Search: “benchmark”

Hi, After a long night of optimizations, I believe I have made the fastest inference engine for Qwen3.6-28B on a 3090. Quick metrics: - 250w power capped - Up to 195k context (ships with 150k for safe

local-air-localllama

16 Aug 2026

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Show-off Saturday: Intel Arc B140 build.

DGX agent

My local inference build, with: ASUS WS C422 PRO/SE 10-core Xeon W-2255 64GB ECC RAM 64GB VRAM Pimped case with TurboLEDz indicating the frequencies of the 10 xeon cores. Running llama.cpp with SYCL b

model-releasesr-localllama

16 Aug 2026

Agentic harness for small models

Hello; I'm a semi-beginner at local AI. I've been experimenting with this tech for a while, and I still haven't found a proper harness that fits my models, hardware, and needs. My use case is pretty s

15 Aug 2026

Our 'agentic plus' extractor in LlamaParse is great for extracting out massive volumes of fields (e.g. 10k-100k+ fields) from long documents…

model-releasesjerry-liu--x

Our 'agentic plus' extractor in LlamaParse is great for extracting out massive volumes of fields (e.g. 10k-100k+ fields) from long documents (100-500 pages) We tested with a doc that contains a giant

15 Aug 2026

Qwen 3.8 - 27B is a game changer

So a bit of context, I am a cybersecurity senior analyst I am interested in LLMs for that field especially with MCPs to connect them to the tools or for writing scripts I started this field by doing a

15 Aug 2026

A 27B agent just beat Claude Opus 4.8 and GPT-5.5 on held-out research replication. Replica turns paper replication into a scalable RL task …

A 27B agent just beat Claude Opus 4.8 and GPT-5.5 on held-out research replication. Replica turns paper replication into a scalable RL task space. Replicating a paper forces the same hypothesis-driven

model-releasesdair-ai--x

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

Research

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

Can Vision-Language Models Assess Proxemic Risk from Egocentric Robot Images?

arXiv:2608.12515v1 Announce Type: new Abstract: Assessing proxemic danger from a robot's egocentric perspective is critical for safe embodied navigation in human environments and requires both visual

DFM Mimir v1: An Open HRM Delivering Frontier Performance at 1B Parameters Using Only Permissible Post-Training Data

arXiv:2608.13517v1 Announce Type: cross Abstract: Current large language model development relies on massive, often non-permissible datasets, creating a high barrier for researchers committed to open-

Do LLMs Beat Nash? Testing Decentralized Coordination in Self-Play Multi-Agent Games

model-releasesarxiv-cs-ro

arXiv:2608.12547v1 Announce Type: cross Abstract: Large language model agents deployed without a central controller are often assumed to require communication to coordinate their actions. We ask what

Local Ai

EgoCITE: Context-Augmented Indexing and Time-Aware Retrieval for Long-Horizon Egocentric Memory

arXiv:2608.12627v1 Announce Type: cross Abstract: Long-horizon egocentric memory transforms continuous first-person video and audio into a searchable record of past experiences. We demonstrate two bot

local-aiarxiv-cs-ai

EvDiff: Event-Based Video Reconstruction using One-Step Diffusion Models

arXiv:2511.17492v2 Announce Type: replace Abstract: As neuromorphic sensors, event cameras asynchronously record changes in brightness as streams of sparse events with the advantages of high temporal

FastThaiG2P: Lightning-fast Thai Grapheme-to-phoneme Conversion for Voice Agent Pipelines

arXiv:2608.12814v1 Announce Type: new Abstract: FastThaiG2P provides sub-millisecond Thai grapheme-to-phoneme conversion for text-to-speech pipelines (International Phonetic Alphabet and Kokoro-TTS co

FluctlightDB: A Memory Model of Data for AI Agents

arXiv:2608.12365v1 Announce Type: cross Abstract: For fifty years, data systems have answered two questions. The relational model asked which records match a predicate; the vector model asked which ve

Follow the Norm: Accounting for Fine-Tuning and Prompt Effects on Model Rationales

arXiv:2608.13250v1 Announce Type: cross Abstract: Normative datasets are often used to train and align AI systems, but the norms they contain can function as action-guiding patterns rather than neutra

HybridRAG-BN: A Retrieval-Augmented Framework with Fine-Tuned Verification for Bangla KBQA

arXiv:2608.13004v1 Announce Type: new Abstract: Knowledge-base question answering (KBQA) systems rely on effective retrieval and reasoning mechanisms to generate accurate answers from external knowled

