STLGT encodes microservice traces as span graphs and applies a structure-aware linear graph transformer with a decoupled temporal module to forecast multi-step p95 tail latencies, reporting 8.5% average MAPE improvement over PERT-GNN and up to 12x faster CPU inference.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
STLGT: A Scalable Trace-Based Linear Graph Transformer for Tail Latency Prediction in Microservices
STLGT encodes microservice traces as span graphs and applies a structure-aware linear graph transformer with a decoupled temporal module to forecast multi-step p95 tail latencies, reporting 8.5% average MAPE improvement over PERT-GNN and up to 12x faster CPU inference.