How to run faster-whisper on gpu on vastai/runpod,etc.

[Vashesh08]

I was trying to run a simple faster-whisper on vastai and runpod but haven't been able to run. The jupyter session keeps on crashing saying:
The kernel for transcription.ipynb appears to have died. It will restart automatically.

It runs fine if I change device from cuda to cpu.

!pip install faster_whisper
from faster_whisper import WhisperModel

model_size = "small"

# or run on GPU with INT8
model = WhisperModel(model_size, device="cuda", compute_type="int8")

segments, info = model.transcribe("video.wav", beam_size=5)

print("Detected language '%s' with probability %f" % (info.language, info.language_probability))

for segment in segments:
    print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))

It seems to be working fine when using openAI whisper from the command line. Working fine on colab as well.

[Vashesh08]

Update: Using the correct docker image worked.