Prediction throuhput
matanox opened this issue · comments
Just curious what throughput people are getting for prediction requests; I can get up to 380 requests per second under highly concurrent load with a clojure based http server wrapper, using a non-quantized fasttext classification model.