MULTIPLE GPUS

[Oheed911]

I am trying to load two models of faster-whisper as i have two gpus,
def load_models_all(gpu_id): whisper_model = WhisperModel( "large-v2", device=device, compute_type="float16")
and calling it like this
`def run_pipeline_on_gpu(gpu_id):
load_models_all(gpu_id)
# Run the pipeline code here

if name == 'main':
process1 = multiprocessing.Process(target=run_pipeline_on_gpu, args=(0,))
process2 = multiprocessing.Process(target=run_pipeline_on_gpu, args=(1,))

process1.start()
process2.start()

process1.join()
process2.join()`


This gives me error ` self.model = ctranslate2.models.Whisper(

ValueError: unsupported device cuda:1
and same for cuda:0.`

when i run it by just giving the device as "cuda" it loads the model successfuly, what could be the possible reason for this error?

[Oheed911]

I was successfull loading the model on the specific gpu by doing this , whisper_model = WhisperModel( "large-v2", device="cuda", compute_type="float16",device_index=[1])