ViewTube

Mixture of Experts (MoE) is everywhere: Meta / Llama 4, DeepSeek, Mistral. But how does it actually work? Do experts specialize?

Many people think that mixture of expert models have domain experts, i.e. math experts, code experts, language experts.

For more information about Stanford's online Artificial Intelligence programs visit: https://stanford.io/ai To learn more about ...

Imagine having a whole team of specialists at your disposal, each an expert in a different field, and a smart coordinator who ...

Mixture-of-Experts (MoE) models now power leading AI systems like GPT-4, Qwen3, DeepSeek-v3, and Gemini 1.5. But behind ...

DeepSeek: The First Open-Weight Reasoning Model! In this video, I'll break down DeepSeek's two flagship models— V3 and R1 ...

In this AI Research Roundup episode, Alex discusses the paper: 'Geometric Regularization in Mixture-of-Experts: The Disconnect ...

Never get stuck without AI again. Run three Small Language Models (SLMs)—also called Local LLMs—TinyLlama, Gemma-3 and ...

Slides: https://drive.google.com/file/d/11OSdPJLZ1v4QH2KHlEYGYCts5qEdR5gN/view?usp=sharing At Ray Summit 2025, ...

Get started now with open source & privacy focused password manager by Proton! https://proton.me/pass/bycloudai In this video, ...

Most developers default to transformers without understanding the alternatives. This video breaks down the intuition behind four ...

Time Stamps: 00:00 Bi-weekly vLLM project update (v0.9.2 and v0.10.0) 14:30 Scaling MoE models with llm-d 55:40 Q&A + ...

I changed 2 settings in LM Studio and I increased my t/s by about 4x. My 8gb gpu (rtx 4060) now runs GPT OSS 120b at 20t/s and ...

In this video we'll go through three methods of running SUPER LARGE AI models locally, using model streaming, model serving, ...

How is it possible for a 120 billion parameter AI model to run on a single consumer GPU? This isn't magic—it's the result of ...

You'll also learn about real-world MoE models like Mixtral, and DeepSeek, which achieve state-of-the-art performance while ...

... models efficiently before I start a quick intro about myself um I am researching LMS at Cerebras one of my recent gigs is MOE ...

Resources: - Understanding and Coding the KV Cache in LLMs from Scratch: ...

NVIDIA Nemotron 3 is an open family of hybrid Mamba Transformer MoE models, designed for agentic AI with long context and ...

Arcee AI is a the startup I've found to be taking the most real approach to monetizing their open models. With a bunch of ...