//
sign in
Profile
by @danabra.mov
Profile
by @dansshadow.bsky.social
Profile
by @jimpick.com
AviHandle
by @danabra.mov
AviHandle
by @dansshadow.bsky.social
AviHandle
by @katherine.computer
EventsList
by @katherine.computer
ProfileHeader
by @dansshadow.bsky.social
ProfileHeader
by @danabra.mov
ProfileMedia
by @danabra.mov
ProfilePlays
by @danabra.mov
ProfilePosts
by @danabra.mov
ProfilePosts
by @dansshadow.bsky.social
ProfileReplies
by @danabra.mov
Record
by @atsui.org
Skircle
by @danabra.mov
StreamPlacePlaylist
by @katherine.computer
+ new component
ProfileReplies









Loading...
Interesting read! Our recent ICML 2026 paper explored a similar question in continual RL under continuously drifting environment dynamics rather than discrete task switches.
We found that under gradual non-stationarity, severe losses of plasticity were indeed much less apparent than in many task-switching settings. However, continual learning remained challenging because stability, and not plasticity, became the primary bottleneck.
This motivated our focus on preserving predictive representations across multiple timescales. Paper: arxiv.org/abs/2605.26357 Post on Bluesky: bsky.app/profile/raym...
Special thanks to @tyrellturing.bsky.social , Doina Precup, Christos Kaplanis, and the anonymous reviewers for their support, expertise, and valuable feedback throughout this project. 14/15
To understand individual consolidation variables, we use cross-attention over multi-timescale SFs (variables = keys/values; reward = query). 10/15
Slower-timescale weights keep contributing during continuous drift, showing long-term memories are vital for adaptation! 11/15