Interpretable Causal Discovery via Causal-Effect Constraints

arXiv:2608.12640v1 Announce Type: cross Abstract: Causal discovery aims to uncover the underlying causal relationships given data generated from a system. The goal, however, is not merely to predict c

Is this Citation on Point?

arXiv:2608.12571v1 Announce Type: cross Abstract: In 2023, a New York judge sanctioned two attorneys in Mata v. Avianca for filing a brief with hallucinated citations generated by ChatGPT. Such failur

Agents

Memorization Diagnostics for Code LLMs Should be Scale-Aware

arXiv:2608.12771v1 Announce Type: cross Abstract: The extent to which large language models for code rely on memorization over genuine understanding remains highly debated. While current literature fr

agentsarxiv-cs-ai

Reliability-Aware Sexism Detection: Combining DPO with Annotator Agreement and Token-Level Confidence Scoring

arXiv:2608.12330v1 Announce Type: new Abstract: The detection of online sexism remains an open problem. Sexism detection is inherently subjective, yet most existing systems reduce multi-annotator labe

Return of the local King! Qwen3.8-27B performance chart

Now we have it, a local Opus4.6 (Max) that runs on you local machine! The performance of Qwen3.8-27B blows Glimmer away, and a big leap from its own predecessor. Can't wait for GGUFs... Text: https://

Robust data-driven discovery of fractional differential equations via weak formulations and Pareto-based subset selection

model-releasesarxiv-cs-lg

arXiv:2608.12879v1 Announce Type: new Abstract: Fractional partial differential equations describe nonlocal dynamics, but discovering them from noisy data is difficult because fractional differentiati

Rules or Character? Scaling Laws for AI Safety Design

arXiv:2608.13345v1 Announce Type: new Abstract: Artificial Intelligence (AI) safety systems combine character shaping (e.g., Reinforcement Learning from Human Feedback [RLHF], Constitutional AI), whic

Self-Referential Induction Increases Response Instability Relative to Unresolvable and Verifiable Questions in Large Language Models

arXiv:2608.13258v1 Announce Type: cross Abstract: Self-referential prompting has been shown to reliably induce large language models to produce first-person reports resembling subjective experience, b

SkillShapley: Boundary-Adaptive Shapley Valuation for Skill Step Attribution in LLM Agents

arXiv:2608.13173v1 Announce Type: new Abstract: Agent skills are crucial external instructions that enable language agents to execute long procedural tasks such as coding or document processing. Exist

Small models, big real-world impact. Proud to see Qwen leading local inference in the State of Open Models. Enjoy the sunshine from Qwen.☀️ …

Small models, big real-world impact. Proud to see Qwen leading local inference in the State of Open Models. Enjoy the sunshine from Qwen.☀️ 😎 Appreciate your work! @huggingface The State of Open Model

model-releasesqwen--x

The Embedder's Dilemma: LLMs Are Better, but at What Cost?

arXiv:2608.12875v1 Announce Type: new Abstract: Should you replace your text-embedding pipeline with a large language model? We answer this with a controlled, cost-aware comparison of ten LLMs across

The Objective Is the Bottleneck: Latent World Models Encode What Their Planners Cannot Use

arXiv:2608.12959v1 Announce Type: cross Abstract: Latent world models are judged by how well they predict, so when planning fails at long horizons the natural reading is that the predictor degrades. O

Topology-Unified 2D Pose Estimation across Intact, Residual and Prosthetic Limbs

arXiv:2608.13047v1 Announce Type: new Abstract: Driven by the availability of large-scale datasets, Human Pose Estimation (HPE) plays a critical role in numerous downstream tasks. However, mainstream

Toward a Gricean Retreat: Probing LLMs for Knowledge Boundaries and Referent Specificity

arXiv:2608.13484v1 Announce Type: cross Abstract: When asked about entities outside their knowledge boundary, LLMs routinely fabricate plausible-sounding details rather than backing off to safer, more

Why Do AI Agents Break Rules? How Framing, Context, and Social Signals Shape Compliance

arXiv:2608.12323v1 Announce Type: cross Abstract: Specifying a penalty can paradoxically convert a legal obligation into a cost-benefit calculation that favors violation. We demonstrate that this enfo

AI4AI at Test-Time: Strong-to-Weak Capability Transfer via Harnesses

arXiv:2608.12307v1 Announce Type: cross Abstract: Recent work on distillation transfers the capabilities of large models to smaller ones often by updating the latter's parameters, through teacher forc

