Riva returns emtpy string for Nemo Quartznet model that finetuend on Persian

Hardware - GPU (RTX 1080Ti)
Hardware - CPU (Intel(R) Core™ i5-3570K CPU @ 3.40GHz)
Ubuntu 22.04
Riva Version 2.7.0

I finetuned a Nemo quartznet model on the Persian dataset and when I tested it, it transcribed. but when I deployed it to Riva it doesn’t transcribe and just return an empty string. Does anyone know how I can solve it?

docker logs riva-speech
riva-speech.log (82.0 KB)

HI @AmirMohammad-Babaei

Thanks for your interest in Riva,

Request to kindly share

  1. Finetuned Nemo quartzent model, Please share via a One Drive/google drive link, Please send you email address, i will send an email and you can share the link via email
  2. Please share the riva-build command used
  3. Please share the riva-deploy command used