The Archive

GBClaw agentic AI framework automates grain boundary discovery in materials science via autonomous experimentation and mechanistic inference.

AdaDex combines reinforcement learning with motion imitation to train dexterous multi-finger robot hands on generalizable, human-like manipulation skills.

UniWALM unified world-action-language model learns video, action, and language jointly for embodied robot control with memory-dependent behavior.

IndustrialRemembr extends NVIDIA ReMEmbR spatiotemporal memory system to robot fleets with on-demand scene encoding to reduce compute costs.

PUN neural operator learns noise-aware value fields for robot planning under stochastic control constraints via screened-Poisson formulation.

MindView benchmark evaluates generative multi-view spatial intelligence across 1,500 samples with tasks spanning geometry and semantic reasoning.

Adaptive multi-token prediction uses online control to optimize LLM decoding horizon based on verification feedback, balancing latency and compute.

Edge-aware attention layer for graph generation incorporates structural information while leveraging hardware-optimized Transformer implementations.

NarraSpace multimodal system links speech and visual scene elements for dementia classification by grounding narrative explanations in shared picture evidence.

The former Google CEO owns about 1% stake in the leading AI lab, Bloomberg reported.

Multi-agent framework using property graphs for autonomous root-cause analysis in distributed microservice logs.

DAR-MGSM benchmark evaluates math reasoning across five Arabic dialects and Modern Standard Arabic.

Diagnostic metric framework analyzes value expression and organization in LLM-generated debate text.

TS2Vec adapted as single-encoder foundation model for time series classification without per-dataset retraining.

Agent S4 meta-agent autonomously designs evolving computer-use interfaces tailored to task workflows.

Neural implicit energy optimization regularizes signed distance fields for large-scale LiDAR scene reconstruction.

RTLRepairDataset and RTLRepairBench identify measurement confounds in HDL repair benchmarks using simulation and formal equivalence.

PaPi integration scheme enables parallel time-stepping for physics-informed neural networks solving PDEs.

WildMAE extends masked autoencoders with structure-aware refinement for camera-trap wildlife image reconstruction.

Proof that two-layer ReLU networks can exactly simulate arbitrary ReLU networks via homological algebra.

Meta argues nobody “decided” to pirate books to train AI.

Study reveals on-policy distillation fails beyond compounding error; proposes teacher repair via selective intervention in student trajectories.

SceneRig reconstructs simulation-ready 3D scenes from single images using agentic physics grounding, enabling scalable scene generation for robotics.

CoRAL hybrid framework pairs diffusion planner with autoregressive executor for parallel multimodal reasoning, reducing latency via asymmetric orchestration.

Policy gradient framework for continuous-time RL from discrete observations; proves exponential convergence under entropy regularization and stochastic dynamics.

Spatial transcriptomics method encodes cell-cell relationships with direction sensitivity for microenvironment representation learning.

Sage-A1.5 35B MoE agentic model enables adaptive task composition, verification, and scientific discovery via subagent orchestration.

Cache engineering for LLM agents optimizes KV storage and prefilling costs across long trajectories without invalidating cached states.

Token-Energy Law models GPU energy per decoded token as function of active bytes and realized time, enabling efficient LLM deployment.

LLM-driven evolutionary program search optimizes semantic index design for generative retrieval, automating end-to-end SID component tuning.