# @package _global_
  
data:
  keep_hf_dataset_in_memory: true
  aggressive_aug: false
  n_duplicate_train: null
  n_duplicate_val: null

  tokenize_vqvae_in_dataloader: false
  enable_cuda_in_tensordict_collate: false
  force_mp_spawn: false
  keep_tensordict_on_disk: false
  move_tensordict_to_shm: false

  fid_dataset: cc12m_tokens_val_256
  image_data_train: null
  image_data_val: null
  data_dir_train: ${data.data_dir_val}
  data_dir_val:
    - dir: ${oc.env:DIFFUSION_DATA_DIR}/tokens/scratch_ssd_tokens/cc12m_tokens_val_256
      weight: 1
      name: ${data.fid_dataset}