Sign up to save your podcasts
Or
Share Becoming self-instruct: introducing early stopping criteria for minimal instruct tuning
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
Becoming self-instruct: introducing early stopping criteria for minimal instruct tuning
The paper introduces the Instruction Following Score (IFS), a metric to measure language models' ability to follow instructions. It can distinguish between base and instruct models and be used as an early stopping criteria for instruct tuning. The IFS can also lead to changes in the underlying base model semantics.
https://arxiv.org/abs//2307.03692
YouTube: https://www.youtube.com/@ArxivPapers
PODCASTS:
Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016
Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers
View all episodes
By Igor Melnyk
5
33 ratings
The paper introduces the Instruction Following Score (IFS), a metric to measure language models' ability to follow instructions. It can distinguish between base and instruct models and be used as an early stopping criteria for instruct tuning. The IFS can also lead to changes in the underlying base model semantics.
https://arxiv.org/abs//2307.03692
YouTube: https://www.youtube.com/@ArxivPapers
PODCASTS:
Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016
Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers
More shows like Arxiv Papers
View all
Exchanges
958 Listeners
Odd Lots
1,977 Listeners
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
438 Listeners
The Daily
112,858 Listeners
All-In with Chamath, Jason, Sacks & Friedberg
10,073 Listeners
Hard Fork
5,535 Listeners
UnHerd with Freddie Sayers
214 Listeners
Unsupervised Learning with Jacob Effron
51 Listeners
Latent Space: The AI Engineer Podcast
98 Listeners
BG2Pod with Brad Gerstner and Bill Gurley
473 Listeners