search

m-bain / whisperX

WhisperX: Automatic Speech Recognition with Word-level Timestamps (& Diarization)
BSD 2-Clause "Simplified" License
12.43k stars 1.31k forks source link

Could not load library libcudnn_ops_infer.so.8. Error: libcudnn_ops_infer.so.8: cannot open shared object file: No such file or directory #865

Closed IgorEzerskiy closed 2 months ago

IgorEzerskiy commented 2 months ago

I'm encountering an issue when using an EC2 g5.2xlarge instance and I'm getting the error mentioned in the title.

Description: I'm using Whisper-X and Mistral in the same Docker container. In this setup, Mistral is working, but Whisper-X is not.

Packages: NVIDIA-SMI 555.42.02 CUDA Version: 12.5 NVIDIA A10G Docker version 27.1.2, build d01f264

Python Packages: ctranslate2 4.3.1 faster-whisper 1.0.0 nvidia-cublas-cu12 12.1.3.1 nvidia-cuda-cupti-cu12 12.1.105 nvidia-cuda-nvrtc-cu2 12.1.105 nvidia-cuda-runtime-cu12 12.1.105 nvidia-cudnn-cu12 9.1.0.70 nvidia-cufft-cu12 11.0.2.54 nvidia-curand-cu12 10.3.2.106 nvidia-cusolver-cu12 11.4.5.107 nvidia-cusparse-cu12 12.1.0.106 nvidia-nccl-cu12 2.20.5 nvidia-nvjitlink-cu12 12.6.20 nvidia-nvtx-cu12 12.1.105 torch 2.4.0 whisperx 3.1.1  

I've found similar errors related to torch, but in my case, torch is working within the container. Please help, I've spent a lot of time on this unsuccessfully.

saveli commented 2 months ago

It's caused by faster-whisper and the easy solution there is to pin torch==2.3.1 - it breaks with 2.4.

The hard solution is library parsing and dynamically adding it to (PYTHON)PATH.

IgorEzerskiy commented 2 months ago

It's caused by faster-whisper and the easy solution there is to pin torch==2.3.1 - it breaks with 2.4.

The hard solution is library parsing and dynamically adding it to (PYTHON)PATH.

Thank you very much, you just saved me, helped to return the torch and torchaudio libraries to version 2.3.1

geekharsh commented 2 months ago

torch==2.3.1

Thanks help me to sort out issue.

MatinEbrahimkhani commented 1 week ago

@saveli you are literally my hero. I've been struggling with this issue for more than 4 days I tested 3 operating systems 3 different CUDA versions and a lot more. thank you soooo much.