Representation Engineering (Activation Hacking)

Practical AI: Machine Learning, Data Science

Recently, we briefly mentioned the concept of “Activation Hacking” in the episode with Karan from Nous Research. In this fully connected episode, Chris and Daniel dive into the details of this model control mechanism, also called “representation engineering”. Of course, they also take time to discuss the new Sora model from OpenAI.

Leave us a comment

Changelog++ members save 4 minutes on this episode because they made the ads disappear. Join today!

Sponsors:

• Neo4j – Is your code getting dragged down by JOINs and long query times? The problem might be your database…Try simplifying the complex with graphs. Stop asking relational databases to do more than they were made for. Graphs work well for use cases with lots of data connections like supply chain, fraud detection, real-time analytics, and genAI. With Neo4j, you can code in your favorite programming language and against any driver. Plus, it’s easy to integrate into your tech stack. Visit Neo4j.com/developer to get started.
• Changelog News – A podcast+newsletter combo that’s brief, entertaining & always on-point. Subscribe today.
• 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:

• Chris Benson – Twitter, GitHub, LinkedIn, Website
• Daniel Whitenack – Twitter, GitHub, Website

Show Notes:

• Data synthesis for SOTA LLMs with Karan Malhotra from Nous Research (Practical AI #255)
• Article: Representation Engineering Mistral-7B an Acid Trip
• OpenAI Sora

Something missing or broken? PRs welcome!

Κεφάλαια

1. Welcome to Practical AI (00:00:00)

2. Keeping you fully connected (00:00:43)

3. Cool TreeHacks inventions (00:01:28)

4. Sponsor: Neo4j (00:05:51)

https://www.neo4j.com/developer?utm_source=changelog&utm_medium=podcast&utm_campaign=practicalai

5. Activation hacking & representation engineering (00:06:58)

6. Prompt hassles (00:08:55)

7. Adding control to prompts (00:10:45)

8. Binary control & hidden states (00:14:35)

9. Comparing traditional approaches (00:20:59)

10. Real world usecase for bias (00:23:13)

11. A zoo of behaviors (00:24:53)

12. Control vectors in anti-jailbreaking (00:27:21)

13. Sponsor: Changelog News (00:28:40)

https://changelog.com/news

14. Fascinating deep dive (00:30:01)

15. OpenAI's Sora (00:30:17)

16. Google's Gemma (00:35:18)

17. AGI magic (00:39:28)

18. Thanks for joining us (00:41:28)

19. Outro (00:42:47)