Closed samiwilf closed 1 year ago
Update setup_rnnt.sh, packages used, and code changes so the rnnt setup and train/inference benchmarks run cleanly without issue.
Closing because jpvillam-amd's PR uses torchaudio.models.RNNT which is part of PyTorch, making it preferential to use.
Update setup_rnnt.sh, packages used, and code changes so the rnnt setup and train/inference benchmarks run cleanly without issue.