Using your model
Hi,
I've tried to use your model directly in whisper but when I try to load the pytorch_model.bin it says a dimension error. Can you help me.
Regards
from whisper.audio import SAMPLE_RATE, N_FRAMES, HOP_LENGTH, pad_or_trim, log_mel_spectrogram
from whisper.decoding import DecodingOptions, DecodingResult
from whisper.tokenizer import LANGUAGES, TO_LANGUAGE_CODE, get_tokenizer
from whisper.utils import exact_div, format_timestamp, optional_int, optional_float, str2bool, write_txt, write_vtt, write_srt
from whisper.transcribe import *
from whisper import *
import torch
import pdb
model=load_model("pytorch_model.bin")
Hi @nicko27 ,
Thanks for you interest. It's indeed a great idea to run with OpenAI's decoding pipelines!
You just need to convert this model to the format of OpenAI before loading it. You can take a look at this script.
A really, really big thank you. I've search this for weeks.
Thanks again