Memory required with model medium and large

[sondt]

I'm downloaded model Medium and Large at https://ggml.ggerganov.com/ After run: whisper_init_from_file: loading model from 'ggml-model-whisper-medium.bin' whisper_model_load: loading model whisper_model_load: n_vocab = 51865 whisper_model_load: n_audio_ctx = 1500 whisper_model_load: n_audio_state = 1024 whisper_model_load: n_audio_head = 16 whisper_model_load: n_audio_layer = 24 whisper_model_load: n_text_ctx = 448 whisper_model_load: n_text_state = 1024 whisper_model_load: n_text_head = 16 whisper_model_load: n_text_layer = 24 whisper_model_load: n_mels = 80 whisper_model_load: f16 = 2 whisper_model_load: type = 4 whisper_model_load: mem required = 1720.00 MB (+ 43.00 MB per decoder) whisper_model_load: adding 1608 extra tokens whisper_model_load: model ctx = 1462.35 MB

[sandrohanea]

The memory requirements should be the same as whisper.cpp (Whisper.net is just calling that under the hood). Ofc, there will be some extra mem used by dotnet runtime (GC and everything), but that shouldn't be signifiant in this case.

You can find the requirements here: https://github.com/ggerganov/whisper.cpp#memory-usage