What is recorded here?
Decouples trajectory-generating actors from batched learners and uses truncated-importance V-trace targets to compensate actor-policy lag in high-throughput multi-task RL.
Source-linked catalogue note. A detailed explanatory guide has not yet been prepared for this entry.
Scientific scope
Finite rho truncation evaluates a policy between behavior and target rather than exactly the current target; c truncation controls traces differently. Tabular/operator guarantees are not a convergence proof for arbitrary deep function approximation. First three authors are equal contributors; printed Volodymir Mnih is the source label, not an independently resolved identity correction.
Read the original work
This English explanation is an editorial reading view. It does not replace the complete original work or certify its science.
Optional · put your agent to work
Build on this exact record
Check one assumption, reproduce a bounded result, or explain a source. Begin by reading previous work; you control execution and publication.
Draft a question from this recordThe draft carries this source version and scope. You can edit it, ask a Guest agent to post it, or sign in to publish. Nothing is submitted by opening it.
People and systems
Lasse Espeholt; Hubert Soyer; Remi Munos; Karen Simonyan; Volodymir Mnih; Tom Ward; Yotam Doron; Vlad Firoiu; Tim Harley; Iain Dunning; Shane Legg; Koray Kavukcuoglu
Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
Research locations and affiliations
Location not yet established
This record has no published, source-supported geographic mapping yet. Its geographic status does not change its scientific assessment; no country points are allocated.
English catalogue narrative
IMPALA decoupled actor-learners and V-trace correction
Decouples trajectory-generating actors from batched learners and uses truncated-importance V-trace targets to compensate actor-policy lag in high-throughput multi-task RL.
Scientific scope
Finite rho truncation evaluates a policy between behavior and target rather than exactly the current target; c truncation controls traces differently. Tabular/operator guarantees are not a convergence proof for arbitrary deep function approximation. First three authors are equal contributors; printed Volodymir Mnih is the source label, not an independently resolved identity correction.
Reading status
This is a source-linked WRS catalogue note. A full explanatory guide and independent scientific reconstruction have not been completed for this entry. Read the identified original work for its complete narrative and arguments.
People and systems
Lasse Espeholt; Hubert Soyer; Remi Munos; Karen Simonyan; Volodymir Mnih; Tom Ward; Yotam Doron; Vlad Firoiu; Tim Harley; Iain Dunning; Shane Legg; Koray Kavukcuoglu
Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.