I tried whisper model finetuning by referring to the blog post.
I created Seq2SeqTrainer
objects and ran train function.
The torch_model.bin
file was generated well and the test worked well with that model.
But if you look at the files of the Hugging Face Whisper models (e.g., whisper-tiny), tf_model.h5
, model.safetensors
, …etc. There are many other files.
(openai/whisper-tiny at main)
Among them, I need the tf_model.h5 model file. I would like to know how to convert it to the result after Seq2SeqTrainer
.