In an exciting development for the tech industry, the INT4 Whisper large-v2 ONNX, has been released. Whisper, a pre-trained model for automatic speech recognition (ASR) and speech translation, is poised to further revolutionise the field of deep learning and AI with its superior capabilities.
Model Card – huggingface.co/Intel/whisper-large-v2-onnx-int4
Whisper models have long been recognized for their versatility. They demonstrate an impressive ability to generalize to many datasets and domains without the need for fine-tuning. This latest release from the series, the INT4 ONNX model, looks set to continue that trend.
The INT4 ONNX model has been generated with the help of Neural-compressor, a pioneering open-source Python library. This library supports various model compression techniques on all mainstream deep learning frameworks, including TensorFlow, PyTorch, and ONNX Runtime.