Search: “inference”

Can LLMs in Draft-Verify-Revise Pipelines Resolve Deictic Ambiguity?

model-releasesarxiv-cs-cl

arXiv:2609.12162v1 Announce Type: cross Abstract: Draft-verify-revise is a common LLM orchestration pattern for scaling inference-time compute. One LLM drafts, a second critiques the draft and provide

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Fixed State, Long Reach: What a Constant-Size Cache Buys Block Diffusion at Scale

DGX agent

arXiv:2609.11998v1 Announce Type: new Abstract: Diffusion language models decode tokens in parallel, but their bidirectional denoiser rules out the naive key--value (KV) cache behind fast autoregressi

safetyarxiv-cs-lg

14 Sept 2026

Google is a leader in The Forrester Wave™: Public Cloud Platforms, Q3 2026

model-releasesgoogle-cloud-ai

We are excited to share that Google Cloud was named a Leader and received the highest score in the ‘current offering’ category in the Forrester Wave™: Public Cloud Platforms, Q3 2026 report, which exa

GSO-Net: Visual State Machines for Hazardous Freight Transfer Compliance at Petrochemical Logistics Nodes

arXiv:2609.12408v1 Announce Type: new Abstract: Hazardous-freight operations at petrochemical logistics nodes are safety-critical for intelligent transportation systems, yet existing vision benchmarks

PEARL: Structural Privacy-Utility Control in Human-Centric CPS via Personalized Early-Exit Deep Reinforcement Learning

arXiv:2403.05864v3 Announce Type: replace Abstract: In human-centric Cyber-Physical Systems (CPS), personalized Deep Reinforcement Learning (DRL) agents must share fine-grained control actions with cl

3k$ 128GB VRAM + 256GB RAM DDR4 Server

I finished my home inference server. First I tried Lenovo p620 workstation and while it’s a good value overall it pissed me off with a ton of proprietary Lenovo shit to deal with and I return it in th

local-air-localllama

13 Sept 2026

I built a serverless hosting platform for LoRA adapters with vLLM

It’s always bothered me that after fine-tuning a model for a project, there isn’t a particularly easy way to host it without either running it locally and keeping a GPU on 24/7 or paying for an entire

13 Sept 2026

A Time-Based Readout for Vector-Matrix Multiplication in Fully Analog Memristive SNNs

arXiv:2609.11713v1 Announce Type: cross Abstract: Artificial neural networks rely on vector-matrix multiplications (VMMs), whose implementation in von Neumann architectures is dominated by costly data

researcharxiv-cs-ai

12 Sept 2026

Can Edge-Deployable Vision-Language Models Identify Species?

arXiv:2609.11916v1 Announce Type: new Abstract: Camera traps often run in the field on edge hardware with limited or no connectivity, making small, locally-deployable vision-language models (VLMs) --

local-aiarxiv-cs-ai

12 Sept 2026

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

12 Sept 2026

A Dynamic Fusion Large Language Model for Traffic Flow Prediction

arXiv:2609.11314v1 Announce Type: new Abstract: Traffic flow prediction is a core supporting technology for intelligent transportation systems. It uses historical data to infer future traffic dynamics

Bio-inspired Learning and Decision-Making with Probabilistic In-Memory Computing Hardware: Part 1

arXiv:2609.11281v1 Announce Type: cross Abstract: Learning and decision-making in animals are often modeled as Bayesian processes, where sensory evidence is integrated with prior beliefs to guide beha

tutorialsarxiv-cs-lg

Compute-Bounded Security Assurance - Coverage, Verification, and Response under Resource Constraints

model-releasesarxiv-cs-ai

arXiv:2609.09229v1 Announce Type: cross Abstract: Additional inference compute can increase the number of correctly resolved security-assurance tasks, but repeated success, unique coverage, accepted e

Detectable Only Where It Is Confounded: What Verified Duplication Counts Say About Membership Evidence in Language Models

arXiv:2609.10830v1 Announce Type: new Abstract: When a language model finds a sentence unusually cheap to predict, it is tempting to conclude that the sentence was in its training data. Almost every p

researcharxiv-cs-cl

GEOSTEER: Geodesic Optimization for Activation Steering in Large Language Models

arXiv:2609.10658v1 Announce Type: new Abstract: Activation steering provides a lightweight way to control large language models (LLMs) by modifying their hidden activations at inference time. Among th

tutorialsarxiv-cs-lg

Learning/RSI through ngrams?

