I am running a web app which I deployed through docker. The web app works perfectly besides one important detail, the whisperx ai model I have takes forever to run a transcription (think hours). I run the same ai function on a “T4 GPU” using Google collab and it takes 5 minutes using cuda as the device. This is clearly a problem so I wanted to know if the same functionality exists within Azure. Is this possible?

Introduced the line
if cuda.is_available(): device = "cuda" else: device = "cpu" to try and switch to cuda but slow times remain.