license: apache-2.0
OpenAI has all the rights for the model
Model Card for Whisper-large-v3_int8_bf16
The model is quantized version of the whisper/large-v3 with int8_bf16 quantization.
Model Details
Model Description
The model being quantized from fp32 using CTranslate2 with the following command:
ct2-transformers-converter --model your_path_to_model --output_dir desired_ouput_path --quantization int8_bf16 --force
It's originally intended to be used with fedirz/faster-whisper-server
