AI Wiki — Knowledge Base

Harness EngineeringからRSIへ AI自身がAIをつくり、改善していく──再帰的自己改善（RSI）はどのように実現するのでしょうか。この問いに対して、Thinking Machines Lab 共同創業者の Lilian Weng 氏（@lilianweng）が…

Harness EngineeringからRSIへ AI自身がAIをつくり、改善していく──再帰的自己改善（RSI）はどのように実現するのでしょうか。この問いに対して、Thinking Machines Lab 共同創業者の Lilian Weng 氏（@lilianweng）が、モデルを取り巻く実行系＝「ハーネス」の設計こそが現実的な出発点だと論じるブログ記事を公開しました。RSI周辺の研究が

researchdavid-ha--x

7 Jul 2026

research

Green for Go, Red for No: Visual Grounding via Semantic Segmentation for VLA Navigation Policies

Greenland isn’t surrounded by Chinese and Russian ships. There hasn’t been $19.2 trillion invested in the US under Trump. Trump hasn’t settl…

GUSH3R: Everyone Everywhere All at Once as Gaussians

Handwriting Trajectory Recovery with Diffusion Models

Harness EngineeringからRSIへ AI自身がAIをつくり、改善していく──再帰的自己改善（RSI）はどのように実現するのでしょうか。この問いに対して、Thinking Machines Lab 共同創業者の Lilian Weng 氏（@lilianweng）が…

HASSL: Hierarchy-Aware Self-Supervised Learning Framework for Single Cell Microscopy

Heterogeneous Graph Condensation via Role-Aware Clustering

Hierarchical Evidence-Driven Reasoning for Long Document Understanding

Hierarchical Sparse Attention Done Right: Toward Infinite Context Modeling

HiMe: Hierarchical Embodied Memory for Long-Horizon Vision-Language-Action Control

How Many Initial Points Does Bayesian Optimization Need?

Hyperparameter Transfer in Graph Neural Networks

iFLYTEK-Embodied-Omni Technical Report

Improved Finite-Particle Convergence Rates for Stein Variational Gradient Descent

Improved generalization bounds for binary linear classification via isoperimetry

In-span learning: adapting reduced-order models using their own predictions

Inclusive KL Gradient Flows: Otto-Wasserstein, Fisher-Rao-Gaussian, and Local-Estimator Dynamics

Incredible WSJ reporting on how European leaders deal with Trump -European leaders workshopped tweets to send Trump, debating which words to…

Individual Parameters in Weight-Sparse Transformers Appear Interpretable

Inelastic Constitutive Kolmogorov-Arnold Networks: A generalized framework for automated discovery of interpretable inelastic material models

Information-Geometric Superposed Vowel Evaluation: Part 1. Moraic Syllabary (Japanese)

InSpace: Structure-Aware 3D Indoor Scene Generation from a Single 360{eg} Image

Integrated Forward-Inverse Network for Lensless Image Reconstruction

Interaction Techniques that Encourage Longer Prompts Can Improve Psychological Ownership when Writing with AI

Interpretable machine learning predicts Parkinson's disease severity using motion-corrected QSM MRI and multiband multiecho fMRI features

Inverse Design of Metainterfaces for Static Friction Control: Beyond the Hertzian Limit

InverseCrafter: Efficient Video ReCapture as a Latent Domain Inverse Problem

InverseScope: Scalable Activation Inversion for Interpreting Large Language Models

Is Generation Required for Data-Efficient Perception?

It's all In the (Exponential) Family: An Equivalence between Maximum Likelihood Estimation and Control Variates for Sketching Algorithms

@__JohnNguyen__ and I are presenting the Beyond Language Modeling paper as an ICML spotlight in 30 minutes at the 10:30 AM poster session! I…

Jointly Improving Dialect Identification and ASR in Indian Languages using Multimodal Feature Fusion

KAN vs LSTM Performance in Time Series Forecasting

KinEMbed: Decoding Kinematics from Electromyography via Cross-Modal Contrastive Learning

Knowledge Knows, Verbalization Tells: Disentangling Latent Directions for Mathematical Solvability in LLMs

Language Generation with Replay: A Learning-Theoretic View of Model Collapse

Language Models as Higher-Order Planning Formalizers

Language Models Represent and Transform Concepts with Shared Geometry

Large-scale dataset of automatically classified rhetorical sections in scientific papers

LAW & ORDER: Adaptive Spatial Weighting for Medical Diffusion and Segmentation

Layer-Parallel Inference Reduces Encrypted Nonlinear Depth in Transformers

LBTCap: A Lightweight Bilateral Transformer for Real-Time Remote Sensing Image Change Captioning

LCPNet: Latent Consistent Proximal Unfolding Network for Infrared Small Target Detection

Learning from Lost Provenance: Multiple Instance Learning for Cancer Registry Tumor Group Classification

Learning on the Manifold: Unlocking Standard Diffusion Transformers with Representation Encoders

Learning rate adaptive stochastic gradient descent optimization methods: numerical simulations for deep learning methods for partial differential equations and convergence analyses

Learning Taxonomic Trees with Hierarchical Representation Regularization for Large Multimodal Models

Learning to Generate Human-Human-Object Interactions from Textual Descriptions

@JohnNguyen and I are presenting the Beyond Language Modeling paper as an ICML spotlight in 30 minutes at the 10:30 AM poster session! I…