metadata
dataset_info:
features:
- name: input_ids
list: int32
length: 2048
- name: token_count
dtype: int32
- name: embedding
list: float32
length: 1024
- name: midi_path
dtype: string
- name: source_file
dtype: string
splits:
- name: train
num_bytes: 1904980224
num_examples: 153232
download_size: 1895516620
dataset_size: 1904980224
configs:
- config_name: default
data_files:
- split: train
path: data/train-*