task:
  hub_module_url: ''
  model:
    encoder:
      type: bigbird
      bigbird:
        use_gradient_checkpointing: false
        # hidden_size: 768
        # num_layers: 12
        # num_attention_heads: 12
        # intermediate_size: 3072
  max_answer_length: 30
  n_best_size: 20
  null_score_diff_threshold: 0.0
  init_checkpoint: 'TODO'
  train_data:
    drop_remainder: true
    global_batch_size: 48
    input_path: 'TODO'
    is_training: true
    seq_length: 1024
  validation_data:
    do_lower_case: true
    doc_stride: 128
    drop_remainder: false
    global_batch_size: 48
    input_path: 'TODO'
    is_training: false
    query_length: 64
    seq_length: 1024
    tokenization: SentencePiece
    version_2_with_negative: false
    vocab_file: 'TODO'
trainer:
  checkpoint_interval: 1000
  max_to_keep: 5
  optimizer_config:
    learning_rate:
      polynomial:
        decay_steps: 3699
        end_learning_rate: 0.0
        initial_learning_rate: 8.0e-05
        power: 1.0
      type: polynomial
    optimizer:
      type: adamw
    warmup:
      polynomial:
        power: 1
        warmup_steps: 370
      type: polynomial
  steps_per_loop: 1000
  summary_interval: 1000
  train_steps: 3699
  validation_interval: 1000
  validation_steps: 226
  best_checkpoint_export_subdir: 'best_ckpt'
  best_checkpoint_eval_metric: 'final_f1'
  best_checkpoint_metric_comp: 'higher'