ACuRL lets computer-use agents continually adapt to new digital environments by generating their own curriculum tasks and using an automatic judge, delivering 3-29% gains without human data or forgetting prior skills.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Autonomous Continual Learning for Environment Adaptation of Computer-Use Agents
ACuRL lets computer-use agents continually adapt to new digital environments by generating their own curriculum tasks and using an automatic judge, delivering 3-29% gains without human data or forgetting prior skills.