Practical AI: Machine Learning, Data Science, LLM

Data synthesis for SOTA LLMs

Feb 6, 2024

46:41

Snipd AI

Karan from Nous, a distributed collective of LLM researchers, discusses the origins of the organization and the success of their data synthesis techniques, particularly with their popular Hermes family of models. They dive into the strategies for fine-tuning models and explain why data synthesis is so effective in machine learning.

AI Summary

Highlights

AI Chapters

Episode notes

Podcast summary created with Snipd AI

Quick takeaways

The podcast discusses the organic growth and collaboration that led to the creation of the Nous research collective, highlighting the importance of specialization and cross communication among its members.

The episode explores advancements in fine-tuning models, including instruction tuning, model merging, and the use of rewards models, with a focus on enhancing control, robustness, and fidelity.

Deep dives

Organic Growth and Collaboration

The podcast episode dives into the organic growth and collaboration that led to the creation of the news research collective. It started with a small group of individuals experimenting with open source language models, and as their work gained attention and popularity, more people joined the collective. They established different channels and categories for collaboration, with a focus on data synthesis, training, and agent projects. The collective encourages specialization and cross communication to leverage the diverse skills and expertise of its members.

Understanding Synthetic Data and Data Distillation in AI Training

03:05

Maximizing Model Training Effectiveness

01:34

Introduction

2min

The Evolution of NUS Research and Open Source AI

26min

Organization and Structure of an AI Community

3min

The Impact of AI on Society and the Internet's Authenticity Problem

1min

Fine-Tuning Models and Recommended Tools

13min

Closing Remarks and Gratitude

2min

Nous Research has been pumping out some of the best open access LLMs using SOTA data synthesis techniques. Their Hermes family of models is incredibly popular! In this episode, Karan from Nous talks about the origins of Nous as a distributed collective of LLM researchers. We also get into fine-tuning strategies and why data synthesis works so well.

Leave us a comment

Changelog++ members save 2 minutes on this episode because they made the ads disappear. Join today!

Sponsors:

Read Write Own – Read, Write, Own: Building the Next Era of the Internet—a new book from entrepreneur and investor Chris Dixon—explores one possible solution to the internet’s authenticity problem: Blockchains. From AI that tracks its source material to generative programs that compensate—rather than cannibalize—creators. It’s a call to action for a more open, transparent, and democratic internet. One that opens the black box of AI, tracks the origins we see online, and much more. Order your copy of Read, Write, Own today at readwriteown.com
Fly.io – The home of Changelog.com — Deploy your apps and databases close to your users. In minutes you can run your Ruby, Go, Node, Deno, Python, or Elixir app (and databases!) all over the world. No ops required. Learn more at fly.io/changelog and check out the speedrun in their docs.

Featuring:

Karan Malhotra – LinkedIn
Chris Benson – Twitter, GitHub, LinkedIn, Website
Daniel Whitenack – Twitter, GitHub, Website

Show Notes:

Something missing or broken? PRs welcome!

Get the Snipd
podcast app

Unlock the knowledge in podcasts with the podcast player of the future.

AI-powered
podcast player

Listen to all your favourite podcasts with AI-powered features

Discover
highlights

Listen to the best highlights from the podcasts you love and dive into the full episode

Save any
moment

Hear something you like? Tap your headphones to save it with AI-generated key takeaways

Share
& Export