Error transcribing file on line parallel_for failed: cudaErrorNoKernelImageForDevice: no kernel image is available for execution on the device

[liaodong]

load models models/Whisper/faster-whisper
Error transcribing file on line parallel_for failed: cudaErrorNoKernelImageForDevice: no kernel image is available for execution on the device
/home/ai/other/miniconda3/envs/whisper-webui/lib/python3.10/site-packages/torch/cuda/memory.py:303: FutureWarning: torch.cuda.reset_max_memory_allocated now calls torch.cuda.reset_peak_memory_stats, which resets /all/ peak memory stats.
warnings.warn(
Traceback (most recent call last):
File "/home/ai/other/miniconda3/envs/whisper-webui/lib/python3.10/site-packages/gradio/queueing.py", line 501, in call_prediction
output = await route_utils.call_process_api(
File "/home/ai/other/miniconda3/envs/whisper-webui/lib/python3.10/site-packages/gradio/route_utils.py", line 253, in call_process_api
output = await app.get_blocks().process_api(
File "/home/ai/other/miniconda3/envs/whisper-webui/lib/python3.10/site-packages/gradio/blocks.py", line 1704, in process_api
data = await anyio.to_thread.run_sync(
File "/home/ai/other/miniconda3/envs/whisper-webui/lib/python3.10/site-packages/anyio/to_thread.py", line 56, in run_sync
return await get_async_backend().run_sync_in_worker_thread(
File "/home/ai/other/miniconda3/envs/whisper-webui/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 2144, in run_sync_in_worker_thread
return await future
File "/home/ai/other/miniconda3/envs/whisper-webui/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 851, in run
result = context.run(func, *args)
File "/home/ai/other/miniconda3/envs/whisper-webui/lib/python3.10/site-packages/gradio/blocks.py", line 1460, in postprocess_data
self.validate_outputs(fn_index, predictions) # type: ignore
File "/home/ai/other/miniconda3/envs/whisper-webui/lib/python3.10/site-packages/gradio/blocks.py", line 1434, in validate_outputs
raise ValueError(
ValueError: An event handler (transcribe_file) didn't receive enough output values (needed: 2, received: 1).

[jhj0517]

Hi, it seems to be a CUDA error. Can you try

nvcc --version

in the cmd and show the result? If you're using CUDA versions below 12.0, faster-whisper is not compatible.

[liaodong]

nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2023 NVIDIA Corporation
Built on Tue_Jul_11_02:20:44_PDT_2023
Cuda compilation tools, release 12.2, V12.2.128
Build cuda_12.2.r12.2/compiler.33053471_0

It looks matched

[jhj0517]

According to here,
Your GPU architecture is incompatible with CUDA 12.
Can you try reinstaliing torch

# After activate venv! 
pip install --force-reinstall torch==1.12.1+cu113 --extra-index-url https://download.pytorch.org/whl/

as the post says?

You have to activate the venv before reinstalling, you can activate the venv with

• Windows
  venv\Scripts\activate.bat or venv\Scripts\Activate.ps1
• Linux or MacOS
  source venv/bin/activate

[liaodong]

I tried 1.12.1, and no matter what version, it's the same error. Maybe it is other reason?