Search: “context”

Physical-State-Guided Diffusion Sampling for Full-Waveform Inversion

arXiv:2609.12899v1 Announce Type: new Abstract: Full waveform inversion (FWI) estimates subsurface velocity from seismic recordings, but its ill-posedness and nonlinearity make accurate reconstruction

researcharxiv-cs-lg

14 Sept 2026

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Research

Predicting Collision Cross Sections with GRACE: Geometric Residual Adduct Conditioning via Early-fusion

DGX agent

arXiv:2609.12223v1 Announce Type: new Abstract: Collision cross section (CCS), derived from ion mobility mass spectrometry, is a common descriptor for molecular annotation. Prediction is challenging f

researcharxiv-cs-lg

14 Sept 2026

R9V Update: now ~100 tok/s in TG on Qwen3.8 Flash Next IQ4_XS on x2 R9700 + 128GB RAM. Fixed crashes with n-gram SSD streaming, improved diagnostics, plus pinned images. Q4_K_XL now supported, 50 tok/s TG.

Pushed out this new update, hopefully decreases the instances of crashes. I torture tested this one for ~12 hours after my fixes and found no instability. Q4 K XL needs more fine tuning, which I will

14 Sept 2026

Dear 24G owners, try VLLM you might be able to run Qwen3.8 27B INT4, 144K FP8 KV on RTX 3090 with better speed. (TLDR VLLM AOT)

VLLM Benchmark: Prefill, Prompt processing - avg, 871.93 tok/s (3 hours constant running xhigh) - 10K prompt, 1000.26 tok/s (16 runs) - 90K prompt, 743,59 tok/s (16 runs) Decode, tok gen - avg, 38.39

13 Sept 2026

Qwen3.8 flash next - untrained svg generation

> 'make an svg of a frog playing on a chello on the back of a whale with carribean island in the back.' interestingly the svg looks different in the OpenWebUi preview then when looked at in preview (o

13 Sept 2026

ROCm 7.2.1 + llama.cpp on mixed Vega GPUs (gfx900 Vega64 + gfx906 MI50) — got 52-57 tok/s on Gemma 3 27B Q4 with patched rocBLAS payloads [repo + Dockerfile]

I'm documenting a reproducible setup for running current llama.cpp on a mixed AMD Vega system with both: Radeon RX Vega 64 / Vega10 / gfx900 (8GB HBM2) Radeon Instinct MI50 (Vega20) / gfx906 (16GB HBM

model-releasesr-ollama

13 Sept 2026

2×RTX 3090 + EPYC box running qwen3.8-flash-next at ~38 tok/s

What I have: - CPU: EPYC 7551 (32c/64T, Zen 1) - Board: Supermicro H11SSL-i (SP3), Rev 2.0 - RAM: 128 GB DDR4-2133 (all 8 channels full) - GPU: 2x RTX 3090 (48 GB total, PCIe 3.0) - 1500 W PSU What I

Safety

Autonomy, Social Norms, and Alignment: Towards a Developmental Framework for Autonomous Artificial Agents

arXiv:2609.11660v1 Announce Type: new Abstract: In recent years, artificial intelligence has made extraordinary progress thanks to large-scale models capable of generalization and the generation of co

safetyarxiv-cs-ai

DeepSeek V4.1 Flash on 8× A40: ~40 tok/s Q2_K and ~32 tok/s Q4_K_M with TensorSharp

Hey everyone — I’m building TensorSharp, an open-source LLM inference engine. Here are the latest DeepSeek V4.1 Flash GGUF results using its native ggml_cuda backend. Setup: 8× NVIDIA A40, layer split

For those of you forced to only use open models from Western labs in production, what are you deploying?

First off, I know that GLM, Qwen, and DeepSeek absolutely dominate in terms of SOTA Open Source models, and that’s what I use in my personal projects and for school, however, I’m also responsible for

MAPLE: Memory-Augmented Planning with Language and Evolution

arXiv:2609.11636v1 Announce Type: new Abstract: Domain practitioners understand their business constraints but may lack operations-research expertise or dedicated support. LLM-based optimization agent

Memory Compression for High-Fanout Agent Sandboxes

arXiv:2609.11294v1 Announce Type: new Abstract: High-fanout agent workloads create a growing memory bottleneck because a single task may spawn many concurrent sandbox sessions. Yet these sandboxes are

terms.txt: A Consent and Compensation Protocol for Agentic Web Access

arXiv:2609.11152v1 Announce Type: cross Abstract: The open web ran on an unwritten bargain: sites admitted crawlers, and search engines sent visitors back. Public measurements show that bargain breaki

A Taxonomy of Architecture Options for Foundation Model-based Agents: Analysis and Decision Model

arXiv:2408.02920v2 Announce Type: replace-cross Abstract: The rapid advancement of AI technology has led to widespread applications of agent systems across various domains. However, the need for detai

Safety

An Autonomous GeoAI Agent for Arctic Eco-Navigation

arXiv:2609.09374v1 Announce Type: new Abstract: Arctic maritime navigation is becoming increasingly important as changing sea-ice conditions expand seasonal accessibility while simultaneously introduc

safetyarxiv-cs-ai

Any 12gb VRAM users out there?