Hey gang, im wondering if you in theory could use ngrams as seen with Qwen 3.8 Flash or DS4.1 in order to dynamically train the model? Normally the ngram embeddings behave similar to a lookup table of

Hardware

Phase-Decoupled, Model-Calibrated Power Control for Disaggregated LLM Serving

arXiv:2609.11133v1 Announce Type: new Abstract: Datacenter GPU power is the binding constraint on LLM serving capacity, and production serving has shifted to prefill/decode (PD) disaggregation. Deploy

hardwarearxiv-cs-lg

Rethinking Radiomap Blind Prediction with Limited Environment and Configuration Representations

arXiv:2609.11255v1 Announce Type: cross Abstract: Radiomap blind prediction infers radiomaps from observable representations of the propagation environment and base station (BS) configuration without

World in World: Explore the World with World Models

arXiv:2609.11548v1 Announce Type: new Abstract: Autoregressive video world models enable interactive, long-horizon exploration, but flexible control remains challenging. Exploring a source video from

tutorialsarxiv-cs-cv

Hardware

d-Matrix Adopts NVIDIA NVLink Fusion for Rack-Scale XPU Deployment

AI inference chipmaker d-Matrix today announced it will use NVIDIA NVLink Fusion to connect its next-generation Raptor XPUs to NVIDIA’s AI infrastructure platform — joining a growing roster of ecosyst

hardwarenvidia-blog

Elastoformer: Enabling Dynamic Adaptivity via Elastic Model Transformation

arXiv:2609.10018v1 Announce Type: new Abstract: EdgeAI systems are increasingly employing computer vision applications to enable intelligent, on-device decision-making in real-time. However, these dep

local-aiarxiv-cs-cv

LightMedSeg-ISLES: Stroke Lesion Segmentation with 81x Fewer Parameters than nnU-Net

arXiv:2609.09634v1 Announce Type: new Abstract: Large networks and ensembles often lead medical image segmentation challenges, but their storage and inference demands complicate deployment. We present

Looped GPT-BERT: Trading Parameters for Computation in Small Language Modeling

model-releasesarxiv-cs-cl

arXiv:2609.09691v1 Announce Type: new Abstract: When training data are limited, increasing parameter count is not the only way to improve language-model performance. A small parameter set, when repeat

Orukeet: Multilingual ASR with Frozen Gabor Kernels

arXiv:2609.10054v1 Announce Type: cross Abstract: Orukeet replaces half of an adapted Parakeet encoder's temporal filters with 12,288 fitted Gabor kernels, freezes these replacements, and trains the r

Quantization Amplifies Determinism, Not Bias: Scale-Dependent Behavioral Effects of Serving-Time Weight Compression

arXiv:2609.07901v1 Announce Type: new Abstract: Weight quantization largely determines the economics of serving open-weight LLMs. Its costs are usually assessed with capability benchmarks, on which 4-

safetyarxiv-cs-ai

Agents

SocialRL: Refining LLMs' Social Intelligence through Multi-turn Reinforcement Learning and Reward Design

arXiv:2609.09764v1 Announce Type: new Abstract: Social intelligence enables agents to read social context, infer intent, and adapt over sustained dialogue. As language models become autonomous collabo

agentsarxiv-cs-cl

Stable Answers, Unfinished Reasoning: Why Self-Consensus Is Not a Safe Early-Exit Signal

arXiv:2609.09989v1 Announce Type: new Abstract: A natural way to cut reasoning-model inference cost is to repeatedly probe a single partial trajectory for its current answer and stop once probes agree

researcharxiv-cs-cl

TRACE: Trajectory-robust Admission with Evidence Ordering for Efficient GUI Agents

arXiv:2609.10297v1 Announce Type: new Abstract: GUI agents accumulate high-resolution screenshots as the trajectory unfolds, increasing inference latency and memory usage. Training-free visual token p

uFlowCSP: Crystal Structure Prediction using Mean flow generative models

arXiv:2609.09799v1 Announce Type: cross Abstract: Crystal structure prediction (CSP) is fundamental to computational materials discovery. Generative models including CDVAE, DiffCSP, FlowMM, and Crysta

tutorialsarxiv-cs-lg

What Does an LLM-Agent Leaderboard Rank Actually Compare?

model-releasesarxiv-cs-ai

arXiv:2609.07785v1 Announce Type: new Abstract: An LLM-agent leaderboard invites a familiar inference: an agent ranked above another is the better agent. Public evaluation logs may not support that co

Agents

WorldAgen: Unified State-Action Prediction with Test-Time World Model Training

arXiv:2609.08162v1 Announce Type: new Abstract: How can vision-language-action (VLA) models adapt to new environments where world dynamics shift? While recent research has combined world modeling and

agentsarxiv-cs-ai

1-bit 27B in the browser: 25–30 tok/s on a 6 GB RTX 3060 Laptop (WebGPU, no install)

mentria.ai is a browser inference engine I've been building solo, from scratch in WebGPU/WGSL. This week it crossed a milestone I had been chasing for a while: a 27B one-bit model answering at up to 3

