Go offline with the Player FM app!
Training Data Attribution via Approximate Unrolled Differentation
Manage episode 419325446 series 3524393
The paper introduces SOURCE, a computationally efficient training data attribution method that combines implicit differentiation and unrolling approaches, outperforming existing techniques in counterfactual prediction.
https://arxiv.org/abs//2405.12186
YouTube: https://www.youtube.com/@ArxivPapers
TikTok: https://www.tiktok.com/@arxiv_papers
Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016
Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers
--- Support this podcast: https://podcasters.spotify.com/pod/show/arxiv-papers/support
1162 episodes
Manage episode 419325446 series 3524393
The paper introduces SOURCE, a computationally efficient training data attribution method that combines implicit differentiation and unrolling approaches, outperforming existing techniques in counterfactual prediction.
https://arxiv.org/abs//2405.12186
YouTube: https://www.youtube.com/@ArxivPapers
TikTok: https://www.tiktok.com/@arxiv_papers
Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016
Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers
--- Support this podcast: https://podcasters.spotify.com/pod/show/arxiv-papers/support
1162 episodes
All episodes
×Welcome to Player FM!
Player FM is scanning the web for high-quality podcasts for you to enjoy right now. It's the best podcast app and works on Android, iPhone, and the web. Signup to sync subscriptions across devices.