Hi! I've been following this community for quite a while and have difficulty figuring out what to put on my 3080 12gb - I know Qwen 3.6 35B 3A was the go-to choice when it first came out, but I'm curi

Augustinian BabyLM: What Ostensive Definition Can and Cannot Teach a Small Language Model

arXiv:2609.11870v1 Announce Type: new Abstract: A language model normally begins training with random word embeddings: whatever 'banana' means must be learned from training corpora. I implement St. Au

City Editing: Hierarchical Agentic Execution for Dependency-Aware Urban Geospatial Modification

arXiv:2602.19326v3 Announce Type: replace-cross Abstract: Urban renewal requires incremental modifications to existing geospatial plans, yet manually updating complex layouts under spatial constraints

Domain-Specific Hallucination Detection in Large Language Models

arXiv:2609.11878v1 Announce Type: new Abstract: Large language models generate fluent text that can contain unfaithful claims -- a phenomenon known as hallucination. We present a multi-signal detectio

HuRo: Robotizing Human Videos for Scalable VLA Pretraining

arXiv:2609.10706v1 Announce Type: cross Abstract: Human video datasets have emerged as a compelling alternative to expensive real-robot data, offering rich diversity at scale. To bridge the human-to-r

Measuring Progress in Reasoning Toward Mathematical Discovery with Automatic Verification

model-releasesarxiv-cs-lg

arXiv:2603.15617v2 Announce Type: replace Abstract: Can AI make progress on important, unsolved mathematical problems? Large language models are now capable of sophisticated mathematical and scientifi

Nuha-Speech: Building General-Purpose Arabic Speech-LLMs

arXiv:2609.11892v1 Announce Type: new Abstract: As Speech Large Language Models (speech-LLMs) become increasingly multilingual, Arabic remains significantly underrepresented, highlighting the need for

OpenResearcher: A Fully Open Pipeline for Long-Horizon Deep Research Trajectory Synthesis

arXiv:2603.20278v2 Announce Type: replace-cross Abstract: Training deep research agents requires long-horizon trajectories that interleave search, evidence aggregation, and multi-step reasoning. Howev

agentsarxiv-cs-cl

Safety

Quantifying Logical Consistency in Transformers via Query-Key Alignment

arXiv:2502.17017v1 Announce Type: cross Abstract: Large language models (LLMs) have demonstrated impressive performance in various natural language processing tasks, yet their ability to perform multi

safetyarxiv-cs-ai

R4Tun: LLM-guided adaptive segmental tunnel lining segmentation in point clouds

arXiv:2609.11360v1 Announce Type: new Abstract: Automated inspection of segmental tunnel linings requires adaptive segmentation from 3D point clouds, yet expert-tuned pipelines often degrade when tunn

Reliable Near-Field Multi-User Positioning Informed by Two-Stage MUSIC

arXiv:2609.09409v1 Announce Type: cross Abstract: Near-field localization is a promising technique for high-resolution multi-user positioning in future wireless systems, but its performance is often d

Scale-Aware 3D Deep Learning for Robust Brain Metastasis Detection in Multimodal MRI

arXiv:2609.10825v1 Announce Type: cross Abstract: Detecting brain metastases in magnetic resonance imaging (MRI) remains challenging because lesions vary widely in size and appearance, with very small

researcharxiv-cs-cv

Seamless Whole Slide Label-Free Virtual Staining

arXiv:2609.10914v1 Announce Type: cross Abstract: Label-free virtual staining offers a compelling, non-destructive alternative to standard histopathology; however, its clinical adoption is hindered by

researcharxiv-cs-cv

Self-Supervised Cardiac Phase Detection via Single-Parameter Latent Orbits

arXiv:2609.11650v1 Announce Type: new Abstract: Accurate identification of end-diastole (ED) and end-systole (ES) in echocardiography underpins the quantification of ventricular function, yet manual s

Structurally Speaking: Motif-Oriented Graph Captioning through Bidirectional Graph-Text Translation

arXiv:2609.10923v1 Announce Type: new Abstract: Graph captions should help readers understand graph structure, rather than simply translate adjacency matrices into long textual edge lists. A useful gr

Tutorials

The information geometry of large language models is shared, learned, and controllable

arXiv:2609.11063v1 Announce Type: cross Abstract: Large language models learn similar behaviours, yet it remains unclear what structure they share or how to change one behaviour without disturbing oth

tutorialsarxiv-cs-cl

This is why we need open-source harnesses + local models

i've been thinking about this more after trying different agent setups. the model isn't the only thing that determines how well an agent performs. The harness around the model matters a lot too. With

