Replies: 1 comment 1 reply
-
@kvrban , hello. You can refer this issue, should copy 'preprocess_config.json' file when converting model. |
Beta Was this translation helpful? Give feedback.
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
I try to use the German Speech Recognition model https://huggingface.co/primeline/distil-whisper-large-v3-german
ct2-transformers-converter --model primeline/distil-whisper-large-v3-german --output_dir primeline/distil-whisper-large-v3-german
But when you load the model:
ValueError: Invalid input features shape: expected an input with shape (1, 128, 3000), but got an input with shape (1, 80, 3000) instead
Beta Was this translation helpful? Give feedback.
All reactions