Distilled model which is 49% smaller and 6.3× faster while maintaining near accuracy, especially on long-form transcription. <metadata> gpu: T4 | collections: ["HF Transformers"] </metadata> - View it on GitHub