RuntimeError: Expected 3-dimensional tensor

[YMLLG]

When I use this model file: whisper_frontend_mesonet_mfcc.yaml, the following error occurs.

RuntimeError: Expected 3-dimensional tensor, but got 2-dimensional tensor for argument #1 'self' (while checking arguments for adaptive_avg_pool1d)

[piotrkawa]

Hi,
can you please provide command you run, along with the related config and error stack trace?

Does this problem occurs when you run src/models/whisper_meso_net.py?

[FreshJoa]

Hi, I have the same issue whenever I use every type of model with mesonet (whisper_mesonet, mfcc_mesonet, whisper_mfcc_mesonet). The problem occurs in file https://github.com/piotrkawa/deepfake-whisper-features/blob/main/src/models/meso_net.py in the 110 line. I will be grateful for help :)

[YMLLG]

You can try these 

…

2024年6月25日 19:10，JGajewska ***@***.***> 写道： Hi, I have the same issue whenever I use every type of model with mesonet (whisper_mesonet, mfcc_mesonet, whisper_mfcc_mesonet). The problem occurs in file https://github.com/piotrkawa/deepfake-whisper-features/blob/main/src/models/meso_net.py in the 110 line. I will be grateful for help :) — Reply to this email directly, view it on GitHub <#11 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AJCDH6MAOEPRRUN7QW2XFL3ZJFFZJAVCNFSM6AAAAABJ3TQSMGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCOBYGY2DSMRXGE>. You are receiving this because you authored the thread.

[FreshJoa]

In my case, the issue was incorrectly configured in the conda environment. Theoretically "conda list" showed loaded correct versions of packages (especially torch, and torchaudio), but in practice, python saw different versions. When I created carefully once again my virtual environment and made sure that Python saw the correct versions, everything started working well. :)