UniH^3: Unifying Hierarchical Homogeneity and Heterogeneity for All-in-One Medical Image Restoration

arXiv:2609.11156v1 Announce Type: new Abstract: All-in-One medical image restoration (MedIR) aims to address diverse tasks across modalities and degradation types using a single universal model. Exist

UniMPA: A Unified Memory-Prediction-Action Model via Action-Grounded Transition Modeling

arXiv:2609.11875v1 Announce Type: new Abstract: Recent advances in Vision-Language-Action (VLA) models have improved robotic manipulation, yet observation-to-action learning remains limited by a funda

researcharxiv-cs-ro

US-hosted, with Zero Data Retention enabled by default https://lmstudio.ai/models/deepseek-v4.1-flash

model-releaseslm-studio--x

LM Studio is a US‑hosted platform that enables “Zero Data Retention” by default, meaning user inputs are not stored after processing. It currently hosts the DeepSeek V4.1 Flash model—a 552 B multimoda

VikingRAG: Accurate and Token-efficient Retrieval-augmented Generation over Structured Documents

arXiv:2609.11390v1 Announce Type: cross Abstract: State-of-the-art retrieval-augmented generation (RAG) methods exploit document structures to acquire sufficient evidence, but often incur substantial

agentsarxiv-cs-cl

What Should an Agent Forget? Separating What Is Stored from What Is Used

arXiv:2609.10263v1 Announce Type: new Abstract: Persistent language agents need stored experience to remain available across time, while each answer requires evidence suited to a particular question.

Whisper-Based Speech Transcription from Videos Across Multiple Languages for Cross-Cultural Understanding

arXiv:2609.11772v1 Announce Type: cross Abstract: Cross-cultural understanding has become increasingly important in today's highly connected, cross-national world. The success of LLM-based technologie

A Hierarchical Consistency Framework for Auditing Retrieval-Augmented Generation Systems

arXiv:2609.07075v1 Announce Type: new Abstract: Retrieval-augmented generation (RAG) is commonly evaluated by whether the final answer is correct. That test is insufficient: an answer can match its re

researcharxiv-cs-ai

A Three-Tier Persona Vector for Controllable User Simulation in Agentic Evaluation

arXiv:2609.08592v1 Announce Type: new Abstract: Evaluating tool-augmented LLM agents requires diverse, realistic user inputs yet most evaluation frameworks use flat role descriptions ('you are an angr

Audio-Maestro: Enhancing Large Audio-Language Models with Tool-Augmented Reasoning

arXiv:2510.11454v2 Announce Type: replace-cross Abstract: Recent advancements in large multimodal models (LMMs) have shown strong capabilities in audio understanding. However, most systems rely solely

Auditable Emergency Triage for Maternal and Newborn Care in India

arXiv:2609.09356v1 Announce Type: new Abstract: At Noora Health, our nurses answer more than 50,000 medical queries per month on our WhatsApp-based service that provides caregivers with on-demand supp

researcharxiv-cs-cl

AutoFyn Technical Report: Non-Parametric Expert Iteration for Long-Horizon Agents

arXiv:2609.05446v1 Announce Type: new Abstract: We introduce AutoFyn, an agent harness inspired by the Expert Iteration algorithm, adapting a frozen model across many rounds by updating persistent sta

Tutorials

BrainTaskonomy: Learning How to Pretrain and What to Transfer in fMRI Foundation Models

arXiv:2609.10518v1 Announce Type: new Abstract: fMRI foundation models increasingly aggregate heterogeneous data across brain states, cohorts, and acquisition settings, yet pretraining domains are com

tutorialsarxiv-cs-cv

CARRE: Counterfactual Action Retrieval and Reason Evaluation for Explainable Churn Prescription

arXiv:2609.09766v1 Announce Type: new Abstract: Churn models typically identify high-risk customers but do not specify which feasible retention action should be considered or why that action is approp

researcharxiv-cs-cl

CyberTiel 35B-A3B’s uncensored 4-bit quant beats Opus 4.6 medium cleanly on real codebase issues, in 27% of the time Qwen3.8-27b medium takes.

The downside of uncensoring a model is that it is known to potentially damage it, but CyberTiel is an even more capable software engineer than its censored TielCoder base, while allowing offensive sec

Detokenization Leaks: Reconstructing Local LLM Outputs From Cache Traces

arXiv:2609.06674v1 Announce Type: cross Abstract: We present a new attack that reconstructs the text generated by locally hosted LLMs by observing CPU cache activity during detokenization. Unlike prio

local-aiarxiv-cs-ai

DiSCo: A Distribution-First Steering and Cultural Prior Evaluation Framework for Measuring Cultural Preference Bias in LLMs

arXiv:2609.10253v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly deployed in globally used assistants, yet their default choices in culturally grounded everyday situations

local-aiarxiv-cs-cl