---
dataset_info:
  features:
  - name: input_ids
    list: int32
    length: 2048
  - name: token_count
    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: 4323156539
    num_examples: 150023
  download_size: 4318282735
  dataset_size: 4323156539
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---