A standard combination of model compression and serving optimization gives 2.4x throughput on a GPU benchmark, but the headline claims of <30% latency and preserved accuracy are not supported by the paper's own data.
In the cloud, models use asynchronous microservices with Kubernetes/Kubeflow, supporting dynamic replica scaling
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Deep Learning Model Acceleration and Optimization Strategies for Real-Time Recommendation Systems
A standard combination of model compression and serving optimization gives 2.4x throughput on a GPU benchmark, but the headline claims of <30% latency and preserved accuracy are not supported by the paper's own data.