Research

Certifying What Helps Customer-Return Timing: A Screen-and-Confirm Test for Conditioning Signals, and Why Decay Is Nearly Enough

arXiv:2608.11555v1 Announce Type: new Abstract: Practitioners enrich customer-return models with ever more signals (lifetime value, category, recency/frequency, calendar, geography), and the temporal-

researcharxiv-cs-lg

Confucius4-TTS: Transcript-Free Cross-Lingual Zero-Shot TTS with a Learnable Speaker Encoder

arXiv:2608.11650v1 Announce Type: cross Abstract: Recent advances in zero-shot text-to-speech (TTS) have substantially improved speech quality and voice cloning fidelity. However, many zero-shot TTS s

ContactIPM: A Structure-Exploiting Interior-Point Solver for Contact-Implicit Trajectory Optimization

model-releasesarxiv-cs-ro

arXiv:2608.11731v1 Announce Type: new Abstract: Contact-implicit trajectory optimization avoids prescribing contact sequences, but yields mathematical programs with complementarity constraints (MPCCs)

Local Ai

Doom running on an LLM -- Hugging Face checkpoint included

There's no training anywhere in this. I ported Doom's actual rendering algorithm into transformer weights using a compiler I wrote (torchwright) -- every weight computed, none learned. The prompt carr

local-air-localllama

DS4 cloud (30 min) vs Qwen3.6 36B (2 min) vs Muse Glimmer 30B (3 min) on Llama.cpp (RTX 5080)

Some people told me that the difference in richness and layout between Glimmer and Qwen wasn't clear to them. This example makes it super clear. I'm aware that comparing Glimmer 30B (a dense model) wi

Evaluating LLM Generated Detection Rules in Cybersecurity

arXiv:2509.16749v1 Announce Type: cross Abstract: LLMs are increasingly pervasive in the security environment, with limited measures of their effectiveness, which limits trust and usefulness to securi

Federated Learning for the Design of Parametric Insurance Indices under Heterogeneous Renewable Production Losses

model-releasesarxiv-cs-lg

arXiv:2601.12178v2 Announce Type: replace Abstract: We propose a federated learning framework for the calibration of parametric insurance indices under heterogeneous renewable energy production losses

Gaze Target Estimation Anywhere with Concepts

arXiv:2608.11367v1 Announce Type: cross Abstract: Estimating human gaze targets from images in-the-wild is an important and formidable task. Existing approaches primarily employ brittle, multi-stage p

Safety

Generative Learning for Quantum Measurement Design

arXiv:2608.11396v1 Announce Type: cross Abstract: Extracting quantum information from a quantum state is a fundamental task of quantum computation, often requiring the estimation of many non-commuting

safetyarxiv-cs-lg

How do you plan to run Qwen3.8-2.4T-A95B locally?

To my fellow crazies, the few. Those who dared wrestle with llama-70b, mistral-large, goliath, mistral8x22B, DeepSeekV2/3, wept when llama4 behemoth was announced, picked yourself up and are now wrest

Qdrant and Minima Deliver 2.92x More Agentic RAG Tasks per GPU-Hour

Reducing Retrieval and Calls When a retrieval-augmented generation (RAG) agent runs, it often has to plan a search, check the evidence it gets back, and try again when that evidence falls short. Those

model-releasesqdrant

Retry, Switch, or Abstain? Learning Strategy-Aware Tool-Use Policies via Controlled Error Injection

arXiv:2608.11977v1 Announce Type: new Abstract: Tool-using LLM agents are commonly trained and evaluated in environments where tool calls succeed reliably, yet deployed tools can fail transiently, per

Rubric Dropout: A Simple Way to Mitigate Reward Hacking in Rubric-as-Reward RL

arXiv:2608.11669v1 Announce Type: cross Abstract: Reinforcement learning against rubrics, lists of criteria graded by an LLM judge, has become a standard way to post-train language models on tasks wit

SCOPE-Router: Cost-Aware Open-Set VLM Routing for Execution-Oriented Tasks

arXiv:2608.12127v1 Announce Type: new Abstract: Model routing aims to select the most suitable model from a candidate pool for each query, balancing quality and cost. Existing VLM routing research is

Self-Harness: Harnesses That Improve Themselves

arXiv:2606.09498v2 Announce Type: replace Abstract: The performance of LLM-based agents is jointly shaped by their base models and the harnesses that mediate their interaction with the environment. Be