The Archive

MAPLE framework disaggregates knowledge-extraction evaluation beyond aggregate accuracy to measure diverse effects of steering and decoding techniques.

LiFi framework generates LiDAR point clouds from multi-view RGB images using geometric and semantic guidance for simulation and perception tasks.

Measures information transmission limits between LLMs via prompt optimization and novel transmission metrics to quantify concept articulation as model scale increases.

Gauge-equivariant GNNs exhibit loop geometry with rapid contour growth and slow spatial expansion independent of network training.

GeoReader improves vision-language models' spatial reasoning by learning to actively use geometric representations aligned with latent features.

RadarGSOcc predicts 3D semantic occupancy for autonomous driving using 4D radar tensors with Doppler-aware geometric-semantic feature extraction.

Language model logits follow consistent log-linear geometry across contexts and scales, emerging during pretraining and constraining decoding behavior.

Graph Metanetworks extends weight-space learning to transformers by accounting for attention-induced continuous symmetries beyond neuron permutations.

Roadside camera-radar 3D object detection adapts end-to-end models for autonomous infrastructure despite varying sensor mounting geometry.

As usage, costs, and customer expectations change, AI companies are moving beyond traditional SaaS pricing. Leaders at Monetize 2026 shared how they’re combining subscriptions with usage-based billing, building spend controls into products, adapting packaging and value metrics, and making pricing readable for both human and AI buyers.

Interactive 4D volumetric liquid forecasting task predicts spatiotemporal fluid dynamics conditioned on solid object motion for robotics and digital twin applications.

Fisher information matrix analysis on CIFAR-10 vision models reveals how top eigenvectors decouple generalization from memorization under controlled noise.

Paper identifies failure modes in multimodal entity alignment for knowledge graphs caused by incomplete modalities and structural asymmetries during fusion.

Steering Hessian analysis explains fragility of contrastive activation addition (CAA) for LLM behavioral control via curvature geometry of steering vectors.

DisCo algebra formalizes joint probability decomposition via symmetric operads and cooperads using block-wise CDFs.

RL framework reduces object hallucination in LVLM image captioning using metric-driven rewards without human feedback.

CASP-VL benchmark reveals VLMs lose spatial reasoning under perturbations; EICS metric detects representation instability in causally-controlled interventions.

Systematic study of design levers in continuous diffusion language models, examining self-conditioning and embedding geometry for accelerated sampling.

Phoneme-level error analysis of four data augmentation methods for low-resource ASR reveals complementary benefits beyond aggregate error rate metrics.

Multi-objective optimization using norm-constrained oracles with entropic weight updates achieves O(T^-1/2) ergodic stationarity.

FedPCD proposes priority-constrained federated optimization to improve fairness across heterogeneous clients without symmetric weighting.

Representational similarity metrics (RSA, CKA, Procrustes) are limited by input selection and grouping choices made before measurement; study across vowel acoustics in 27 representations.

Systematic bibliometric survey of Tibetan NLP literature (2004–2026) reveals rapid growth since 2018 across corpora, segmentation, and speech tasks.

Preparing CAD assets for robotics simulation requires more than converting geometry to OpenUSD: developers must configure and validate materials, collision... Preparing CAD assets for robotics simulation requires more than converting geometry to OpenUSD: developers must configure and validate materials, collision geometry, joints, and other physics properties before testing robot behavior. NVIDIA Omniverse libraries, guided by SimReady Foundation specifications and agentic NVIDIA skills, provide a structured workflow for converting and validating… Source

Statistical reanalysis of METR's AI time-horizon benchmark using splines and IRT reveals non-linear task-difficulty scaling, questioning validity of capability extrapolation.

Psychometric audit of HELM Safety benchmark questions whether harmful refusal constitutes a measurable, isolated attribute in safety evaluations.

ViSkill framework enables VLM agents to learn visual-native skills from trajectories while optimizing policy, preserving geometric structure over text linearization.

SpaceFlow provides training-free pipeline for locally controllable 3D generation using geometric primitives as proxies for object parts with per-region control.

GeoReform framework improves MLLM geometry problem-solving through reflective formalization evolution, addressing redundant/ambiguous constraint representation.

GPD distillation method improves vision-language model spatial reasoning by leveraging 3D geometry during training without inference overhead.