Using a fine-tuned model for the pipeline for the long-form transcription
#104
by
Hossep
- opened
The README explains how to use the pipeline method to run whisper for a long-form transcription:
pipe = pipeline(
"automatic-speech-recognition",
model="openai/whisper-large-v2",
chunk_length_s=30,
device=device,
)
I have a fine-tuned version of whisper-large-v3. The version is saved as a local file whisper_v3_finetuned.pt. How do I load this fine-tuned version into the pipeline?