metadata
dataset_info:
features:
- name: input_ids
list: int32
- name: token_count
dtype: int32
- name: chunk_index
dtype: int32
- name: num_chunks
dtype: int32
- name: embedding
list:
list: float32
length: 1024
length: 5
- name: embedding_mask
list: bool
length: 5
- name: midi_path
dtype: string
- name: source_file
dtype: string
splits:
- name: train
num_bytes: 24182377657
num_examples: 840816
download_size: 24098457387
dataset_size: 24182377657
configs:
- config_name: default
data_files:
- split: train
path: data/train-*