Media Summary: I read papers during my weekly livestream, Saturdays at 8AM Pacific Time. Before this video, I did a 1 minute "speed date" of this ... Original Paper & Resources: Paper Title: " Mohamed Osman from MindsAI (now Tufa Labs) joins Tim to discuss how his team achieved the highest score on the ARC ...

Neural Thickets Diverse Task Experts - Detailed Analysis & Overview

I read papers during my weekly livestream, Saturdays at 8AM Pacific Time. Before this video, I did a 1 minute "speed date" of this ... Original Paper & Resources: Paper Title: " Mohamed Osman from MindsAI (now Tufa Labs) joins Tim to discuss how his team achieved the highest score on the ARC ... In this AI Research Roundup episode, Alex discusses the paper: DeepSeek-V3 runs 671 billion parameters. Kimi K3 runs 2.8 trillion. Almost none of that runs per token. So how does Mixture of ... These short, interactive tutorials are designed to teach you how to use NeuralTools software. This is the video Step 5.

Most devs are using LLMs daily but don't have a clue about some of the fundamentals. Understanding tokens is crucial because ... Earn a Generative AI certificate today → Learn more about watsonx: AI promises to ...

Photo Gallery

Neural Thickets: Diverse Task Experts Are Dense Around Pretrained Weights
Neural Thickets: Diverse Task Experts Are Dense Around Pretrained Weights (Mar 2026)
Read the Viral Neural Thickets Paper with an AI Scientist
Don't Train Your AI, Just Guess! (MIT's RandOpt & Neural Thickets)
Pretrained LLMs Are Surrounded by Task Experts feat Yulu Gan from MIT
[Podcast] Neural Thickets
Neural Thickets
Test-Time Adaptation: the key to reasoning with DL
RandOpt: Improving LLMs via Random Weight Sampling
Mixture of Experts Explained Visually: How Trillion-Parameter Models Actually Work
NeuralTools Quick Start - Step 5: Interpret the Results
Most devs don't understand how LLM tokens work
View Detailed Profile
Neural Thickets: Diverse Task Experts Are Dense Around Pretrained Weights

Neural Thickets: Diverse Task Experts Are Dense Around Pretrained Weights

Paper:

Neural Thickets: Diverse Task Experts Are Dense Around Pretrained Weights (Mar 2026)

Neural Thickets: Diverse Task Experts Are Dense Around Pretrained Weights (Mar 2026)

Title:

Read the Viral Neural Thickets Paper with an AI Scientist

Read the Viral Neural Thickets Paper with an AI Scientist

I read papers during my weekly livestream, Saturdays at 8AM Pacific Time. Before this video, I did a 1 minute "speed date" of this ...

Don't Train Your AI, Just Guess! (MIT's RandOpt & Neural Thickets)

Don't Train Your AI, Just Guess! (MIT's RandOpt & Neural Thickets)

Original Paper & Resources: Paper Title: "

Pretrained LLMs Are Surrounded by Task Experts feat Yulu Gan from MIT

Pretrained LLMs Are Surrounded by Task Experts feat Yulu Gan from MIT

But after going through the paper "

[Podcast] Neural Thickets

[Podcast] Neural Thickets

https://arxiv.org/pdf/2603.12228

Neural Thickets

Neural Thickets

https://arxiv.org/pdf/2603.12228

Test-Time Adaptation: the key to reasoning with DL

Test-Time Adaptation: the key to reasoning with DL

Mohamed Osman from MindsAI (now Tufa Labs) joins Tim to discuss how his team achieved the highest score on the ARC ...

RandOpt: Improving LLMs via Random Weight Sampling

RandOpt: Improving LLMs via Random Weight Sampling

In this AI Research Roundup episode, Alex discusses the paper:

Mixture of Experts Explained Visually: How Trillion-Parameter Models Actually Work

Mixture of Experts Explained Visually: How Trillion-Parameter Models Actually Work

DeepSeek-V3 runs 671 billion parameters. Kimi K3 runs 2.8 trillion. Almost none of that runs per token. So how does Mixture of ...

NeuralTools Quick Start - Step 5: Interpret the Results

NeuralTools Quick Start - Step 5: Interpret the Results

These short, interactive tutorials are designed to teach you how to use NeuralTools software. This is the video Step 5.

Most devs don't understand how LLM tokens work

Most devs don't understand how LLM tokens work

Most devs are using LLMs daily but don't have a clue about some of the fundamentals. Understanding tokens is crucial because ...

Five Steps to Create a New AI Model

Five Steps to Create a New AI Model

Earn a Generative AI certificate today → https://ibm.biz/BdKUNX Learn more about watsonx: https://ibm.biz/BdvDnr AI promises to ...