Offline-trained world-model agents in DreamerV3 underperform online agents due to out-of-distribution states at test time; adding about 10% self-generated data or exploratory data largely recovers performance.
A Implementation Details A.1 Runtime Overview Our experiments comprised approximately 2000 runs, totaling 20000 GPU hours
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
extension 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
extension 1polarities
extend 1representative citing papers
citing papers explorer
-
Offline vs. Online Learning in Model-based RL: Lessons for Data Collection Strategies
Offline-trained world-model agents in DreamerV3 underperform online agents due to out-of-distribution states at test time; adding about 10% self-generated data or exploratory data largely recovers performance.