10
2
Next Concept Models (arxiv.org)
1
Gemma at 120000 Tk/S (twitter.com/osanseviero)
1
Correction Training for Diffusion Models (sunovivid.github.io)
3
Leading Chinese mathematicians unite behind use of AI in research (scmp.com)
1
The Unreasonable Effectiveness of Linear Prediction as a Perceptual Metric (arxiv.org)
1
Residual Matrix Transformers: Scaling the Size of the Residual Stream (arxiv.org)
0
Skate 3 in the Browser (aaddpp.lol)
29
Integer multiplication below n log n (github.com/openai)
44
Dust: Pretraining Transformers Without Backpropagation (qlabs.sh)
0
Language Models Can Control Their Own Attention (arxiv.org)
0
Inverse Rendering for Modeling with Line Primitives (Good for Fur) (kenji-tojo.github.io)
0
The Spectral Neuron (arxiv.org)
7
Universality of Gradient Descent Neural Network Training (arxiv.org)
1
Mobius: Foundation Model with Decoupled Knowledge and Reasoning (arxiv.org)
1
Idea for Expository AI
46
Simplifying and Refactoring Introductory Calculus (2018) (arxiv.org)
2
3B1B Virtual Career Fair (3blue1brown.com)
1
Geometric Fluid Mechanics (SGP'26 Course) (youtube.com)
2
Unified Representation for Continuous-Latent Diffusion Language Modeling (arxiv.org)
1
Minimum Description Length Meets Singular Learning Theory (arxiv.org)
2
Elastic Curves via Geometric Mechanics (arxiv.org)
2
Explorative Modeling: Unlocking a Third Pretraining Axis and E2E Generation (explorative-modeling.github.io)
2
Minimax H3 on Anime (reddit.com)
2
Neural Representation of Minimal Surfaces (arxiv.org)
3
Walk on Decomposed Subdomains (clementjambon.github.io)
3
Vakonomic Fluids (arxiv.org)
1
Diffusion ReRoll (seonsoo-p1.github.io)
1
Librrd Playground (epfl.ch)
3
A Controlled Study of Attention-Only Transformers (arxiv.org)
2
Three-Body Scattering for Generative Modeling (arxiv.org)
1
Proof of Fermat Last Theorem from Scratch (youtube.com)
5
99% on ARC-AGI 3 (public eval, with harness) (xcancel.com)
3
Audio Transport: A Generalized Portamento via Optimal Transport (arxiv.org)
2
Video Generation Models Are General-Purpose Vision Learners (genception.github.io)
2
Universal Learning of Nonlinear Dynamics (arxiv.org)
2
Learning-to-Optimize via Deep Unfolded Flows (mit-realm.github.io)
2
A Theory of Contrastive Learning with Natural Images (arxiv.org)
1
Lingbot Vision (robbyant.com)
1
Zero-Flow Encoders (arxiv.org)
16
Dispersion loss counteracts embedding condensation in small language models (chenliu-1996.github.io)
1
Meshtryoshka: Differentiable Mesh Rendering for Unbounded Scenes (danielxu9393.github.io)
1
Simplified Sparse Attention via Gist Tokens (arxiv.org)
1
Continuity-Enhancing Degree Elevation and Splits (utah.edu)
1
Forward Self Models (jagilley.github.io)
1
Tapered Language Models (arxiv.org)
78
Everything is logarithms (alexkritchevsky.com)
1
Explaining Attention with Program Synthesis (arxiv.org)
1
Sparsely gated tiny linear experts (arxiv.org)
2
Training-Free Single-Image Diffusion Models (haojunqiu.github.io)
2
Simple input–output dependencies explain neuronal activity (nature.com)
2
Self Teaching Autoencoder (the-puzzler.github.io)
1
Distance Marching for Generative Modeling (arxiv.org)
2
A Theory of Generalization in Deep Learning (arxiv.org)
3
Neural similarity predicts whether strangers become friends (nature.com)
1
Generation Is Required for Data-Efficient Perception (arxiv.org)
10
Flow Map Learning via Nongradient Vector Flow [pdf] (openreview.net)
2
How to Generate Text in One Step (one-step-lm.github.io)
1
Neural Particle Automata: Learning Self-Organizing Particle Dynamics (arxiv.org)
3
Open AI's competition spammed by AI slop (github.com/openai)
2
Effective RNG in ARC-AGI 3 (xcancel.com)
2
Matrix Valued Residuals (arxiv.org)
2
GE's Walking Truck – ca. 1965 (youtube.com)
1
Training-Free Infinite Video Generation via Evolving Memory Tokens (memrope.github.io)
3
30k Tok/S (Allegedly) (percepta.ai)
1
Towards a Neural Debugger for Python (arxiv.org)
2
Rectified Flow Auto Coder (world-snapshot.github.io)
20
Speculative Speculative Decoding (SSD) (arxiv.org)
1
One-Instruction Set Computer (wikipedia.org)
1
Mechanical Galloping (2013) (youtube.com)
2
Graphs are maximally expressive for higher-order interactions (arxiv.org)
1
Seedance 2.0 Emulating Arcane (twitter.com/nachos2d_)
4
Portable 3D Printer (ifdesign.com)
1
Generative Modeling via Drifting (lambertae.github.io)
1
Distance Marching for Generative Modeling (arxiv.org)
3
Schmidhuber Pleads Epstein for Money (xcancel.com)
5
Adobe Just Killed My Career [video] (youtube.com)
1
One-step pixel space image generation (arxiv.org)
1
Direct Manipulation of Procedural Implicit Surfaces (Siggraph Asia '24) [video] (youtube.com)
3
K-Surfaces: Bézier-Splines Interpolating at Gaussian Curvature Extrema (acm.org)
3
DiffusionBlocks: Block-Wise Neural Network Training (arxiv.org)
6
Former DeepMind engineer (possibly) experiences LLM pyschosis (xcancel.com)
1
Segment Anything Audio (meta.com)
1
Realtime Rendering of Caustics (youtube.com)
1
NeuralRemaster: Phase-Preserving Diffusion for Structure-Aligned Generation (youtube.com)
1
Evolution Strategies at the Hyperscale (eshyperscale.github.io)
1
Inner Workings of Neural Circuits (youtube.com)
1
Jelly Slider (swmansion.com)
1
Subspace-Accelerated Coordinate Descent for Physics-Based Simulation (arxiv.org)
3
Equilibrium Matching (raywang4.github.io)
1
IT³: Idempotent Test-Time Training (youtube.com)
1
Globally Aware Optimization with Resurgence (arxiv.org)
1
Neural Reparameterization Improves Structural Optimization (greydanus.github.io)
2
VR with: 1,400 nits – 180° FOV – 90 pixels per degree (meta.com)
2
3D Printing Radiance Fields (arxiv.org)
8
Stochastic Transparency [pdf] (luebke.us)
12
Percentage of one-person households, 1960 to 2018 (ourworldindata.org)
2
An Efficient Function Representation Without Neural Networks (arxiv.org)
1
Ask HN: Thoughts on Wireheading
1