Update Example Code Snippets
#6
by
sanchit-gandhi
HF staff
- opened
Fixes #5. Note that the SpeechEncoderDecoderModel allows for arbitrary combinations of speech encoders and text decoders. Hence, we have arbitrary combinations of feature extractors and tokenizers, meaning it's not possible to define a processor class (which requires fixed feature extractor and tokenizer classes). Thus, we explicitly define which feature extractor and tokenizer we are using the AutoFeatureExtractor and AutoTokenizer classes.
cc @Changhan - it would be great if you could merge this simple README update for Transformers usage! Thanks!