configs/dataset/osv5m_emb.yaml

defaults:
  - train_transform: empty
  - test_transform: empty
  - _self_

name: osv5m_${dataset.embedding_name}
full_batch_size: 1024
eval_batch_size: 4096
cond_dim: 1024
output_type: emb
embedding_name: street_clip

train_dataset:
  _partial_: true
  _target_: data.webdataset.GPSWebdataset
  root: ${data_dir}/osv5m/train/
  train: true
  embedding_name: ${dataset.embedding_name}
  return_image: false
  metadata_attributes: []

val_dataset:
  _partial_: true
  _target_: data.webdataset.GPSWebdataset
  root: ${data_dir}/osv5m/val/
  train: false
  embedding_name: ${dataset.embedding_name}
  return_image: false
  metadata_attributes: ["unique_country", "unique_region", "unique_sub-region", "unique_city"]

test_dataset:
  _partial_: true
  _target_: data.webdataset.GPSWebdataset
  root: ${data_dir}/osv5m/test/
  train: false
  embedding_name: ${dataset.embedding_name}
  return_image: false
  metadata_attributes: ["unique_country", "unique_region", "unique_sub-region", "unique_city"]