schema-observations.json

[
  {
    "path": "h5_6mer_annotation_fungi/seq_fungi_part10_sub1.h5",
    "arrays": {
      "input_ids": {
        "shape": [
          10000,
          16384
        ],
        "dtype": "uint16"
      },
      "label_minus": {
        "shape": [
          10000,
          98304
        ],
        "dtype": "int8"
      },
      "label_plus": {
        "shape": [
          10000,
          98304
        ],
        "dtype": "int8"
      }
    },
    "first_row_label_values": {
      "label_plus": [
        -1,
        0,
        1
      ],
      "label_minus": [
        -1,
        0,
        1
      ]
    }
  },
  {
    "path": "h5_6mer_annotation_invertebrate/seq_invertebrate_part10_sub1.h5",
    "arrays": {
      "input_ids": {
        "shape": [
          10000,
          16384
        ],
        "dtype": "uint16"
      },
      "label_minus": {
        "shape": [
          10000,
          98304
        ],
        "dtype": "int8"
      },
      "label_plus": {
        "shape": [
          10000,
          98304
        ],
        "dtype": "int8"
      }
    },
    "first_row_label_values": {
      "label_plus": [
        -1,
        0
      ],
      "label_minus": [
        0
      ]
    }
  },
  {
    "path": "h5_6mer_annotation_plant/seq_plant_part10_sub1.h5",
    "arrays": {
      "input_ids": {
        "shape": [
          10000,
          16384
        ],
        "dtype": "uint16"
      },
      "label_minus": {
        "shape": [
          10000,
          98304
        ],
        "dtype": "int8"
      },
      "label_plus": {
        "shape": [
          10000,
          98304
        ],
        "dtype": "int8"
      }
    },
    "first_row_label_values": {
      "label_plus": [
        0
      ],
      "label_minus": [
        0,
        1
      ]
    }
  },
  {
    "path": "h5_6mer_annotation_protozoa/seq_protozoa_part1_sub1.h5",
    "arrays": {
      "input_ids": {
        "shape": [
          10000,
          16384
        ],
        "dtype": "uint16"
      },
      "label_minus": {
        "shape": [
          10000,
          98304
        ],
        "dtype": "int8"
      },
      "label_plus": {
        "shape": [
          10000,
          98304
        ],
        "dtype": "int8"
      }
    },
    "first_row_label_values": {
      "label_plus": [
        -1,
        0,
        1
      ],
      "label_minus": [
        0,
        1
      ]
    }
  },
  {
    "path": "h5_6mer_annotation_vertebrate/seq_vertebrate_mammalian_part102_sub1.h5",
    "arrays": {
      "input_ids": {
        "shape": [
          10000,
          16384
        ],
        "dtype": "uint16"
      },
      "label_minus": {
        "shape": [
          10000,
          98304
        ],
        "dtype": "int8"
      },
      "label_plus": {
        "shape": [
          10000,
          98304
        ],
        "dtype": "int8"
      }
    },
    "first_row_label_values": {
      "label_plus": [
        0
      ],
      "label_minus": [
        0
      ]
    }
  },
  {
    "path": "eval/fungi1/GCF_000002945.2.h5",
    "arrays": {
      "input_ids": {
        "shape": [
          126,
          16384
        ],
        "dtype": "uint16"
      },
      "label_minus": {
        "shape": [
          126,
          98304
        ],
        "dtype": "int8"
      },
      "label_plus": {
        "shape": [
          126,
          98304
        ],
        "dtype": "int8"
      }
    },
    "first_row_label_values": {
      "label_plus": [
        -2,
        -1,
        0,
        1
      ],
      "label_minus": [
        -1,
        0,
        1
      ]
    }
  }
]