Open a Whisper checkpoint (encoder-decoder speech model) through the same
profile resolution as text models. Whisper never enters the chat loop, so
it is not a RuntimeModel: the transcription engine
(packages/inference/src/transcription/whisper/)
owns it.
Open a Whisper checkpoint (encoder-decoder speech model) through the same profile resolution as text models. Whisper never enters the chat loop, so it is not a RuntimeModel: the transcription engine (packages/inference/src/transcription/whisper/) owns it.