diff --git a/init2winit/dataset_lib/criteo_terabyte_dataset.py b/init2winit/dataset_lib/criteo_terabyte_dataset.py index db1268b8..65ce0e51 100644 --- a/init2winit/dataset_lib/criteo_terabyte_dataset.py +++ b/init2winit/dataset_lib/criteo_terabyte_dataset.py @@ -142,9 +142,9 @@ def criteo_tsv_reader( ds = ds.repeat() ds = ds.interleave( tf.data.TextLineDataset, - cycle_length=128, + cycle_length=64, block_length=batch_size // 8, - num_parallel_calls=128, + num_parallel_calls=64, deterministic=False) if is_training: ds = ds.shuffle(buffer_size=524_288 * 100, seed=data_shuffle_seed)