PinnedNiksa JakovljevicIntroducing HuggingBench: A Path to Optimized Model ServingHow to find the best model serving setup to increase your model throughput and maximize resource utilizationAug 11, 20231Aug 11, 20231
Niksa JakovljevicBERT inference throughput deathmatch 🥊BERT is one of the most popular models on HuggingFace hub based on the number of downloads.This Transformer model, created by Google in…Sep 11, 2023Sep 11, 2023
Niksa JakovljevicOptimizing Resnet-50: 8X inference throughput with just a few commandsOur exploration has taken us from a starting point of mere hundreds to achieving thousands of inferences per second…Aug 18, 2023Aug 18, 2023
Niksa JakovljevicinTimescaleHow to manage Prometheus high-availability with PostgreSQL + TimescaleDBIn this post we describe how PostgreSQL + TimescaleDB can help with managing high-availability.Oct 4, 20181Oct 4, 20181
Niksa JakovljevicinTimescaleUniting SQL and NoSQL for Monitoring: Why PostgreSQL is the ultimate data store for PrometheusHow to use Prometheus, PostgreSQL + TimescaleDB, and Grafana for storing, analyzing, and visualizing metricsJul 12, 20181Jul 12, 20181