scaling_metadata.json

{
  "source_path": "/root/.cache/huggingface/hub/datasets--roman-bushuiev--GeMS/blobs/a3bf23ce812fd30a9060ec053e17234f9014b25798f16ecdb3b66d6d239e6595",
  "tfrecord_dir": "/root/md4/data/gems_tfrecord",
  "validation_fraction": 0.05,
  "examples_per_shard": 500000,
  "split_seed": 42,
  "total_records": 41951922,
  "valid_records": 41951908,
  "train_size": 39854312,
  "validation_size": 2097596,
  "rt_percentile_1": 8.605621013641358,
  "rt_percentile_95": 874.5380249023438,
  "rt_clip": [
    0.0,
    1.0
  ],
  "precursor_min": 50.0,
  "precursor_max": 1500.0,
  "mz_channel_bins": 1450,
  "loss_channel_bins": 1450,
  "bin_width": 1.0,
  "splat_ppm": 20.0,
  "min_sigma_bins": 0.75,
  "num_write_workers": 96,
  "train_files": [
    "train/gems_train.tfrecord-00000-of-00080",
    "train/gems_train.tfrecord-00001-of-00080",
    "train/gems_train.tfrecord-00002-of-00080",
    "train/gems_train.tfrecord-00003-of-00080",
    "train/gems_train.tfrecord-00004-of-00080",
    "train/gems_train.tfrecord-00005-of-00080",
    "train/gems_train.tfrecord-00006-of-00080",
    "train/gems_train.tfrecord-00007-of-00080",
    "train/gems_train.tfrecord-00008-of-00080",
    "train/gems_train.tfrecord-00009-of-00080",
    "train/gems_train.tfrecord-00010-of-00080",
    "train/gems_train.tfrecord-00011-of-00080",
    "train/gems_train.tfrecord-00012-of-00080",
    "train/gems_train.tfrecord-00013-of-00080",
    "train/gems_train.tfrecord-00014-of-00080",
    "train/gems_train.tfrecord-00015-of-00080",
    "train/gems_train.tfrecord-00016-of-00080",
    "train/gems_train.tfrecord-00017-of-00080",
    "train/gems_train.tfrecord-00018-of-00080",
    "train/gems_train.tfrecord-00019-of-00080",
    "train/gems_train.tfrecord-00020-of-00080",
    "train/gems_train.tfrecord-00021-of-00080",
    "train/gems_train.tfrecord-00022-of-00080",
    "train/gems_train.tfrecord-00023-of-00080",
    "train/gems_train.tfrecord-00024-of-00080",
    "train/gems_train.tfrecord-00025-of-00080",
    "train/gems_train.tfrecord-00026-of-00080",
    "train/gems_train.tfrecord-00027-of-00080",
    "train/gems_train.tfrecord-00028-of-00080",
    "train/gems_train.tfrecord-00029-of-00080",
    "train/gems_train.tfrecord-00030-of-00080",
    "train/gems_train.tfrecord-00031-of-00080",
    "train/gems_train.tfrecord-00032-of-00080",
    "train/gems_train.tfrecord-00033-of-00080",
    "train/gems_train.tfrecord-00034-of-00080",
    "train/gems_train.tfrecord-00035-of-00080",
    "train/gems_train.tfrecord-00036-of-00080",
    "train/gems_train.tfrecord-00037-of-00080",
    "train/gems_train.tfrecord-00038-of-00080",
    "train/gems_train.tfrecord-00039-of-00080",
    "train/gems_train.tfrecord-00040-of-00080",
    "train/gems_train.tfrecord-00041-of-00080",
    "train/gems_train.tfrecord-00042-of-00080",
    "train/gems_train.tfrecord-00043-of-00080",
    "train/gems_train.tfrecord-00044-of-00080",
    "train/gems_train.tfrecord-00045-of-00080",
    "train/gems_train.tfrecord-00046-of-00080",
    "train/gems_train.tfrecord-00047-of-00080",
    "train/gems_train.tfrecord-00048-of-00080",
    "train/gems_train.tfrecord-00049-of-00080",
    "train/gems_train.tfrecord-00050-of-00080",
    "train/gems_train.tfrecord-00051-of-00080",
    "train/gems_train.tfrecord-00052-of-00080",
    "train/gems_train.tfrecord-00053-of-00080",
    "train/gems_train.tfrecord-00054-of-00080",
    "train/gems_train.tfrecord-00055-of-00080",
    "train/gems_train.tfrecord-00056-of-00080",
    "train/gems_train.tfrecord-00057-of-00080",
    "train/gems_train.tfrecord-00058-of-00080",
    "train/gems_train.tfrecord-00059-of-00080",
    "train/gems_train.tfrecord-00060-of-00080",
    "train/gems_train.tfrecord-00061-of-00080",
    "train/gems_train.tfrecord-00062-of-00080",
    "train/gems_train.tfrecord-00063-of-00080",
    "train/gems_train.tfrecord-00064-of-00080",
    "train/gems_train.tfrecord-00065-of-00080",
    "train/gems_train.tfrecord-00066-of-00080",
    "train/gems_train.tfrecord-00067-of-00080",
    "train/gems_train.tfrecord-00068-of-00080",
    "train/gems_train.tfrecord-00069-of-00080",
    "train/gems_train.tfrecord-00070-of-00080",
    "train/gems_train.tfrecord-00071-of-00080",
    "train/gems_train.tfrecord-00072-of-00080",
    "train/gems_train.tfrecord-00073-of-00080",
    "train/gems_train.tfrecord-00074-of-00080",
    "train/gems_train.tfrecord-00075-of-00080",
    "train/gems_train.tfrecord-00076-of-00080",
    "train/gems_train.tfrecord-00077-of-00080",
    "train/gems_train.tfrecord-00078-of-00080",
    "train/gems_train.tfrecord-00079-of-00080"
  ],
  "validation_files": [
    "validation/gems_validation.tfrecord-00000-of-00009",
    "validation/gems_validation.tfrecord-00001-of-00009",
    "validation/gems_validation.tfrecord-00002-of-00009",
    "validation/gems_validation.tfrecord-00003-of-00009",
    "validation/gems_validation.tfrecord-00004-of-00009",
    "validation/gems_validation.tfrecord-00005-of-00009",
    "validation/gems_validation.tfrecord-00006-of-00009",
    "validation/gems_validation.tfrecord-00007-of-00009",
    "validation/gems_validation.tfrecord-00008-of-00009"
  ],
  "train_shard_lengths": [
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    500000,
    354312
  ],
  "validation_shard_lengths": [
    250000,
    250000,
    250000,
    250000,
    250000,
    250000,
    250000,
    250000,
    97596
  ],
  "massspec_test_files": [
    "massspec_test/gems_massspec_test.tfrecord-00000-of-00001"
  ],
  "massspec_test_shard_lengths": [
    231104
  ],
  "massspec_total_records": 231104
}