Closed captainyugi00 closed 2 months ago
Distilling large-v3
now! It's a ~2 week process if you include pseudo-labelling the data and training the model
@sanchit-gandhi will batch transcribe be supported like it is supported in large v3?
Distilling
large-v3
now! It's a ~2 week process if you include pseudo-labelling the data and training the model
This will change the game!
...Especially once @ggerganov gets his hands on it (for https://github.com/ggerganov/whisper.cpp )
looking forward for large-v3 to be distilled! :)
wait
Running distil-large-v3
with some updates that will mean better long-form WER performance using OpenAI's long-form algorithm. This should also translate to WER improvements in other libraries like faster-whisper
and Whisper cpp!
Largely speaking, the changes are:
decode_token_ids
here)Training run logs: https://wandb.ai/sanchit-gandhi/distil-whisper?workspace=user-sanchit-gandhi
Is distil-large-v3 finished?
When distill largev3?
Good things take a little time ;)
I'd say ETA is about 1-2 weeks. Getting really promising long-form WER results now (within 1.3% WER of large-v3 using OpenAI's long-form transcription algorithm) - currently training the model so that it works with condition_on_prev_text
Does Distill Large v3 support multilingual, or does it only support English?
我想说预计到达时间约为 1-2 周。现在获得了真正有希望的长格式 WER 结果(使用 OpenAI 的长格式转录算法,WER 在大型 v3 的 1.3% 以内) - 目前正在训练模型,以便它可以与
condition_on_prev_text
加油
Any news on largeV3 ?
Wish
great work! thx https://huggingface.co/distil-whisper/distil-large-v3
Hello, is this model supporting large-v3?