Hi from OctoML -- our team just put out some benchmarking results using GPT-J with a few model config/hw combinations that have relatively low latency: https:/o

Hi from OctoML -- our team just put out some benchmarking results using GPT-J with a few model config/hw combinations that have relatively low latency: https:/octoml.ai/blog/evaluating-gpt-j-deployment-options/