Hi! This model is missing a normalizer.json file which is contained in all other model versions https://huggingface.co/distil-whisper/distil-large-v3/blob/main/normalizer.json. The missing file caused processor.tokenizer.normalize() to not do the expected job and ends up in a high Word Error Rate for evaluation.

Thanks @zifei9 !
Could you take a look @Steveeeeeeen ? I no longer have access to merge

Ready to merge
This branch is ready to get merged automatically.

Sign up or log in to comment