A critic-free one-step RL method with a scaling-graph decoder improves sample efficiency for simulation-guided DNN accelerator co-design, reportedly beating GA and HASCO by over an order of magnitude.
Archgym: An open-source gymnasium for machine learning assisted architecture design
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CORE: Constraint-Aware One-Step Reinforcement Learning for Simulation-Guided Neural Network Accelerator Design
A critic-free one-step RL method with a scaling-graph decoder improves sample efficiency for simulation-guided DNN accelerator co-design, reportedly beating GA and HASCO by over an order of magnitude.