configs/dataset/combined_emb.yaml

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

name: iNaturalist_OSV5M_YFCC100M_${dataset.embedding_name}
full_batch_size: 2048
cond_dim: 1024
eval_batch_size: 4096
output_type: emb
embedding_name: dinov2_vitl14_registers

train_dataset:
  _partial_: true
  _target_: data.webdataset.GPSWebdataset
  root: ${data_dir}/YFCC100M/train/ ${data_dir}/osv5m/train/ ${data_dir}/inaturalist/train/ ${data_dir}/osv5m/train/ ${data_dir}/inaturalist/train/
  train: true
  embedding_name: ${dataset.embedding_name}
  return_image: false
  metadata_attributes: []

val_dataset:
  _partial_: true
  _target_: data.webdataset.GPSWebdataset
  root: ${data_dir}/YFCC100M/yfcc4k/
  train: false
  embedding_name: ${dataset.embedding_name}
  return_image: false
  metadata_attributes: []

test_dataset:
  _partial_: true
  _target_: data.webdataset.GPSWebdataset
  root: ${data_dir}/YFCC100M/yfcc4k/
  train: false
  embedding_name: ${dataset.embedding_name}
  return_image: false
  metadata_attributes: []