Architectural and Regularization Components in Deep Learning Medical Image Registration: Systematic Ablation Study

arXiv:2609.05484v1 Announce Type: new Abstract: Deep learning registration methods routinely stack two kinds of enhancement on a base network: architectural additions such as affine pre-alignment stag

safetyarxiv-cs-cv

Hardware

CASCADE: A Spatio-Temporal-Causal Reasoning Representation and Dataset for Driving

arXiv:2609.07094v1 Announce Type: new Abstract: Reasoning is a promising route to the generalization that autonomous driving requires in the long tail, as it can infer how the elements of a scene depe

hardwarearxiv-cs-cv

Compressed Recurrent Feedback in Tsetlin Machines: A Reproducible Boolean-FSM Study

arXiv:2609.06133v1 Announce Type: new Abstract: Sequential inference on small devices requires a model to retain useful history without repeatedly processing a long input record. A Recurrent Tsetlin M

Contextual Observer Grounding: Evaluating Situated Spatial Reasoning in Vision-Language Models

arXiv:2609.06880v1 Announce Type: new Abstract: Reasoning over language instructions in embodied tasks such as robotics often requires understanding spatial relations from a speaker's situated perspec

Decoder-Side Semantic Conditioning for Low-Bitrate Neural Speech Compression

arXiv:2512.21653v2 Announce Type: replace-cross Abstract: Speech codecs are usually optimized for waveform fidelity, allocating bits to acoustic detail that can be inferred from linguistic structure.

Disentangling Steering Vectors

arXiv:2609.07037v1 Announce Type: new Abstract: Activation steering has emerged as a lightweight, inference-time approach to control the behavior of Large Language Models (LLMs). However, traditional

FILT3R: Latent State Adaptive Kalman Filter for Streaming 3D Reconstruction

arXiv:2603.18493v2 Announce Type: replace Abstract: Streaming 3D reconstruction maintains a persistent latent state that is updated online from incoming frames, enabling constant-memory inference. A k

researcharxiv-cs-cv

Geometry-Aware Test-Time Learning for Quantitative Spatial Reasoning

arXiv:2609.06004v1 Announce Type: new Abstract: Quantitative spatial reasoning in visual-language models (VLMs) aims to infer spatial distances and directional relationships among objects in 3D space

researcharxiv-cs-cv

Applications

HoneyRoute: Honeypot-Model Routing for Adversarial LLM Serving

arXiv:2609.08306v1 Announce Type: cross Abstract: We introduce HoneyRoute, an inference-serving layer that detects whether an incoming request is malicious and, if so, routes it to a dedicated honeypo

applicationsarxiv-cs-lg

KDFlow: A User-Friendly and Efficient Knowledge Distillation Framework for Large Language Models

arXiv:2603.01875v4 Announce Type: replace-cross Abstract: Knowledge distillation (KD) is widely used to compress and post-train large language models (LLMs), yet many existing frameworks execute teach

Learning functional components of PDEs from data using neural networks

arXiv:2602.13174v2 Announce Type: replace Abstract: Partial differential equation (PDE) models frequently contain unknown functional terms that cannot be measured directly, limiting their predictive u

Let It Go or Learn to Self-Correct: Continuous Diffusion for Constrained Discrete Tasks

arXiv:2609.09009v1 Announce Type: new Abstract: Denoising Diffusion Probabilistic Models (DDPMs) generate samples by starting from noise and repeatedly denoising while keeping each update close to the

Norms at a Price: Why RL-Based Alignment Can Promise Conditional Compliance at Best

arXiv:2609.07627v1 Announce Type: cross Abstract: AI agents sometimes act aligned when they infer they are being tested, and differently when not. We argue this is not an anomaly but what current trai

Novel hybrid protein scaffold gap filling using weighted machine learning ensemble, beam search, and mass-constrained reranking

arXiv:2609.05436v1 Announce Type: cross Abstract: Protein scaffold gap filling is an important computational task in protein sequence reconstruction, where missing amino acid regions must be inferred

Object-Aware Background-Controlled Editing via Weighted Velocity Guidance

arXiv:2609.06288v1 Announce Type: new Abstract: Training-free image editing steers diffusion or flow-matching generative models at inference time by modifying prompt-conditioned denoising velocities.

local-aiarxiv-cs-cv

PAC-Private Autoregressive Generation: Calibrating Noise to Ensemble Disagreement

arXiv:2609.05676v1 Announce Type: new Abstract: Language models adapted on private text are often served through APIs, so privacy leakage occurs through generated outputs rather than exposed weights.