I am doing a study on the NSynth Dataset (TFR train and test) using codelab.
But I am not interesed to load the full dataset as I run out of storage.
I am only interesed in the following subset:
- instrument source = acoustic
- instrument family = guitar.
Is there a way to load and select a subset train dataset with codelab or I need to manually download the TRF and create the subset myself and load the subset TFR file into codelab?
for now i am doing:
ds_train, ds_test = tfds.load(
name="nsynth", split=["train", "test"], data_dir=DATA_DIR
Which result into a ResourceExhaustedError
Thanks for your advice....I am quite new to this topic, feel free to recommend some reading.