived on Together AI on Day 0 @Zai_org pushed the GLM-5.2 base further with scaled post-training, delivering stronger coding and long-horizon agent performance. Across benchmarks, it nea

agentstogether-ai--x

GRAIN: Bridging Name and Narrative Shifts in Real-World Graph Reasoning through Invariance-Rewarded Agentic RL

arXiv:2608.27142v1 Announce Type: new Abstract: Despite their potential in standardized graph tasks, Large Language Models (LLMs) remain brittle to real-world shifts in node identifiers and task formu