Yujia-Yan / Skipping-The-Frame-Level

A simple yet effective Audio-to-Midi Automatic Piano Transcription system
MIT License
75 stars 8 forks source link

About the model size #8

Closed pangtouyuqqq closed 1 year ago

pangtouyuqqq commented 1 year ago

Hello, I have a question about the model size. The pretrained model you gave is 40MB (https://github.com/Yujia-Yan/Skipping-The-Frame-Level/tree/main/transkun/pretrained) But when I run the training script using this config, I got a checkpoint which is 170MB.

Yujia-Yan commented 1 year ago

The provided one is only 1 * model size.

Let's break down what's needed to save the state during training:

  1. current model: 1* model size
  2. best model: 1* model size
  3. Adam kind of optimizer: 2 * model size

So during training it's 4 times the model size.

pangtouyuqqq commented 1 year ago

Got it~! Thank you