essentialweb_clean/2000/checkpoint_metadata.json

{
  "custom_metas": null,
  "dp": 32,
  "metas": {
    "consumed_train_samples": 1024000,
    "data_stages": [
      {
        "consumed_tokens_per_dataset_folder": {
          "/scratch/joel_niklaus/tmp/finephrase/dataset/essentialweb_clean": 4194304000
        },
        "consumed_train_samples": 1024000,
        "name": "stable",
        "start_training_step": 1
      }
    ],
    "last_stage_idx": 0,
    "last_train_step": 2000
  },
  "tp": 1,
  "version": "1.4"
}