[
  {
    "dataset": "pdllm_core",
    "splits": {
      "train": {
        "rows": 66560,
        "unique_sequences": 66560,
        "unique_canonical": 66558,
        "sequence_lengths": {
          "300": 66560
        },
        "canonical_label_conflicts": 0,
        "labels": {
          "0": 33280,
          "1": 33280
        }
      },
      "test": {
        "rows": 8320,
        "unique_sequences": 8320,
        "unique_canonical": 8320,
        "sequence_lengths": {
          "300": 8320
        },
        "canonical_label_conflicts": 0,
        "labels": {
          "1": 4160,
          "0": 4160
        }
      },
      "validation": {
        "rows": 8320,
        "unique_sequences": 8320,
        "unique_canonical": 8320,
        "sequence_lengths": {
          "300": 8320
        },
        "canonical_label_conflicts": 0,
        "labels": {
          "0": 4160,
          "1": 4160
        }
      }
    },
    "overlap": {
      "train__test": {
        "unique_canonical_shared": 0,
        "affected_rows_in_second": 0,
        "fraction_of_second": 0.0
      },
      "train__validation": {
        "unique_canonical_shared": 1,
        "affected_rows_in_second": 1,
        "fraction_of_second": 0.0001201923076923077
      },
      "validation__test": {
        "unique_canonical_shared": 1,
        "affected_rows_in_second": 1,
        "fraction_of_second": 0.0001201923076923077
      }
    }
  },
  {
    "dataset": "promoter_strength_leaf",
    "splits": {
      "train": {
        "rows": 58179,
        "unique_sequences": 58179,
        "unique_canonical": 58178,
        "sequence_lengths": {
          "170": 58179
        },
        "canonical_label_conflicts": 1,
        "labels": {
          "1.1406206759846684": 43,
          "-1.0525369285461719": 37,
          "0.1406206759846682": 23,
          "0.16823974298282396": 22,
          "-0.05253692854617209": 21,
          "0.5556581752635121": 20,
          "-2.374465023433535": 18,
          "0.5311882253288209": 17,
          "-1.6374994292673286": 15,
          "1.4760506844638266": 15
        }
      },
      "test": {
        "rows": 7154,
        "unique_sequences": 7154,
        "unique_canonical": 7154,
        "sequence_lengths": {
          "170": 7154
        },
        "canonical_label_conflicts": 0,
        "labels": {
          "1.1406206759846684": 10,
          "-0.7306088336588099": 5,
          "-1.0525369285461719": 5,
          "0.5556581752635121": 4,
          "0.6551938488144264": 4,
          "2.278124199734603": 3,
          "0.05785140721832582": 3,
          "1.811548814803964": 3,
          "1.555658175263512": 3,
          "0.5069469035579497": 3
        }
      },
      "validation": {
        "rows": 6825,
        "unique_sequences": 6825,
        "unique_canonical": 6825,
        "sequence_lengths": {
          "170": 6825
        },
        "canonical_label_conflicts": 0,
        "labels": {
          "1.1406206759846684": 7,
          "-1.4167227577383323": 3,
          "0.5311882253288209": 3,
          "-2.4310485517999023": 3,
          "1.0610131851849829": 3,
          "1.0950691111659103": 3,
          "1.2188154546620071": 3,
          "-2.374465023433535": 3,
          "-1.6374994292673286": 3,
          "1.62604750315491": 3
        }
      }
    },
    "overlap": {
      "train__test": {
        "unique_canonical_shared": 2,
        "affected_rows_in_second": 2,
        "fraction_of_second": 0.0002795638803466592
      },
      "train__validation": {
        "unique_canonical_shared": 0,
        "affected_rows_in_second": 0,
        "fraction_of_second": 0.0
      },
      "validation__test": {
        "unique_canonical_shared": 0,
        "affected_rows_in_second": 0,
        "fraction_of_second": 0.0
      }
    }
  }
]