Join Ads Marketplace to earn through podcast sponsorships.
Manage your ads with dynamic ad insertion capability.
Monetize with Apple Podcasts Subscriptions via Podbean.
Earn rewards and recurring income from Fan Club membership.
Get the answers and support you need.
Resources and guides to launch, grow, and monetize podcast.
Stay updated with the latest podcasting tips and trends.
Check out our newest and recently released features!
Podcast interviews, best practices, and helpful tips.
The step-by-step guide to start your own podcast.
Create the best live podcast and engage your audience.
Tips on making the decision to monetize your podcast.
The best ways to get more eyes and ears on your podcast.
Everything you need to know about podcast advertising.
The ultimate guide to recording a podcast on your phone.
Steps to set up and use group recording in the Podbean app.
Join Ads Marketplace to earn through podcast sponsorships.
Manage your ads with dynamic ad insertion capability.
Monetize with Apple Podcasts Subscriptions via Podbean.
Earn rewards and recurring income from Fan Club membership.
Get the answers and support you need.
Resources and guides to launch, grow, and monetize podcast.
Stay updated with the latest podcasting tips and trends.
Check out our newest and recently released features!
Podcast interviews, best practices, and helpful tips.
The step-by-step guide to start your own podcast.
Create the best live podcast and engage your audience.
Tips on making the decision to monetize your podcast.
The best ways to get more eyes and ears on your podcast.
Everything you need to know about podcast advertising.
The ultimate guide to recording a podcast on your phone.
Steps to set up and use group recording in the Podbean app.
"Steering GPT-2-XL by adding an activation vector" by TurnTrout et al.
Summary: We demonstrate a new scalable way of interacting with language models: adding certain activation vectors into forward passes. Essentially, we add together combinations of forward passes in order to get GPT-2 to output the kinds of text we want. We provide a lot of entertaining and successful examples of these "activation additions." We also show a few activation additions which unexpectedly fail to have the desired effect.
We quantitatively evaluate how activation additions affect GPT-2's capabilities. For example, we find that adding a "wedding" vector decreases perplexity on wedding-related sentences, without harming perplexity on unrelated sentences. Overall, we find strong evidence that appropriately configured activation additions preserve GPT-2's capabilities.
Our results provide enticing clues about the kinds of programs implemented by language models. For some reason, GPT-2 allows "combination" of its forward passes, even though it was never trained to do so. Furthermore, our results are evidence of linear feature directions, including "anger", "weddings", and "create conspiracy theories."
We coin the phrase "activation engineering" to describe techniques which steer models by modifying their activations. As a complement to prompt engineering and finetuning, activation engineering is a low-overhead way to steer models at runtime. Activation additions are nearly as easy as prompting, and they offer an additional way to influence a model’s behaviors and values. We suspect that activation additions can adjust the goals being pursued by a network at inference time.
https://www.lesswrong.com/posts/5spBue2z2tw4JuDCx/steering-gpt-2-xl-by-adding-an-activation-vector
Create your
podcast in
minutes
It is Free