faster whisper workloads failing - Python kernel unresponsive

[shaunck96]

workloads failing since the last 2 days. OOM Error.
Executing in databricks env leveraging V100 GPU cluster (1 driver, 4 workers).

whisper medium.

GPU SPECS:
CUDA Version = Python version: 3.10.12 (main, Jul 29 2024, 16:56:48) [GCC 11.4.0]
CUDA available: True
CUDA version: 12.4
cuDNN version: 90100
cuDNN enabled: True
GPU Device: Tesla V100-PCIE-16GB
Number of GPUs: 2

installation command: !pip install faster-whisper --upgrade
tried executing post installing prior commits but still failing.

please advice

[shaunck96]

!pip install faster-whisper==1.0.3 ctranslate2==4.4.0 av==12.3.0
this fixed it as I reverted to older installation method