Saranraj Nambusubramaniyan @memoryencoder

AI Architect | Neuro + Math + CS | Building thinking machines sarannns.github.io Chennai, India Joined May 2014

Tweets

228
Followers

306
Following

641
Likes

3K

Arthur Gretton @ArthurGretton

3 weeks ago

Your drifting model is secretly a fixed point for the Wasserstein gradient flow on... ...the KL? ...an approximation to the Sinkhorn? ...Is it even a Wasserstein gradient flow at all? arxiv.org/abs/2605.05118 @liwenliang @agalashov @JamesTThorn @ValentinDeBort1 @ArnaudDoucet1

2 82 441 64K 355

View Details

Marco Fumero @marco_fumero

a year ago

Inductive biases in the training process cause the formations of attractors in the latent vector field, characterizing memorization and generalization regimes of the network.

1 4 15 3K 3

View Details

Stat.ML Papers @StatMLPapers

2 months ago

Random Matrix Theory for Deep Learning: Beyond Eigenvalues of Linear Models ift.tt/EQOHiNt

0 52 369 24K 351

View Details

CLaE @leafs_s

2 months ago

The Brain May Have Been Critical Before It Became Predictive: Self-Organized Criticality and the Physical Basis of Action-Readiness zenodo.org/records/193250…

0 24 93 9K 65

View Details

SIAM Activity Group on Dynamical Systems @DynamicsSIAM

2 months ago

Tutorial article: "Energy-Based Dynamical Models for Neurocomputation, Learning, and Optimization" (by Arthur N. Montanari, Francesco Bullo, Dmitry Krotov, Adilson E. Motter): arxiv.org/abs/2604.05042

2 17 75 4K 53

View Details

Christos Constantinidis @c_constan

2 months ago

New preprint from the lab! Prefrontal Mechanisms of Rule Learning doi.org/10.64898/2026.… (1/10)

1 14 39 4K 20

View Details

Ido Amos @AmosaurusRex

4 months ago

Can LLMs reason internally while processing their inputs, similar to how humans can think ahead as we process information? Our latest work introduces Thinking States, a novel architectural adaptation that transforms reasoning into a internal recurrent process. By training models to maintain a dynamic thinking state, we achieve significant inference speedups over Chain-of-Thought while substantially outperforming existing latent reasoning methods. Paper: arxiv.org/abs/2602.08332

5 27 132 13K 105

View Details

Carlos E. Perez @IntuitMachine

3 months ago

With the emergence of AI, we need to change the question from "what is intelligence?" to "what is agency?" Here is my model of agency:

5 10 35 3K 40

View Details

Simone Foti @simo_foti

3 months ago

It's time to bring 3D meshes into modern machine learning properly! 🛸 Our work solves the non-differentiability of the Exp map on meshes, enabling gradients to flow directly through geodesics. It’s differentiable, GPU-fast, and fully parallelised. circle-group.github.io/research/DSG

7 79 684 69K 608

View Details

Anna Ciaunica PhD @annaciaunica.bsky.social @AnnaCiaunica

a year ago

Cooked a new paper 😎 The No Body Problem - Intelligence and Selfhood in Biological and Artificial Systems osf.io/preprints/psya…

15 35 169 13K 96

View Details

Aran Nayebi @aran_nayebi

3 months ago

1/ As AI agents become increasingly capable, what must *inevitably* emerge inside them? We prove selection theorems: strong task performance forces world models, belief-like memory and—under task mixtures—persistent variables resembling core primitives associated with emotion.

3 38 215 41K 168

View Details

The Brain Prize @BrainPrize

3 months ago

Meet the winners of The Brain Prize 2026. The announcement film introduces Professors David Ginty and Patrik Ernfors, awarded the prize for their groundbreaking discoveries on the neuroscience of touch and pain. Learn more about the winners: brainprize.org/winners/touch-…

1 13 34 2K 5

View Details

PNASNews @PNASNews

4 months ago

One of the most-viewed PNAS articles in the last week is “Quantifying the compressibility of the human brain.” Explore the article here: ow.ly/3TPc50Y6heR For more trending articles, visit ow.ly/AziX50Y6heS.

4 36 239 48K 173

View Details

Learning Systems and Robotics Lab (is hiring!) @learnsyslab

3 months ago

Our work, "A Primer on SO(3) Action Representations in Deep Reinforcement Learning," was accepted to #ICLR2026! We provide a systematic study of action representation choices in RL, showing that they fundamentally impact training stability and performance. #Robotics #AI #RL

3 53 472 50K 353

View Details

elvis @omarsar0

3 months ago

First empirical study on how developers are actually writing AI context files across open-source projects. Researchers scanned 10,000 repositories and found only 466 (5%) have adopted AI configuration files like AGENTS dot md, CLAUDE dot md, or Copilot instructions. Why does it matter? Of the 155 AGENTS dot md files analyzed, 50% were never modified after the initial commit. Only 6% had 10 or more revisions. The most common content in these files were conventions, contribution guidelines, and architecture overviews. But there is no standard structure, wide variation in what teams encode, and most files are written once and left to decay. The conventions for this new form of documentation are still in flux. Paper: arxiv.org/abs/2510.21413 Learn to build effective AI agents in our academy: academy.dair.ai

42 62 329 55K 394

View Details

Zhuoran Yang @zhuoran_yang

4 months ago

New Paper -- "On the Mechanism and Dynamics of Modular Addition: Fourier Features, Lottery Ticket, and Grokking" We give a complete mechanistic and dynamic picture of how neural networks learn modular addition f(x,y) = (x+y) mod p. We answer three questions: (1) What does the trained network compute? (2) How do Fourier features emerge during training? (3) Why does grokking happen? Each answer comes with a mathematical characterization backed by theory and experiments. Paper: arxiv.org/abs/2602.16849 Blog: y-agent.github.io/posts/modular_… Demo: huggingface.co/spaces/y-agent… Code: github.com/Y-Agent/modula…

7 52 325 38K 259

View Details

Jason Stock @itsstock

4 months ago

Generative Models via Drifting in MLX. A minimal single file implementation showing how we evolve a pushforward distribution with a drifting field during training for single-step inference. No ODE/SDE solvers, adversarial loss, or complex nets. code: github.com/stockeh/mlx-dr…

14 36 327 23K 207

View Details

Rosinality @rosinality

4 months ago

JiT + MeanFlow!

2 44 373 19K 255

View Details

george @georgeyw_

5 months ago

“Neural networks are grown, not programmed” We’re changing that. Mechinterp investigates how models generalize beyond their training data by studying the resulting internal structure. We introduce patterning as the dual: given desired structure, determine what data produces it.

19 157 1K 92K 960

View Details

Andrew Lampinen @AndrewLampinen

5 months ago

Is it better to go to academia or industry for research in AI or cognitive science? It's the most frequent question I get asked by PhD students, and I've finally written up some of my thoughts on the answers, as an epilogue to my more research-focused posts on cog sci and AI: