ltuncay's picture
Submission to the Interspeech 2026 Audio Encoder Capability Challenge
eca55dc verified
_target_: src.data.yt1b_datamodule.YT1BDataModule
data_dir: ${paths.data_dir}/YT-Temporal-1B
batch_size: 64
num_workers: 4
pin_memory: True
train_parquet: train_metadata.parquet
val_parquet: val_metadata.parquet
test_parquet: val_metadata.parquet
max_audio_length_sec: 10.0
min_duration_sec: 10.0
target_sample_rate: 16000
collate_mode: pad
decode_window_sec: null