flâneur

augustus odena on X: "I have a bunch of thoughts about continual learning and nothing to do with them (I'm working on something else) so I figured I'd just turn them into a post: First: I think people use "continual learning" to point at a cluster of issues that are related but distinct. I'll list" / X

x.com · 676 words · saved by 1 readers

To view keyboard shortcuts, press question mark View keyboard shortcuts Home Explore Notifications Chat Grok Lists Bookmarks Communities Premium Profile More Post Atticus Wang @atticuswzf Post See new posts Conversation augustus odena @gstsdn I have a bunch of thoughts about continual learning and nothing to do with them (I'm working on something else) so I figured I'd just turn them into a post: First: I think people use "continual learning" to point at a cluster of issues that are related but distinct. I'll list the issues and then speculate about what might fix them. a) Catastrophic Forgetting: If you train on a distribution D_1 and then do SFT on another distribution D_2, you'll often find that your performance on D_1 degrades. The extent of this issue is maybe overstated and is more true for SFT than for RL, but it's still real. There's also an important limit case that IMO is a "smell" for the way we train models currently: repeated data can seriously harm model performance. H

@gstsdn: I have a bunch of thoughts about continual learning and nothing to do with them (I'm working on something else) so I figured I'd just turn them into a post: First: I think people use "continual learning" to point at a cluster of issues that are related but distinct. I'll list the issues and then speculate about what might fix them. a) Catastrophic Forgetting: If you train on a distribution D_1 and then do SFT on another distribution D_2, you'll often find that your performance on D_1 degrades. The extent of this issue is maybe overstated and is more true for SFT than for RL, but it

saved by

related reading