How can I use this on my fine tunned model

[mursaleen-sengr]

so basically i just fine tunned whisper on arabic to detect harakats now the model was saved in .pt , I used to map fine tunned weights on the medium model and then transcribe , now i want real time transcription of whisper , how can i do that using this.

[HatterTheMad]

Save it in a standard hugginface format then run

ct2-transformers-converter --model ./model location --output_dir faster-whisper-large-v3
--copy_files tokenizer.json preprocessor_config.json --quantization the quantization you want like float16