AI Wiki — Knowledge Base

Introducing 𝐒𝐩𝐞𝐜𝐮𝐥𝐚𝐭𝐢𝐯𝐞 𝐃𝐞𝐜𝐨𝐝𝐢𝐧𝐠: 𝐇𝐨𝐰 𝐈𝐭 𝐄𝐯𝐨𝐥𝐯𝐞𝐝, 𝐖𝐡𝐞𝐧 𝐈𝐭 𝐒𝐭𝐚𝐲𝐬 𝐋𝐨𝐬𝐬𝐥𝐞𝐬𝐬, 𝐚𝐧𝐝 𝐖𝐡𝐚𝐭'…

Introducing 𝐒𝐩𝐞𝐜𝐮𝐥𝐚𝐭𝐢𝐯𝐞 𝐃𝐞𝐜𝐨𝐝𝐢𝐧𝐠: 𝐇𝐨𝐰 𝐈𝐭 𝐄𝐯𝐨𝐥𝐯𝐞𝐝, 𝐖𝐡𝐞𝐧 𝐈𝐭 𝐒𝐭𝐚𝐲𝐬 𝐋𝐨𝐬𝐬𝐥𝐞𝐬𝐬, 𝐚𝐧𝐝 𝐖𝐡𝐚𝐭'𝐬 𝐍𝐞𝐱𝐭. An interactive tutorial @Madisonkanna and I built for the NeurIPS Education Track. Every LLM you use generate

tutorialsswyx--x

6 Sept 2026

Search: “engineering”

I dropped out of a PhD in AI safety partially for the opposite reason. I didn’t believe AI existential risk was as important as the doomers …

Machine learning doesn't end when the model is trained. You still need to evaluate it, deploy it, and build the infrastructure to keep it wo…

MetaRSI / RSI2: A Meta-Recursive Self-Improving System for Recursive Self-Improving Systems Themselves

Miles v0.1: Production-Level Post-Training

mjorbit: A Simulation Framework for Space Robotics

Modal Logic Neural Networks

On the Recall Scaling Laws in Mamba: A Theoretical and Mechanistic Study via Hashing

PatchFormer: A Patch-Based Time Series Foundation Model with Hierarchical Masked Reconstruction and Cross-Domain Transfer Learning for Zero-Shot Multi-Horizon Forecasting

PhysFlow: Physics-Aware Optical Flow for Motion Controllable Video Generation

Riemannian Optimization for Multi-Player Quantum Games on Product Unitary Manifolds

RoMu4o: A Robotic Manipulation Unit For Orchard Operations Automating Proximal Hyperspectral Leaf Sensing

Scene Graph-Driven Haptic Feedback for Safety Enhancement in Robotic Ophthalmic Surgery via Physically Simulated iOCT

SeisBench DAS: A machine learning framework for Distributed Acoustic Sensing

Sparse Oblique Rule Boosting for Simpler Additive Rule Ensembles

Spatiotemporal Heterogeneity of AI-Driven Traffic Flow Patterns and Land Use Interaction: A GeoAI-Based Analysis of Multimodal Urban Mobility

This is a really neat benchmark. It's super interesting that Astra starts strong and holds the lead for up to 19 hours, but Fable 5.1 catche…

Toward Sustainable Distributed LLM Inference: A Systems Synthesis and Research Agenda for an Energy-, Carbon-, and Cache-Aware llm-d Control Plane

Uni-Light: An Ultra-Lightweight Framework via Uncertainty-Aware Knowledge Distillation for Brain Tumour Segmentation

Unifying Physics-Based Humanoid Interaction with a Context-Conditioned Interaction Prior

Vision: Data-Centric Anchoring for Robust and Interpretable Agentic AI

What settings do you use for running Qwen3.8-Flash-Next in llama.cpp?

Great to see Stripe’s platform and agent for knowledge work, Kai, get featured on Claire’s show! My favorite use case was asking it question…

How HPE Zerto built an agentic troubleshooting system with Amazon Bedrock

How KDDI built Buffmee, a faster, reliable consumer RAG app

How many agents can 2×4090 actually run at once? Three weeks of llama.cpp concurrency data — soft cap 5 @ 64k, hard cap 9, and why.

Introducing CUDA Rust: Two Tracks for Writing GPU Kernels

Abstraction Agent

After over a year of my nights and weekends, the Jenny app is done!

CUA-Universe: A Scalable and Dynamic Environment for Hybrid GUI+CLI Agents

🚨EXCLUSIVE: David Silver is assembling an absurd reinforcement learning team at Ineffable Intelligence. Six new “cofounders,” including fou…

ExLlamaV3 is underrated

Fascinating paper from Anthropic and colleagues. They study whether models can tell when they are being tested. It turns out that capable mo…

Fast Gauss Sums via Flash Attention

Fractal basins trap latent reasoning

GEPARD - Generative, Prosody-aware, Autoregressive text-to-speech model for Realtime Dialogue

I made Warrior Quest, a local LLM-powered dark-fantasy RPG where the model only plays NPCs and the actual game state stays deterministic

My Qwen3.8-27B task-aware quant reaches 99% of BF16 reasoning performance at 15% of the size.

Partial Inverse Design of High-Performance Concrete Using Cooperative Neural Networks for Constraint-Aware Mix Generation

Persistent Robot World Models: Stabilizing Multi-Step Rollouts via Reinforcement Learning

Qlippy: A Retrieval-Augmented GenAI Assistant for Reproducible Quantum Workflows and Experiment Tracking

The Sample Complexity of Learning Lipschitz Operators with respect to Gaussian Measures

This is actually an example of why Astra is not artificial general intelligence (AGI). I see the same issues with visual perception demos ie…

ToPos: Automated Optimal Positioning on Topographic Manifolds using Constrained Geodesic Voronoi Decomposition

48 tg/s 440 prefill on my grandma's cluster (2xP40) (sort of)

Coding benchmarks that are quickly showcasing deep capability

Introducing 𝐒𝐩𝐞𝐜𝐮𝐥𝐚𝐭𝐢𝐯𝐞 𝐃𝐞𝐜𝐨𝐝𝐢𝐧𝐠: 𝐇𝐨𝐰 𝐈𝐭 𝐄𝐯𝐨𝐥𝐯𝐞𝐝, 𝐖𝐡𝐞𝐧 𝐈𝐭 𝐒𝐭𝐚𝐲𝐬 𝐋𝐨𝐬𝐬𝐥𝐞𝐬𝐬, 𝐚𝐧𝐝 𝐖𝐡𝐚𝐭'…

PINNStudio: A free, open-source no-code GUI for setting up, training, and visualizing PINNs [P]

Research acceleration: The view inside OpenAI