Hi all, I’m using Rasa through the standard API on an Azure Virtual Machine. Everything works fine with small models but if I try to train a big model, after 20 minutes (more or less) the api doesn’t answer anymore. The training has been completed (checking in the Rasa log) but the response is never sent to the caller. Is there any setting blocking the api after a certain time? I checked and response timeout is set at 1 hour. Thanks, Paolo
Hi @PaoloRossi79, what version of Rasa are you running in Azure?
Hi Ben, I’m running Rasa 1.9.5 (it’s for a client so I’m stick to this version). On Azure, I’m using a Data Science Virtual Machine (NVidia GPU). THe issue happens only with the train API, because it’s taking to long, the response is syncronous and the response size is quite big (the entire data model tar.gz). Thanks, Paolo